Irreversible inactivation of 5-lipoxygenase by leukotriene A4. Characterization of product inactivation with purified enzyme and intact leukocytes.

We report that leukotriene A4, the electrophilic product of 5-lipoxygenase catalysis, irreversibly inactivates the enzyme. Leukotriene A4 inhibits 5-hydroxyeicosatetraenoic acid formation by human neutrophils and differentiated granulocytic HL-60 cells in a concentration-dependent manner with IC50 values = 22.4 +/- 2.5 and 29.0 +/- 8.0 microM, respectively. Recovery of cellular enzymatic activity is negligible (< 6%) following inactivation. Leukotriene A4 inactivates cellular 5-lipoxygenase without inhibiting its translocation from the cytosol to the membrane, suggesting that it impairs catalysis without impairing formation of the complex between 5-lipoxygenase and its membrane-associated activating protein. Consistent with this, leukotriene A4 inactivates purified 5-lipoxygenase from human neutrophils, via saturable, pseudo first-order kinetics with a rate constant, ki = 0.14 min-1 and a dissociation constant, Ki = 2.1 +/- 0.7 microM. Purified 5-lipoxygenase incubated with [3H]arachidonic acid incorporated a radiolabeled species that was not removed by electrophoresis under reduced denaturing conditions. Preincubation with leukotriene A4 diminished the incorporation of radiolabeled material, consistent with irreversible modification of 5-lipoxygenase by its metastable product, leukotriene A4. This unusual product inactivation mechanism may contribute to the decline in 5-lipoxygenase activity observed during catalysis.

$ Supported by an Upjohn Co. sabbatical award.
we hypothesized that LT&, an electrophilic epoxide, is also capable of reacting with certain other macromolecules. LTA, is the product of 5-LOB-LO catalysis, therefore it could exist proximal to the enzyme active site at some stage in the catalytic cycle. Thus, we investigated the inactivation of 5-LO by LTA,. The results with intact neutrophils, HL-60 cells, and with isolated enzyme suggest that LTA, participates directly in the inactivation process which accompanies 5-LO catalysis.
Inactivation of Cellular 5-LO by LTApHuman neutrophils (1.0 ml, 2 x lo7 celldml) or differentiated HL-60 cells, both prominent sources of 5-LO and FLAP (14,15,(19)(20)(21), were incubated with 0-100 p~ LTA., for 5 min at 37 "C. An aliquot (50 pl) was removed for quantitation of LTBl to verify that LTA., reached the cytosol and to verify that no 5-HETE was formed during this incubation. The cells were then stimulated with 5 p~ Ca2+ ionophore A23187 for 3 min to activate &LO, reflected by 5-HETE formation. The reaction was quenched by acidification to pH 1; samples (1.0 ml) were diluted with 1 ml of 0.9% (w/v) NaCI; extracted with ethyl acetaaexane (3 x 2 ml, l/ Irreversible Inactivation of 5-Lipoxygenase pg/ml soybean trypsin inhibitor, and disrupted by sonication for 3 x 20 s (75% duty cycle, power setting 3). &r centrifugation at 100,000 x g for 1 h at 4 "C, 200 pl of the membrane fraction (4-5 pg/pl) and cytosolic fraction (1-2 pg/pl) was immediately mixed with 100 pl of 3 x electrophoresis buffer (20 m~ Tris-HCI, pH 6.8, 0.4% (w/v) SDS, 4% glycerol, 0.24 M P-mercaptoethanol and bromphenol blue), boiled for 5 min, and a 100-p1 portion was applied to a 10% acrylamide gel with a 4% stacking gel. After gel electrophoresis, proteins were transferred to nitrocellulose membranes. Membranes were saturated with 5% non-fat dried milk to reduce nonspecific binding, then incubated with LO32 anti-5-LO antibody (14, 151, 1:400 dilution for 2 h. The membrane was then washed 4 x 15 min in Tween TBS and incubated for 1 h with goat anti-rabbit antiserum conjugated with horseradish peroxidase (1:100,000 in Tween-TBS containing 1% (v/v) fish skin gelatin). Membranes were subsequently washed 4 x 15 min in TBS with Triton X-100, 0.1% (v/v); (22). incubated in ECL reagents for 60 s and exposed with Kodak XAR-5 film Inactivation ofpurified 5-Lipoxygenase by LTA,: Kinetics and Dissociation Constant-5-LO from human leukocytes was purified by affinity chromatography (23,24) on agarose with ATP attached via a 9-carbon spacer at the C-8 adenine. The column was equilibrated with 100 m M Hepes, pH 7.3, 1 m~ EDTA, and 24 pg/ml phosphatidylcholine (buffer A). Buffy coat leukocytes were isolated from human blood (5-7 units). Residual erythrocytes were lysed by suspending cells in (NH4)C1 for 7 min. Cells were centrifuged at 300 x g for 15 min at 22 "C. Leukocytes (approximately 108/ml) in 50 r m potassium phosphate buffer, pH 7.1, 0.1 M NaCl, 2 m~ EDTA, 0.5 m~ phenylmethylsulfonyl fluoride, and 60 pg/ml soybean trypsin inhibitor were lysed sonically for 3 x 20 s at 4 "C (power setting 3,75% duty cycle). The lysate was centrifuged at 100,000 x g for 60 min at 4 "C and the supernatant fraction (10 ml, 2 mg/ml), containing cytosolic 5-LO activity, was applied at a flow rate of 15 ml h-'. The column was washed consecutively with 6 ml of equilibration buffer A (15 ml h-l) and 30 ml of high salt buffer (equilibration buffer A with 0.5 M NaCI). 5-LO was then eluted with 15 ml of buffer A containing 20 m~ ATP. Fractions (15 x 1 ml) were collected and 5-LO activity was monitored by incubating 50 pl of each fraction with 20 p~ arachidonic acid for 5 min at 25 "C and determining 5-HETE formation by RP-HPLC (24). Typically, 5-LO was purified 400-fold: specific activity in the 100,000 x g lysate was 2.5 x lo-, nmol of 5-HETE-min" p~' protein; specific activity in the eluate from the ATP affinity column was 1 x 10" nmol of 5-HETE min-' pg' protein. The identify of the purified enzyme was also established by Western blotting with LO-32 anti-5-LO antiserum. Enzyme from HL-60 cells (1 x 10Vml) was purified 300-fold: specific activity in the 100,000 x g lysate was 2-3 x lo4 nmol of 5-HETE-min-' pg-l protein; specific activity in the eluate was 250-500 x nmol of 5-HETE min-' pg' protein.
To determine the rate of inactivation we monitored 5-LO activity as a function of time after addition of LTA,. LT& (0-5 p~) was mixed with purified 5-LO (3040 pg/ml); samples (100 pl) were withdrawn at 15,30, 45,60,90, and 120 s and transferred to 900 pl of buffer containing 20 p~ arachidonic acid. This step dilutes the LTA, to negligible levels without impairing quantitation of 5-LO activity. After 15 s at 25 "C the 5-LO reaction was quenched by acidifying to pH 1-2, and quantifymg 5-HETE formation. The time required for half-inactivation (t,) and the first-order rate constants were calculated. Data were analyzed by plotting tu, versus the reciprocal of LTA, concentration. The rate constant for inactivation (kin,,) corresponds to the intercept with the ordinate; the LTA, dissociation constant (K,) corresponds to the intercept with the abscissa (25). The dissociation constant, Kr, was also calculated from Dixon plots [lNinitial uersus p~ LTA,). In this case 5-LO (45 pg of proteidml) in 1.0 ml of assay buffer (25 m~ potassium phosphate, 0.5 m M CaC12, 24 pg/ml L-a-phosphatidylcholine, and 1 m~ ATP) was incubated for 3 min at 25 "C with 0-5 p~ LTA,, prior to addition of arachidonic acid substrate. In similar experiments we used LTA, methyl ester To determine if inactivation was reversible we incubated 5-LO with 5 p~ and 50 p~ LT& for 3 min, then monitored its activity before and after gel filtration through a PD-10 column eluted with 5 ml of 100 m M Hepes, 1 m M EDTA, and 24 pg/ml phosphatidylcholine. We also determined if the inactivated enzyme retained its affinity for the ATP affinity column by applying inactive 5-L0, as described above, and monitoring its elution profile by Western blotting with anti-5-LO antiserum. Finally, we determined if Zileuton, a reversible inhibitor of 5-L0, protected the enzyme from inactivation by LTA,. Purified 5-LO (0.2 ml, 54 pg/ml) was incubated with 10 p~ Zileuton at 25 "C for 3 min prior to the addition of 5 p~ LT&. A portion (100 pl) of the sample was transferred to 900 pl of assay buffer containing 20 p~ arachidonic acid. This step dilutes the Zileuton to levels which have a negligible effect on 5-LO Covalent Modification of 5-LO: Incorporation of Radiolabeled Material during Catalysis of PHlArachidonic Acid-Purified 5-LO (100 pl, 145 pg/ml) was incubated for 2 min at 25 "C with 20 p~ arachidonic acid containing 0.20 pCi of [3Hlarachidonic acid. Samples were denatured by boiling for 5 min in a 3 x concentrate of 0.02 M Tris-HC1, pH 6.8, 4% glycerol (v/v), 0.4% (w/v) SDS, and 0.24 M P-mercaptoethanol and then analyzed by electrophoresis on a 10% polyacrylamide gel with a 4% polyacrylamide stacking gel. Radiolabeled protein was detected by autofluorography with Kodak X-AR film exposed for 4 weeks at -70 "C. In corresponding experiments 5-LO was incubated with 5 p~ LTA, for 3 min, prior to incubation with L3H1arachidonic acid, to determine if this reduced incorporation of radiolabeled metabolites into the enzyme.

RESULTS
LTA,, a product of 5-LO catalysis, inhibited 5-HETE formation by human neutrophils in a concentration-dependent manner. Half-maximal inhibition (IC5o) required 22.4 * 2.5 p~ LTA, (Fig. 1, lower panel). Inhibition originated from exogenously added LTA,, not from LTB4 generated by the neutrophils. First, LTB, formation from LTA, was always 50.5 p~ in these experiments. However, LTB4, even at a 10-fold higher concentration, had no significant effect on 5-HETE formation. For instance, neutrophils incubated with 1, 3, or 6 p~ LTB4 formed  (Fig. 1, lower panel). Results were equivalent with HL-60 cells, a granulocyte-like cell line containing 5-LO and FLAP, but distinct from neutrophils in other respects. LTA, and LTA, methyl ester each inhibited 5-HETE formation in a concentration-dependent manner with ICs0 = 29.0 * 8.0 and 19.9 * 5.2 p~, respectively ( Fig.   1, upper panel). LTA, acted irreversibly: HL-60 cells incubated with 50 PM LTA,, then washed twice with Ca2+-free buffer prior to stimulation with 5 PM A23187, formed only 0.02 nmol of 5-HETE/107 cells, a negligible amount (<6%) compared to the control value of 0.32 nmol of 5-HETE/107 cells in this particular experiment. Results were similar for LTA, methyl ester. 5-LO enzyme occurs predominantly in the cytosol of HL-60 cells under basal conditions (Fig. 2, lane 1 ). Stimulation with A23187 promotes its translocation from the cytosol to the cell membrane, consistent with results by others (15) (Fig. 2, Zane 2). We verified that this process was blocked by MK-886, a novel leukotriene biosynthesis inhibitor which impairs the translocation of 5-LO and its subsequent interaction with FLAP (Fig. 2, lane 3). In contrast, LTA, a t concentrations which inhibited 5-HETE formation >90% did not inhibit the translocation of 5-LO from cytosol to membranes (Fig. 2, Zane  4 ) . The ratio of cytosolic 5-LO/membrane-associated 5-LO in cells incubated with 50 p~ LTA,, prior to stimulation with A23187 (lane 41, was indistinguishable from corresponding control cells (lane 2). This implies that LTA, inactivated 5-LO in the cytosol prior to its translocation, and that its inactivation does not impair formation of the FLAP.5-LO complex (27).
Experiments with purified enzyme isolated from human leukocytes confirmed that LTA, inactivated 5-LO in the absence of FLAP. Activity declined exponentially as a function of time after addition of LTA, (Fig. 3, upper panel 1. For three separate enzyme preparations, half-maximal loss of activity ( t~) occurred 10.7 2 2.0 s after addition of 2 PM LTA,. The time for half-inactivation (tnh) plotted as a function of LTA, concentration showed that inactivation was saturable, with an apparent rate constant, kinact = 0.14 s-l and a dissociation constant, KI = 1.1 PM LTA, (Fig. 3, upper inset). Results were similar for LTA, methyl ester: 5-LO activity declined exponentially with halfmaximal loss of activity at 8.9 s after addition of 10 PM LTA, methyl ester. Inactivation was saturable with an apparent rate constant, kinact = 0.08 s-l and a dissociation constant, KI = 5.7 p~ (Fig. 3, lower panel and inset). Dixon plots (PM inhibitor uersus l/vinitial) were linear and yielded similar estimates for the dissociation constants of LTA, and LTA, methyl ester (Table I).  Gel filtration of the LTA,.5-L0 complex did not restore enzymatic activity, consistent with irreversible binding of LTA, to 5-LO. Enzyme incubated with 5 p~ LTA, for 3 min a t 25 "C, then gel-filtered through PD-10 columns, had a specific activity of 1.1 x nmol of 5-HETE pg" m i d after gel filtration. This was a 54% reduction compared to a corresponding control: 2.4 x nmol of 5-HETE-pg-' min-l. Enzyme incubated with 50 PM LTA, for 3 min, and then gel-filtered, had a specific activity of 0.78 x nmol of 5-HETE pg-' min-', a 68% reduction compared to the control. Gel electrophoresis indicated that 5-LO was radiolabeled during catalysis of i3H1arachidonic acid (Fig. 4) and that preincubation of 5-LO with LTA, prior to incubation with r3H]arachidonic acid reduced the in- Control and inactivated 5-LO each bound to the column and each emerged in fractions 4-6 when the column was eluted with buffer containing 20 mM ATP (data not shown). Zileuton, a reversible inhibitor which binds to the active site of 5-L0, did not protect it from inactivation by LTA, (Table 11).

DISCUSSION
Inactivation of lipoxygenase activity has been investigated periodically since its discovery by Smith and Lands (7). With 15-LO from plants (7,28,29) or reticulocytes (8-10) the decline in enzyme activity accompanying catalysis is rapid under aerobic conditions. The exact cause of the suicide process is uncertain; however, it is usually attributed to radical species derived from homolytic cleavage of the hydroperoxy lipid by a peroxidase activity intrinsic to lipoxygenase enzymes. In support of this hypothesis, exogenously added lipid hydroperoxides, identical to those generated during catalysis, will inactivate 15-LO by a pseudo first-order process (291, and radical scavengers will protect the enzyme from suicide inactivation. It should be stressed that radical-mediated inactivation is one likely mechanism, but not the only one. Catalysis of polyunsaturated fatty acids by mammalian and soybean 15-LO generates chemically reactive products, including allylic epoxides and a$-unsaturated ketones (30-33) which could also contribute to inactivation by covalently modifying the enzyme.
Mammalian 5-LO also undergoes suicide inactivation but there is less information available about this enzyme (31, and precedents from 15-LO or vegetable 5-LO (34) may have limited  mixed with 900 pl of buffer containing substrate arachidonic acid. This a Enzyme-Zileuton mixtures (100 pl) were incubated for 3 min, then step dilutes the Zileuton by IO-fold, a level which has a minimal effect on 5-LO activity. For instance, activity declines by 15% in the Zileutontreated sample relative to the control. Zileuton did not protect the enzyme from inactivation by LT&. 5-LO activity in each sample declined by 36%.

5-LO + LT&
value, considering the differences in the enzymology of the proteins. For example, the isolated mammalian 5-LO requires ATP and Ca2+ for catalysis; it has two regiospecific lipoxygenase activities, 5-LO and 8-LO; and in cells its translocation from the cytosol to the membrane is necessary for expression of maximal activity. 5-LO generates two products, 5(S)-HPETE and LTA,, which could each contribute to irreversible inactivation. The former could act via a radical mediated process, the latter via covalent attachment to susceptible nucleophiles essential for lipoxygenase activity. Aharony et al. (3), using crude cytosolic preparations, have established that 5(S)-HPETE can inactivate 5-LO. Using enzyme purified to near homogeneity, we conclude that LTA, is also capable of inactivating 5-LO via a saturable, time-dependent, first-order process. Our results are compatible with those of Aharony et al. (3) who acknowledged the possibility that additional intermediates in the reaction pathway from the enzyme.5-HPETE complex (E-S) to the enzyme-product complex (E-P) could be part of the inactivating mechanism. Aharony et al. (3) attributed a thiol-mediated protection of 5-LO to the destruction of 5-HPETE, but it is equally plausible that thiols protects a nucleophile on the enzyme from reaction with LTA,. LTA, inactivates isolated 5-LO 10 times faster than 5-HPETE, while 5-HPETE is more potent (31, suggesting that there are two discrete reactions involved. Two types of data suggest that LTA, acts via covalent attachment to 5-LO. These include: (i) detection of 5-LO labeled during catalysis of radioactive substrate and the stability of this labeled species during electrophoresis under reduced, denaturing conditions; and (ii) irreversibility of the LTA,.8LO complex by gel-filtration chromatography. There are precedents showing that eicosanoids can covalently modify the enzymes participating in their biosynthesis. LTA, hydrolase is the most thoroughly characterized example (12,(35)(36)(37). Other examples, based on detection of enzyme labeled irreversibly during catalysis of radioactive substrate are platelet cyclooxygenase (38) and thromboxane synthase (39). Our results show that 5-LO can be included in this category. The chemical nature of the modification is uncertain but the available data are consistent with nucleophilic attack by LTA,. At present, using the reliable, one-step affinity purification procedure we have been unable to obtain enough enzyme of suitable purity (99%) for electrospray mass spectrometric characterization of the modified enzyme.
This would be useful to establish the stoichiometry of the reaction.
It is notable that LTA, inactivates 5-LO in leukocytes. Its lower apparent potency in cells, compared to purified enzyme, probably originates from more complex metabolism and disposition. In neutrophils and HL-60 cells exogenously supplied LTA, decays rapidly via both enzymatic and nonenzymatic hydration, thus the concentration which crosses the membranes, reaches the cytosol, and encounters 5-LO is proportional to, but less than the amount of LTA, added. In contrast to a recent report (40) showing a partial inhibition of LTB4 formation by LTA,, we found a complete, dose-dependent inhibition. Furthermore, we found no detectable effect of LT& on sulfidopeptide formation. The increase in sulfidopeptide formation reported (40) in neutrophils treated sequentially with LTA,, then A23187, is incompatible with our data showing that LTA, inactivates 5-LO. Contamination of the neutrophil preparations in Ref. 40 with platelets could explain some of the discrepancies.
Our data with cells and with purified 5-LO indicate that the enzyme itself is the target for inactivation by LTA,. Consistent with this conclusion, LTA, did not inhibit 5-LO translocation implying that 5-LOFLAP interactions can occur with inactive enzyme. However, recent data from Evans and collegues (27) suggest that the translocation and activation of 5-LO are discrete events. In view of this, we cannot exclude the possibility that LTA, reacts with FLAP and impairs the activation step without impairing the formation of the 5-LO.FLAP complex. In summary, our results are consistent with the following scheme for irreversible, product inhibition of 6-lipoxygenase.