Mutation analysis of the human 5-lipoxygenase C-terminus: Support for a stabilizing C-terminal loop

Biochimica et Biophysica Acta 1749 (2005) 123 – 131 http://www.elsevier.com/locate/bba

Mutation analysis of the human 5-lipoxygenase C-terminus: Support for a stabilizing C-terminal loop Hisayo Okamotoa,b, Tove Hammarberga, Ying-Yi Zhangc, Bengt Perssona,d, Takashi Watanabeb, Bengt Samuelssona, Olof R3dmarka,T a

Department of Medical Biochemistry and Biophysics, Division of Physiological Chemistry II, Karolinska Institutet, S-171 77 Stockholm, Sweden b Division of Neurosurgery, Institute of Neurological Sciences, Faculty of Medicine, Tottori University, Yonago, Japan c Whitaker Cardiovascular Institute and Evans Department of Medicine, Boston University School of Medicine, Boston, MA 02118, USA d IFM Bioinformatics, Linkoping University, S-581 83 Linkoping, Sweden Received 12 January 2005; received in revised form 9 March 2005; accepted 9 March 2005 Available online 24 March 2005

Abstract Lipoxygenases contain prosthetic iron, in human 5-lipoxygenase (5LO) the C-terminal isoleucine carboxylate constitutes one of five identified ligands. ATP is one of several factors determining 5LO activity. We compared properties of a series of 5LO C-terminal deletion mutants (one to six amino acid residues deleted). All mutants were enzymatically inactive (expected due to loss of iron), but expression yield (in E. coli) and affinity to ATP–agarose was markedly different. Deletion of up to four C-terminal residues was compatible with good expression and retained affinity to the ATP-column, as for wild-type 5LO. However when also the fifth residue was deleted (Asn-669) expression yield decreased and the affinity to ATP was markedly diminished. This was interpreted as a result of deranged structure and stability, due to loss of a hydrogen bond between Asn-669 and His-399. Mutagenesis of these residues supported this conclusion. In the structure of soybean lipoxygenase-1, a C-terminal loop was pointed out as important for correct orientation of the C-terminus. Accordingly, a hydrogen bond appears to stabilize such a C-terminal loop also in 5LO. D 2005 Elsevier B.V. All rights reserved. Keywords: 5-Lipoxygenase; Lipoxygenase; Leukotriene; Arachidonic acid; Mutagenesis

1. Introduction 5-Lipoxygenase (5LO) catalyzes the two initial steps in leukotriene biosynthesis, converting arachidonic acid to 5(S)-hydroperoxy-6-trans-8,11,14-cis-eicosatetraenoic acid (5-HPETE) and subsequently into the epoxide leukotriene A4 [1]. The leukotrienes are inflammatory mediators; several findings now imply a role for 5LO and leukotrienes in atherosclerosis, see Refs. [2,3] for reviews, and recently in development of aortic aneurysm [4]. Other recent connections to pathophysiology concern pulmonary hyper-

Abbreviations: 5H(P)ETE, 5(S)-hydro(pero)xy-6-trans-8,11,14-cis-eicosatetraenoic acid; 5LO, 5-lipoxygenase; SLO-1, soybean lipoxyganese-1 T Corresponding author. Tel.: +46 8 728 7624; fax: +46 8 736 0439. E-mail address: [email protected] (O. R3dmark). 1570-9639/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2005.03.005

tension [5], multiple organ failure [6] and ischemia– reperfusion injury [7]. 5LO is also discussed in relation to cancer [8]. A model of the 5LO structure, based on the crystal structure of the ferrous form of rabbit reticulocyte 15LO [9], consists of an N-terminal h-barrel (residues 1–114) and a larger C-terminal catalytic domain (residues 121–673); compare Fig. 5. Ca2+ binds to the C2-like h-barrel [10– 12] leading to association with the nuclear membrane [12,13] and thus increased enzyme activity. MAP kinases phosphorylate 5LO on Ser-271 and Ser-663 (in the catalytic domain) and p38 MAPK pathways can activate 5LO during cell stress [14–17]. On the other hand, phosphorylation on Ser-523 by protein kinase A inhibited 5LO activity [18]. ATP is another factor stimulating 5LO activity, which also binds to 5LO [19,20], but the ATP binding site remains undefined. The large catalytic domain contains the iron,

124

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

which is required for catalysis. Four amino acid side chains constitute iron ligands (His-367, His-372, His-550, and Asn-554), while one of the oxygens in the carboxylate group of the C-terminal Ile-673 is the fifth ligand. Of the five ligands, His-372, His-550, and the C-terminal Ile appear to constitute a 2-His-1-carboxylate facial triad [21]. These are necessary to anchor the prosthetic iron; for review, see Ref. [22]. The first lipoxygenase to be crystallized was soybean lipoxygenase-1 (SLO-1) [23,24] and mammalian lipoxygenases (including 5LO) have been modelled also on the basis of this structure [25]. It was thus suggested that mammalian lipoxygenases, as SLO-1, should contain a C-terminal loop important for the geometry of the C-terminal iron ligand. Here we report mutagenesis results which support the presence of a hydrogen bond connecting His-399 with Asn-669, leading to a C-terminal loop in human 5LO. In this study, also Trp-75 and Trp-201, previously found to react with ATP analogues [20] were subjected to mutagenesis.

2. Materials and methods 2.1. Reagents Restriction enzymes, T7 sequencing kit, and other molecular biology reagents were from Amersham Pharmacia. Oligonucleotides were purchased from Scandinavian Gene Synthesis (Kfping, Sweden) or from Pharmacia. Solvents for HPLC were from Rathburn Chemicals (Walkerburn, Scotland). Other chemicals were from Sigma. 2.2. Oligonucleotide-directed mutagenesis Oligonucleotide-directed mutagenesis was performed using the selection principle described by Kunkel using the Muta-Gene kit from Bio-Rad. The plasmid pT3-5LO [26] was transformed into the Dut Ung E. coli strain CJ236 and single-strand uracil containing pT3-5LO cDNA was extracted after infection with the helper phage M13K07. Mutated cDNA was transformed into the Dut+ Ung+E. coli strain MV1190. Plasmid DNA was extracted from several clones using a Magic Miniprep kit (Promega, Madison, WI). Mutated clones were identified by sequencing. See Table 1 for list of mutants. 2.3. Expression in E. coli and extraction of protein E. coli MV 1190 harbouring mutated plasmid was grown in modified M9 medium (Na2HPO4, 6 mg/ml; KH2PO4, 3 mg/ml; NaCl, 0.5 mg/ml; NH4Cl, 1 mg/ml; MgSO4, 2 mM; FeSO4, 5 AM; glycerol, 2% (w / v); and casein hydrolysate, 2 mg/ml) containing ampicillin, 150 Ag/ml. A suitable volume (0.1 or 1 l) was inoculated with overnight culture (0.5 or 10 ml, in LB) and incubated at 18 or 27 8C. When cells had grown to an OD of circa 0.2 (620 nm), isopropyl h-d-

Table 1 5LO mutants included in this study Mutant

Codon change

C-2 C-3 C-4 C-5 C-6 H399A H399G P668E N669A W75R W75F W75S W75A W201R W201F W201S W201A

Nucleotide 2017–2022 deleted Nucleotide 2014–2022 deleted Nucleotide 2011–2022 deleted Nucleotide 2008–2022 deleted Nucleotide 2005–2022 deleted CACYGCC CACYGGC CCGYGAG AACYGCC TGGYAGG TGGYTTC TGGYTCG TGGYGCG TGGYAGG TGGYTTC TGGYTCG TGGYGCG

thiogalactopyranoside (IPTG) was added to a concentration of 0.2 mM. At OD circa 0.7 (620 nm), the cells were harvested by centrifugation at 10,000 g for 10 min, the cell pellet was snap-frozen and kept at 20 8C. For extraction, frozen pellets were suspended in 10 / 50 ml of sonication buffer (for 0.1/1 l cultures) and incubated for 5 min at room temperature followed by 25 min on ice. The sonication buffer contained 50 mM Triethanolamine– HCl, pH 8.0, 5 mM EDTA, 2 mM dithiothreitol, 60 Ag/ml soybean trypsin inhibitor, 500 Ag/ml lysozyme, and 1 mM phenylmethanesulfonylfluoride. The cell suspension (cooled in ice-water) was sonicated with a MSE MK2 150 W ultrasonic disintegrator (3  15 s, amplitude 18). The resulting homogenate was centrifuged at 15,000 g for 15 min, giving supernatant S15. For 1 l cultures, the yield was increased when the 15,000 g pellet was subjected to repeated sonication (after resuspension in 25 ml of sonication buffer). Thus, for 1 l cultures sup 15 from two centrifugations were pooled (total 75 ml). 2.4. Ammonium sulfate precipitation Proteins in S15 (10 or 75 ml, see above) were precipitated by ammonium sulfate. To the cold sample, a saturated solution of (NH4)2SO4 was added, to reach 60% (v / v). After stirring on ice for 40 min, the precipitate was collected by centrifugation at 15,000 g for 30 min, snapfrozen, and stored at 20 8C. Just prior to chromatography on agarose–ATP, the ammonium sulfate pellet was resuspended in 7 ml (for 100 ml culture) or 25 ml ( for 1 l culture) of buffer A (see below) and centrifuged at 100,000 g for 60 min, giving supernatant S100. 2.5. Chromatography on ATP–agarose For agarose–ATP affinity chromatography of samples from 100 ml cultures, a column (0.5  5, 1 ml) was packed

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

125

with C8-linked agarose–ATP resin (Sigma, A2767) and connected to a FPLC system (Pharmacia). The column was equilibrated with buffer A (50 mM Triethanolamine–HCl, pH 7.35, 2 mM EDTA, and 14 mM 2-mercaptoethanol). Buffer B was buffer A plus NaCl (1 M). During loading of the 7 ml S100 sample, 1 ml fractions were collected (Lfractions). Protein not binding to the ATP-column typically eluted in fractions L1, L2, and L3; compare Fig. 3. After loading, a 13 ml wash [consisting of 1 ml buffer A, 1 ml of linear gradient up to 100% buffer B (1 M NaCl in buffer A), 5 ml of buffer B, 1 ml of linear gradient up to 100% buffer A, and then 5 ml buffer A] was performed twice. The bound proteins were eluted by applying 3 ml of AMP solution (15 mM AMP and 100 mM NaCl in buffer A) at a flow rate of 0.2 ml/min, followed by 5 ml of ATP solution (12 mM ATP and 100 mM NaCl in buffer A) at 0.1 ml/min. The ATP eluate was collected in 12 fractions of 0.5 ml (T-fractions). 5LO typically eluted in fractions T4, T5, and T6; compare Figs. 2 and 3. To fractions containing 5LO protein, stabilizing agents were added (0.1 U/ml of glutathione peroxidase and 1 Ag/ml of superoxide dismutase) [27]. For agarose–ATP affinity chromatography of Trp-75 and Trp-201 mutant protein samples from 1 l cultures, a column (1  10 cm, 8 ml), was used. S100 (ca 32 ml) was loaded onto the column in three aliquots at a flow rate of 1 ml/min. After every 10.5 ml loading, there was a 10 ml wash (consisting of 2.5 ml buffer A, 2.5 ml of linear gradient up to 100% buffer B, 2.5 ml of linear gradient up to 100% buffer A, and then 2.5 ml buffer A). After the last load–wash cycle, there was an additional wash with 5 ml of buffer A. During the loading procedure seven L-fractions were collected (6  10 ml and 1  8 ml). The bound proteins were eluted first with 15 ml of AMP solution (12 mM AMP and 100 mM NaCl in buffer A) at a flow rate of 0.8 ml/min, followed by 5 ml of buffer A. Then, the column was eluted with 25 ml of ATP solution (12 mM ATP and 100 mM NaCl in buffer A) at 0.4 ml/min.

For gelfiltration, a small column (1  5 cm) packed with Sephadex G75 (Pharmacia) was eluted by gravity with 50 mM Tris–HCl, pH 7.5, 1 mM EDTA, and 50 mM KCl (buffer A). Elution of protein and ATP was monitored by UV-spectroscopy of collected fractions. Anion exchange chromatography was performed on a MonoQ column ¨ KTA system (Pharmacia). (0.5  5 cm) connected to a A Starting buffer was buffer A (compare above, agarose–ATP column). For elution the concentration of KCl was increased by a linear gradient up to 230 mM. 5LO eluted at circa 145 mM KCl.

2.6. Gel filtration and anion exchange chromatography

Protein concentration was determined by the Bradford method with bovine serum albumin as standard using a BioRad protein assay kit.

In order to test the effect of ATP on enzyme activity, it was required to remove ATP from the pooled 5LO protein containing fractions (from the agarose–ATP column, eluted with ATP). For nonmutated 5LO and for W75 mutants this was accomplished by repeated gelfiltration (twice) on G75, followed by anion exchange chromatography (MonoQ). The concentration of remaining ATP was determined by UVspectroscopy (extinction coefficient 15,400 at 259 nm); the highest value measured was 20 AM. For W201 mutants (obtained in low yields) only repeated gelfiltration was performed; the highest remaining ATP concentration measured was then 23 AM. These protein pools were diluted in the enzyme assay (minimum dilution 10 ). Thus the contribution of ATP, from the 5LO preparations to the enzyme assay mixtures, was negligible (at most ca 2 AM).

2.7. 5-Lipoxygenase enzyme activity assay The assay mixture (final volume 100 Al) contained 100 AM arachidonic acid, 10 AM 13-hydroperoxy-9,11-octadecadienoic acid (enzyme activator), 25 Ag/ml phosphatidylcholine, 1.9 mM CaCl2, 1.2 mM EDTA, 2.5 mM ATP, 72 mM Tris–HCl, pH 7.5, and 15 mM 2-mercaptoethanol. Suitable aliquots of 5LO samples were included, i.e. usually 0.5–10 Al of S15 or fractions from ATP- or Mono Q columns (compare below). Incubation was done at room temperature for 5 min and stopped by addition of stop solution (300 Al of MeCN/H2O/HOAc (2 : 1 : 0.008, v / v)) containing 1.1 nmol 17-OH-22 : 4, as internal standard. After incubations precipitated material was removed by centrifugation (10,000 g for 10 min) and the sample was subjected to HPLC without further purification. Aliquots were injected on a Nova-Pak C18 reverse-phase HPLC column (Waters) and eluted with MeCN/H2O/HOAc (1.1 : 0.9 : 0.004 or 1.2 : 0.8 : 0.004, v / v) at a flow rate of 1.2 ml/min. The absorbance of the eluate was monitored at 234 nm and peak areas were measured with a Waters (model 745) integrator. 5-HPETE and 5-HETE were quantified by their absorbances relative to the internal standard 17-OH22 : 4. 2.8. Protein concentration

2.9. SDS/PAGE and Western blot SDS/PAGE was performed using 10–15% gradient PhastGels (Pharmacia). Separated proteins were stained by Coomassie Brilliant Blue or subjected to Western blot. The Phast machine was used also for transfer to nitrocellulose membrane (Amersham Hybond-C). An in-house 5LO antiserum was obtained after immunization of rabbits with recombinant human 5LO protein, purified to apparent homogeneity. This antiserum was affinity purified. First, antibodies recognizing E. coli proteins in general were removed by passage through a column containing mixed soluble E. coli proteins. This was followed by enrichment of

126

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

antibodies recognizing 5LO, by affinity to a 5LO protein column. After blocking in milkpowder suspension (0.05 g/ ml), membranes were incubated with the anti-5LO antiserum, followed by peroxidase or alkaline phosphatase conjugated goat anti-rabbit IgG (Sigma). Peroxidase activity was detected with 50 mM Tris–HCl, pH 7.5, 100 mM NaCl, 0.03% hydrogen peroxide and 0.3 mg/ml 3,3V-diaminobenzidine tetrahydrochloride. Alkaline phosphatase was detected with nitro blue tetrazolium and 5-bromo-4chloro-3-indolylphosphate in 100 mM Tris/HCl, pH 9.5, 100 mM NaCl, and 5 mM MgCl2.

3. Results and discussion 3.1. The C-terminus We had found that deletion of six amino acid residues at the C-terminus of 5LO (mutant denoted C-6) resulted in complete loss of activity and decreased expression [26]. On the other hand, the mutant C-1 was expressed and purified successfully by agarose–ATP chromatography [28]. Thus, it appeared that residues within the C-terminus could be of importance for expression of a stable 5LO molecule. We determined the properties also for the 5LO mutants C-2, C3, C-4, and C-5. The mutants C-2, C-3, and C-4 were expressed in E. coli at 27 8C in about the same amounts as nonmutated 5LO (Fig. 1, upper panel), while C-5 mutant protein was undetectable in the E. coli supernatants from two different expression cultures at 27 8C, indicating decreased stability (Fig. 1, middle panel). However, when expressed at 18 8C

Fig. 1. Expression of 5LO and C-terminus deletion mutants, determined by Western blots. The indicated amounts of crude soluble proteins (S100), obtained after expressions at 27 or 18 8C, were applied to the gels. Membranes were incubated with 5LO antibody followed by second antibody connected to horseradish peroxidase (PX) or alkaline phosphatase (AP), as indicated. 5LO-std indicates purified 5LO applied as standard. Neg. indicates S100 protein from E. coli transformed with pT7T3 (no insert).

the yield of C-5 was almost the same as for the nonmutated 5LO (Fig. 1, lower panel). Also the C-6 mutant could be expressed at 18 8C, but the amount of 5LO protein in S100 was less than for C-5 (Fig. 1, lower panel). All these mutants were enzymatically inactive as expected, since they lack the C-terminal iron ligand, and thus as C-1 should not contain iron. These differences in expression were reproducible findings, for C-5 two different clones were expressed six times at 27 8C and five times at 18 8C. The affinity to ATP is an established 5LO purification procedure [29,30]. In this study we have also used the behaviour of 5LO mutants on the agarose–ATP column as a method to judge structural integrity. C-4 could be purified on the ATP–column, with about the same yield as nonmutated 5LO, according to SDS–PAGE analysis (Coomassie stain) of fractions from the ATP-column. Also the mutant C-3 appeared in T-fractions, similar to control (not shown). Thus, identical aliquots of T-fractions from 100 ml expressions of C-4 or wild-type 5LO gave bands of similar intensities, while no 5LO protein was seen for C-5 or C-6 (Fig. 2). When ATP-column fractions were analysed by Western blot, weak bands were detectable for C-5 in the Tfractions, but none for C-6 (Fig. 3). For these mutants, relatively stronger 5LO immunoreactive bands were obtained for the pass-through L-fractions. This pattern is opposite as compared to nonmutated 5LO or the mutant C4, which display more 5LO protein in T-fractions as compared to L-fractions (Fig. 3). For the negative control, no 5LO bands were detected in L- or T-fractions. The weak immunoreactive bands obtained for supernatants S15 and S100 for mutants C-5 and C-6 could in principle be due to decreased immunoreactivity (which in itself would indicate considerable structural change) instead of decreased expression. However, the altered distribution of 5LO protein between L- and T-fractions, for 5LO mutants C-5 and C-6 (Fig. 3), should be independent of immunoreactivity. Thus, deletion of up to four residues of the 5LO Cterminus (IPNSVAI673) did not affect expression yield or affinity to the ATP–agarose column. However, when also the fifth residue (Asn-669) was deleted, mutant protein could be expressed at 18 8C but not at 27 8C (indicating decreased stability of the C-5 protein) and the affinity to the ATP-column was severely compromised. These findings indicate that the overall structure of C-4 was intact, while the folding of C-5 (and C-6) had been deranged. Loss of iron itself should not be a reason for reduced ATP affinity since the mutant C-1 was purified by ATP–agarose chromatography and also deliberately inactivated 5LO (O2 exposure leading to loss of iron) retained affinity to ATP [31]. Modelling of mammalian lipoxygenase structures suggested that as in SLO-1, a C-terminal loop should exist also in mammalian lipoxygenases including 5LO and rabbit 15lipoxygenase (15LO) [25]. This loop would be important for correct orientation of the C-terminal iron ligand, i.e. Ile-673 in 5LO. A hydrogen bond should connect the carbonyl

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

Fig. 2. Protein in T-fractions from ATP–agarose chromatographies of 5LO and mutants. The expression volume was 100 ml for all, except 5LOH399A which was 1 litre. For each mutant, expression temperature, OD at harvest, and amount of soluble protein (S100) applied to the ATP-column are indicated. ATP-column (1 ml column volume) purifications were performed as described in Materials and methods. Two microliter aliquots of the fractions T4–T7 were applied to 10–15% gradient PhastGels and stained by Coomassie Brilliant Blue. When bands are seen, these appeared at retentions compatible with 5LO (MW 78,000).

oxygen of Asn-669 with Ny of the side chain of His-399 and this hydrogen bond should be important for the loop structure. We have modelled the 5LO structure on the basis of the crystral structure reported for rabbit 15LO [9]. Also in this model structure (Fig. 5) the Asn-669 carbonyl oxygen and His-399 Ny are close; the distance 2.64 2 is suitable for a hydrogen bond. Both Asn-669 and His-399 are conserved in all mammalian LOs and in plant LOs. In rabbit 15LO (PDB 1LOX) the corresponding distance (residues Asn-658 and His-393) is 2.87 2 and in SLO-1 (PDB 1F8N) the corresponding distance (residues Asn-835 and His-531) is 2.77 2. Apparently, such a loop could be important in all lipoxygenase structures. We reasoned that our results with the C-terminal deletion mutants could be due to interruption of a putative C-terminal loop.

127

His-399 in 5LO is one of the six conserved histidine residues typical for lipoxygenases and this residue was previously mutated to Asn, Gln, Ser, and Ala [26,32,33]. Of these, H399N, H399Q, and H399S were active and H399Q and H399S could be purified by agarose–ATP chromatography [34,35]. However, the mutant 5LO-H399A was inactive and expression (at 26 8C) was reduced [33]. The side chains of Asn, Gln and Ser should be able to take part in a hydrogen bond to Asn-669. The already published data clearly showed that exchange of His-399 to any of these three amino acids allowed for expression of comparatively stable and active 5LO proteins. To confirm that mutation of His-399 to a residue with side chain not forming a hydrogen bond leads to an unstable 5LO molecule, we attempted to express 5LOH399A and 5LO-H399G, at 18 8C or 27 8C. For both mutants no 5LO protein was found in T-fractions by SDS– PAGE and Coomassie staining (Fig. 2). For 5LO-H399A the expression was low or undetectable also by Western blot (Fig. 3), while 5LO-H399G could be expressed at 18 8C (although in reduced amounts compared to wild type 5LO, see S15 and S100 in Fig. 3). Furthermore, for 5LOH399G, the affinity to the ATP-column was much decreased. Relatively stronger 5LO immunoreactive bands were obtained for the pass-through L-fractions, compared to the T-fractions (as found also for mutants C-5 and C-6; compare in Fig. 3). The small amount of 5LO-H399G protein obtained in T-fractions had a low specific activity (fraction T5: approx. 2 Amol/mg protein, compare wildtype 5LO 20–30 Amol/mg). In both 5LO-H399A and in 5LO-H399G, the H-bond to the carbonyl oxygen of Asn669 should be absent and thus the C-terminal loop. Since 5LO-H399A was hardly expressed at all (while 5LOH399G was) it appears that His-399 is important for structure and stability also by other means, in addition to the H-bond to Asn-669. 5LO-N669A was expressed at 27 8C, in similar amounts as nonmutated 5LO. The Western blot signal for the crude E. coli supernatant (S15) was comparable to that of wild type 5LO (Fig. 4) and SDS–PAGE analysis of T-fractions showed strong bands at the position of 5LO (Fig. 2). Thus, for this mutant which could still have the putative loop structure, the affinity to the ATP-column was unaffected and the specific activity was actually higher than for nonmutated 5LO (49 Amol/mg). The neighbouring Pro-668 was mutated to Glu. Also 5LO-P668E was expressed in good yield, and the affinity to the ATP-column was similar as for nonmutated 5LO (Figs. 2 and 4). However the specific activity for 5LO-P668E was low, 3 Amol/mg. Previously, when Val672 was changed to Met in mouse 5LO this led to much reduced 5LO enzyme activity, while protein expression (in HEK 293 cells) was similar to control [36]. Probably the two latter mutations distorted the geometry of the Cterminal iron ligand more locally, leading to reduced activity, without disrupting overall protein folding and stability.

128

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

Fig. 3. Western blots of 5LO protein in crude supernatants (S15 and S100), L- and T-fractions from ATP–agarose chromatographies of 5LO and mutants. The expression batches analyzed are the same as in Fig. 2, expression volume was 100 ml for all, except H399A (1 l). For each mutant, expression temperature, OD at harvest, and amount of soluble protein (S100) applied to the ATP-column are indicated. ATP-column (1 ml column volume) purifications were performed as described in Materials and methods. For crude soluble proteins (S15, S100), the indicated amounts of protein were applied to the gels. For L- and T-fractions (L1–L3, T4–T7), 2 Al aliquots were applied. After SDS–PAGE on 10–15% gradient PhastGels, proteins were electroblotted to nitrocellulose membranes and analyzed for 5LO. 5LO-std indicates purified 5LO applied as standard. 5LO-S100 indicates a crude 5LO sample (S100) also used as standard. Neg. indicates E. coli transformed with pT7T3 (no insert), while none indicates no sample in lane.

Taken together, the results support the existence of a hydrogen bond bridging His-399 and Asn-669 in 5LO, resulting in a so-called C-terminal loop as illustrated in Fig. 5. The derangement of overall structure and stability that can be expected when such a loop cannot be formed is a probable reason for the reduced expression and/or binding to the

Fig. 4. Expression of 5LO mutants N669A and P668E determined by Western blots. The indicated amounts of crude soluble proteins (S15), obtained after expression at 27 8C, were applied to the gels. Membranes were incubated with 5LO antibody followed by second antibody (PX). Indicated amounts of S15 from expression of 5LO were applied as standards. Neg. indicates S15 protein from E. coli transformed with pT7T3 (no insert).

agarose–ATP column of the 5LO mutants C-5, C-6, H399A, and H399G. On the other hand, a suggested association of Ser-670 with His-550 (via a water molecule) [25] does not seem crucial since the mutant C-4 was well expressed. Also, a suggested involvement of the Asn-669 side chain in H-bond formation appears less important for 5LO since the mutant 5LO-N669A was well expressed and active. Finally, since the mutant C-1 (lacking iron) could be well expressed and purified [28] while this was not the case for C-5 (presumably lacking the C-terminal loop), it may be concluded that the iron itself is not crucial for structure. Apparently, the hydrogen bond between His-399 and Asn-669 is more important. It is also reassuring that the data fit with the model structure of 5LO, adding to the credibility of this structure. 3.2. Tryptophanes 75 and 201 In a previous study, the ATP analogue 2-azido-ATP was covalently bound to two residues in purified recombinant human 5LO, i.e. Trp-75 and Trp-201, indicating that these

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

129

Fig. 5. Left: model of 5LO calculated on the structure of rabbit 15-lipoxygenase [9]. In the h-barrel, Trp-75 is indicated. In the large catalytic domain iron ligands His-367, His-372, His-550, Asn-554, the C-terminus, His-399, and Trp-201 are indicated. The blow-up (middle) shows residues 398–400 (ball and stick) and 667–673 (sticks). The H-bond between the carbonyl oxygen of Asn-669 and Ny of the side chain of His-399 is indicated by the dotted line. This Hbond, and the C-terminal loop anchoring the C-terminal iron ligand Ile-673, is also shown in the cartoon (right).

residues are close to an ATP binding site [20]. We mutated these Trp residues to Arg, Phe, Ser, and Ala and determined the effects on expression and activity. The effects of mutations are summarized in Table 2. Exchanges of Trp-75 did not affect 5LO enzyme activity, expression, or purification yield appreciably. For both crude and purified proteins, all Trp-75 mutants were activated by the addition of Ca2+ only and as for nonmutated 5LO ATP added together with Ca2+ gave maximum activities. Thus, although Trp-75 is close to (or part of) an ATP binding site [20] this residue does not appear to be critical for ATP binding or stimulation of enzyme activity by ATP. Mutations of Trp-201 had more profound effects. Exchange to charged Arg prohibited expression of a stable 5LO protein (four clones tested); only very weak bands were seen in Western blots of crude soluble protein (S15, not shown). Exchange to Ala or Ser drastically reduced enzyme activity (Table 2). Activity was low both in response to Ca2+ and in response to Ca2+ plus ATP. In the 5LO model, Trp-201 is distant from the Ca2+ binding h-barrel (Fig. 5). It thus appears that the low activity for W201A/S was due to deranged structure, possibly in addition to an effect on ATP

binding. If only ATP binding would be prevented without an effect on overall structure, one could imagine that Ca2+ should still stimulate enzyme activity, which was not the case. Also, the behaviour on the ATP-column was changed. For 5LO-W201A and 5LO-W201S the amount of protein recovered from the column was about 20% as compared to nonmutated 5LO, although similar amounts of 5LO protein were applied (Table 2). This could be a result of defective overall structure leading to decreased ATP affinity. On the other hand, exchange of Trp-201 to Phe (5LO-W201F) gave intermediate enzyme activity and the behaviour on the ATPcolumn was more similar to that of nonmutated 5LO. Possibly, the aromatic ring structure of Phe could resemble that of Trp-201 in the 5LO structure and/or in binding to ATP in a so-called bbase stackQ. The data allow for a distinction between Trp-75 and Trp201, the two residues that were covalently modified by 2azido-ATP. Trp-75 appears only in mammalian 5LOs, except for rat 5LO, which has Arg at this position. Located on the tip of one of the solvent exposed loops of the h-barrel (Fig. 5), Trp-75 is one of three Trp residues involved in membrane association of the human 5LO h-barrel [12].

Table 2 Amounts and enzyme activities of 5LO mutants at residues Trp-75 and Trp-201, expressed at 27 8C Protein

Soluble protein (S15) from 1 l expression culture Amount (mg)

5LO W75R W75F W75A W75S W201F W201A W201S

398 346 372 455 474 338 441 462

Western blot + + + + + + + +

Purified 5LO protein

Activity (nmol/mg) Ca2+

Ca2+ + ATP

56 45 37 33 51 13 2 1

459 426 122 267 486 63 5 7

Amount in T-fractions (mg) 2.1 2.5 0.5 1.8 2.4 0.7 0.4 0.4

Activity (Amol/mg) Ca2+

Ca2+ + ATP

4 3 4 4 7 2 0.5 0.1

29 22 35 28 55 13 1 1

In Western blots, similar band intensities were obtained for about 10 Ag aliquots of crude soluble protein (S15) compared to standards of purified 5LO. Before determinations of activity of purified proteins, ATP was removed by gelfiltration on G75 and for Trp-75 mutants also by anion exchange on a MonoQ column, see Materials and methods. For the mutants with low/no activity (W201A/S) protein in T-fractions was identified as 5LO by Western blot.

130

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131

However, Trp-75 could be mutated without clear changes in activity and ATP affinity. Trp-201 on the other hand is conserved in almost all mammalian lipoxygenases (except mouse epithelium 12-lipoxygenase). 5LO-W201R was barely expressed and mutagenesis of Trp-201 to Ala or Ser resulted in severely reduced activity and decreased ATPcolumn yield. Trp-201 should be close to (or part of) an ATP binding site [20]. Apparently Trp-201 is also important for the overall structure of 5LO.

[14]

[15]

[16]

Acknowledgements

[17]

This study was supported by the Swedish Research Council (03X-217) and EU (QLG1-CT-2001-01521, LSHMCT-2004-0050333).

[18]

References [1] B. Samuelsson, Leukotrienes: mediators of immediate hypersensitivity reactions and inflammation, Science 220 (1983) 568 – 575. [2] O. Radmark, 5-Lipoxygenase-derived leukotrienes: mediators also of atherosclerotic inflammation, Arterioscler. Thromb. Vasc. Biol. 23 (2003) 1140 – 1142. [3] L. Zhao, C.D. Funk, Lipoxygenase pathways in atherogenesis, Trends Cardiovasc. Med. 14 (2004) 191 – 195. [4] L. Zhao, M.P. Moos, R. Grabner, F. Pedrono, J. Fan, B. Kaiser, N. John, S. Schmidt, R. Spanbroek, K. Lotzer, L. Huang, J. Cui, D.J. Rader, J.F. Evans, A.J. Habenicht, C.D. Funk, The 5-lipoxygenase pathway promotes pathogenesis of hyperlipidemia-dependent aortic aneurysm, Nat. Med. 10 (2004) 973 – 996. [5] J.E. Jones, J.L. Walker, Y. Song, N. Weiss, W.V. Cardoso, R.M. Tuder, J. Loscalzo, Y.Y. Zhang, Effect of 5-lipoxygenase on the development of pulmonary hypertension in rats, Am. J. Physiol, Heart Circ. Physiol. 286 (2004) H1775 – H1784. [6] S. Cuzzocrea, A. Rossi, I. Serraino, R. Di Paola, L. Dugo, T. Genovese, D. Britti, G. Sciarra, A. De Sarro, A.P. Caputi, L. Sautebin, Role of 5-lipoxygenase in the multiple organ failure induced by zymosan, Intensive Care Med. 30 (2004) 1935 – 1943. [7] N.S. Patel, S. Cuzzocrea, P.K. Chatterjee, R. Di Paola, L. Sautebin, D. Britti, C. Thiemermann, Reduction of renal ischemia–reperfusion injury in 5-lipoxygenase knockout mice and by the 5-lipoxygenase inhibitor zileuton, Mol. Pharmacol. 66 (2004) 220 – 227. [8] M. Romano, J. Claria, Cyclooxygenase-2 and 5-lipoxygenase converging functions on cell proliferation and tumor angiogenesis: implications for cancer therapy, FASEB. J. 17 (2003) 1986 – 1995. [9] S.A. Gillmor, A. Villasenor, R. Fletterick, E. Sigal, M.F. Browner, The structure of mammalian 15-lipoxygenase reveals similarity to the lipases and the determinants of substrate specificity, Nat. Struct. Biol. 4 (1997) 1003 – 1009. [10] T. Hammarberg, O. R3dmark, 5-Lipoxygenase binds calcium, Biochemistry 38 (1999) 4441 – 4447. [11] T. Hammarberg, P. Provost, B. Persson, O. Radmark, The Nterminal domain of 5-lipoxygenase binds calcium and mediates calcium stimulation of enzyme activity, J. Biol. Chem. 275 (2000) 38787 – 38793. [12] S. Kulkarni, S. Das, C.D. Funk, D. Murray, W. Cho, Molecular basis of the specific subcellular localization of the C2-like domain of 5lipoxygenase, J. Biol. Chem. 277 (2002) 13167 – 13174. [13] X.S. Chen, C.D. Funk, The N-terminal bbeta-barrelQ domain of 5-

[19]

[20]

[21]

[22] [23]

[24]

[25]

[26]

[27]

[28]

[29]

[30] [31] [32]

[33]

lipoxygenase is essential for nuclear membrane translocation, J. Biol. Chem. 276 (2001) 811 – 818. O. Werz, J. Klemm, B. Samuelsson, O. Radmark, 5-Lipoxygenase is phosphorylated by p38 kinase dependent MAPKAP kinases, Proc. Natl. Acad. Sci. U. S. A. 97 (2000) 5261 – 5266. O. Werz, J. Klemm, O. Radmark, B. Samuelsson, p38 MAP kinase mediates stress-induced leukotriene synthesis in a human B-lymphocyte cell line, J. Leukoc. Biol. 70 (2001) 830 – 838. O. Werz, E. Burkert, L. Fischer, D. Szellas, D. Dishart, B. Samuelsson, O. Radmark, D. Steinhilber, Extracellular signal-regulated kinases phosphorylate 5-lipoxygenase and stimulate 5-lipoxygenase product formation in leukocytes, FASEB. J. 16 (2002) 1441 – 1443. O. Werz, E. Bqrkert, B. Samuelsson, O. Radmark, D. Steinhilber, Activation of 5-lipoxygenase by cell stress is calcium independent in human polymorphonuclear leukocytes, Blood 99 (2002) 1044 – 1052. M. Luo, S.M. Jones, S.M. Phare, M.J. Coffey, M. Peters-Golden, T.G. Brock, Protein kinase A inhibits leukotriene synthesis by phosphorylation of 5-lipoxygenase on serine 523, J. Biol. Chem. 279 (2004) 41512 – 41520. K. Ochi, T. Yoshimoto, S. Yamamoto, K. Taniguchi, T. Miyamoto, Arachidonate 5-lipoxygenase of guinea pig peritoneal polymorphonuclear leukocytes. Activation by adenosine 5V-triphosphate, J. Biol. Chem. 258 (1983) 5754 – 5758. Y.Y. Zhang, T. Hammarberg, O. Radmark, B. Samuelsson, C.F. Ng, C.D. Funk, J. Loscalzo, Analysis of a nucleotide-binding site of 5lipoxygenase by affinity labelling: binding characteristics and amino acid sequences, Biochem. J. 351 (2000) 697 – 707. E.L. Hegg, L. Que Jr., The 2-His-1-carboxylate facial triad—an emerging structural motif in mononuclear non-heme iron(II) enzymes, Eur. J. Biochem. 250 (1997) 625 – 629 (Review 35 refs). O. Radmark, Arachidonate 5-lipoxygenase, Prostaglandins Other Lipid Mediat. 68–69 (2002) 211 – 234. J.C. Boyington, B.J. Gaffney, L.M. Amzel, The 3-dimensional structure of an arachidonic acid 15-lipoxygenase, Science 260 (1993) 1482 – 1486. W. Minor, J. Steczko, B. Stec, Z. Otwinowski, J.T. Bolin, R. Walter, B. Axelrod, Crystal structure of soybean lipoxygenase L-l at 1.4 angstrom resolution, Biochemistry 35 (1996) 10687 – 10701. S.T. Prigge, J.C. Boyington, B.J. Gaffney, L.M. Amzel, Structure conservation in lipoxygenases: structural analysis of soybean lipoxygenase-1 and modeling of human lipoxygenases, Proteins: Struct., Funct. Genet. 24 (1996) 275 – 291. Y.Y. Zhang, O. Radmark, B. Samuelsson, Mutagenesis of some conserved residues in human 5-lipoxygenase—effects on enzyme activity, Proc. Natl. Acad. Sci. U. S. A. 89 (1992) 485 – 489. Y.Y. Zhang, M. Hamberg, O. Radmark, B. Samuelsson, Stabilization of purified human 5-lipoxygenase with glutathione peroxidase and superoxide dismutase, Anal. Biochem. 220 (1994) 28 – 35. T. Hammarberg, Y.Y. Zhang, B. Lind, O. Radmark, B. Samuelsson, Mutations at the C-terminal isoleucine and other potential iron ligands of 5-lipoxygenase, Eur. J. Biochem. 230 (1995) 401 – 407. M. Furukawa, T. Yoshimoto, K. Ochi, S. Yamamoto, Studies on arachidonate 5-lipoxygenase of rat basophilic leukemia cells, Biochim. Biophys. Acta 795 (1984) 458 – 465. J.S. Wiseman, Alpha-secondary isotope effects in the lipoxygenase reaction, Biochemistry 28 (1989) 2106 – 2111. M.D. Percival, Human 5-lipoxygenase contains an essential iron, J. Biol. Chem. 266 (1991) 10058 – 10061. T. Nguyen, J.P. Falgueyret, M. Abramovitz, D. Riendeau, Evaluation of the role of conserved His and Met residues among lipoxygenases by site-directed mutagenesis of recombinant human 5-lipoxygenase, J. Biol. Chem. 266 (1991) 22057 – 22062. S. Ishii, M. Noguchi, M. Miyano, T. Matsumoto, M. Noma, Mutagenesis studies on the amino acid residues involved in the

H. Okamoto et al. / Biochimica et Biophysica Acta 1749 (2005) 123–131 iron-binding and the activity of human 5-lipoxygenase, Biochem. Biophys. Res. Commun. 182 (1992) 1482 – 1490. [34] Y.Y. Zhang, B. Lind, O. Radmark, B. Samuelsson, Iron content of human 5-lipoxygenase, effects of mutations regarding conserved histidine residues, J. Biol. Chem. 268 (1993) 2535 – 2541. [35] M.D. Percival, M. Ouellet, The characterization of 5 histidine–serine

131

mutants of human 5-lipoxygenase, Biochem. Biophys. Res. Commun. 186 (1992) 1265 – 1270. [36] X.S. Chen, T.A. Naumann, U. Kurre, N.A. Jenkins, N.G. Copeland, C.D. Funk, cDNA cloning, expression, mutagenesis, intracellular localization, and gene chromosomal assignment of mouse 5-lipoxygenase, J. Biol. Chem. 270 (1995) 17993 – 17999.