Conditional Expression of 15-Lipoxygenase-1 Inhibits the Selenoenzyme Thioredoxin Reductase

The selenoenzyme thioredoxin reductase regulates redox-sensitive proteins involved in inflammation and carcinogenesis, including ribonucleotide reductase, p53, NFκB, and others. Little is known about endogenous cellular factors that modulate thioredoxin reductase activity. Here we report that several metabolites of 15-lipoxygenase-1 inhibit purified thioredoxin reductase in vitro. 15(S)-Hydroperoxy-5,8,11-cis-13-trans-eicosatetraenoic acid, a metastable hydroperoxide generated by 15-lipoxygenase-1, and 4-hydroxy-2-nonenal, its non-enzymatic rearrangement product inhibit thioredoxin reductase with IC50 = 13 ± 1.5 μm and 1 ± 0.2 μm, respectively. Endogenously generated metabolites of 15-lipoxygenase-1 also inhibit thioredoxin reductase in HEK-293 cells that harbor a 15-LOX-1 gene under the control of an inducible promoter complex. Conditional, highly selective induction of 15-lipoxygenase-1 caused an inhibition of ribonucleotide reductase activity, cell cycle arrest in G1, impairment of anchorage-independent growth, and accumulation of the pro-apoptotic protein BAX. All of these responses are consistent with inhibition of thioredoxin reductase via 15-lipoxygenase-1 overexpression. In contrast, metabolites of 5-lipoxygenase were poor inhibitors of isolated thioredoxin reductase, and the overexpression of 5-lipoxygenase did not inhibit thioredoxin reductase or cause a G cell cycle arrest. The influences of 15-lipoxygenase-1 on 1inflammation, cell growth, and survival may be attributable, in part, to inhibition of thioredoxin reductase and several redox-sensitive processes subordinate to thioredoxin reductase.

The catalytic mechanism of mammalian TrxR is well understood (5). However, less is known about endogenous cellular factors that modulate its activity. Recent reports have described novel interactions between TrxR and products of the COX or LOX enzymes (6 -8). TrxR can directly reduce the lipid hydroperoxide, 15(S)-HpETE, and potentially limit its accumulation in cells that express 15-LOX (6). In addition, several electrophilic lipids, including 4-HNE, a derivative of lipid peroxidation by 15-LOX-1 or 12-LOX (9 -11) can irreversibly inhibit cellular TrxR activity (8,12). Other HpETE regioisomers and reactive lipid carbonyl compounds might also have effects similar to 15(S)-HpETE and 4-HNE. Thus, LOX enzymes (12) might modulate TrxR, and the regulatory control it exerts over other proteins and processes, such as ribonucleotide reductase and the cell cycle.

EXPERIMENTAL PROCEDURES
Materials-A full-length clone of 15-LOX-1, accession number NM_001140, and 5-LOX, accession number NM_000698 were generous gifts from Dr. Colin Funk, University of Pennsylvania. HEK-293-EcR cell lines with a pVgRXR vector were obtained from Invitrogen, Carlsbad, CA. The pIND vector was used to insert the 15-LOX-1 or 5-LOX gene 3Ј to the ecdysone response elements. HEK-293 cells were engineered similarly to permit conditional expression of COX-2 (14). Ponasterone (Invitrogen) was used for the induction of 15-LOX-1,5-LOX or * This work was supported by the Huntsman Cancer Foundation and by United States Public Health Services Grant R01 AI26730 (to F. A. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Conditional Expression of 15-LOX-1 and 5-LOX Enzymes-Stable cell lines for the conditional expression of 15-LOX-1 and 5-LOX were created using a system based on VgEcR, a chimeric protein composed of the VP16 activation domain fused to an ecdysone receptor with altered DNA-binding specificity. Upon exposure to ponasterone, this protein dimerizes with the retinoid X receptor (RXR), and the VgEcR/RXR heterodimer induces the expression of any gene inserted 3Ј to the ecdysone response element (13,14,22). We cloned the full-length cDNA of 15-LOX-1 and 5-LOX into the inducible expression vector pIND. We transfected this vector into EcR 293 cells stably transformed with the regulatory vector pVgRXR. We then isolated a population of G418resistant cells that stably expressed ponasterone-inducible 15-LOX-1 and 5-LOX.
Measurement of TrxR Activity in Lox15-1 ϩ , Lox5 ϩ , and Cox2 ϩ Cells-Lox15-1 ϩ cells were incubated for 0 -96 h with vehicle, or with 3 M ponasterone. This concentration of ponasterone was determined by a dose-response experiment (Fig. 2B). Vehicle or ponasterone was added every 24 h to maintain 15-LOX-1 expression; 20 -100 M AA was added to initiate cellular 15-LOX-1 metabolism. The corresponding cellular TrxR activity was quantified at intervals from 10 min to 8 h after the addition of AA or LA. Cox-2 ϩ and Lox5 ϩ cells were treated analogously to quantify the effect of COX-2 and 5-LOX on cellular TrxR activity.
Cells were lysed in the usual lysis buffer, sonicated intermittently for 15 s at 4°C, centrifuged at 10,000 ϫ g for 5 min, and the supernatant was harvested. 20 l of supernatant (4 g protein/l) was added to 80 l of a reaction mixture containing 50 mM Trism pH 7.4, 143 M insulin, 171 M NADPH, and 1 mM EDTA, then incubated at 25°C for 10 min. 66.6 M Trx substrate was then added to start the TrxR enzymatic reaction. The oxidation of NADPH was monitored spectrophotometrically at 340 nm in microcuvettes at 25°C for 5 min. TrxR activity corresponded to the formation of mol of NADP ϩ /min/mg of cellular protein.
Cell Cycle Analysis-Approximately 6 ϫ 10 5 cells in 2 ml of complete DMEM media were plated in 6-well plates. Lox15-1 ϩ cells were incubated daily with 3 M ponasterone or vehicle for 96 h. 20 M AA was added for 4 h prior to cell fixation, and the distribution of cells in the G 1 , S, and G 2 /M phases of the cell cycle was determined by FACS analysis. Adherent cells were released by treatment with 0.25% trypsin. The trypsin was inactivated with 10% serum in complete DMEM media. Each sample was washed twice with 1ϫ PBS and fixed overnight with ice-cold 70% ethanol at 4°C. Cells were rehydrated with 1ϫ PBS, then stained with 10% propidium iodide in 0.1% Triton X-100, 3.7% EDTA, and 100 units/ml of RNase in 1ϫ PBS for 60 min at 25°C. Cells were filtered through 35-m filters before analysis. The flow cytometer was configured to track the number of events with the FL2 parameter (BD Pharmingen FACSCAN®. The DNA content was analyzed using a nonlinear least-squares algorithm (23). Lox5 ϩ cells were treated analogously in separate experiments.
Measurement of Ribonucleotide Reductase Activity in Lox15-1 ϩ Cells-Approximately 1 ϫ 10 5 cells were plated on a T-25 flask and rested overnight. They were then treated with 3 M ponasterone every 24 h, and harvested after 96 h. Four hours prior to harvesting, cells were treated with 20 M AA. Cells were treated with 1 mM hydroxyurea, a ribonucleotide reductase inhibitor that served as a procedural control (25). Hydroxyurea was added at the same time as the AA.
Adherent cells were removed with 0.25% trypsin. To enhance cell permeability, cells were washed twice in solution A (150 mM sucrose, 80 mM KCl, 35 mM HEPES (pH 7.4), 5 mM MgCl 2, 0.5 mM CaCl 2 ), then suspended in 500 l of cold Solution A containing 0.25 mg/ml lysolecithin, and incubated for 1 min at 4°C. To measure ribonucleotide reductase activity, 1 ϫ 10 6 cells were incubated at 37°C for 10 min in 300 l of reaction mixture containing 50 mM HEPES, pH 7.4, 0.75 mM CaCl 2 , 10 mM phosphoenolpyruvate, 0.2 mM [ 3 H]rCDP, 0.2 mM rGDP, 0.2 mM rADP, and 0.2 mM dTDP. After 10 min, the incubation mixture was added to 60 l of 60% percholic acid/0.1% Na 4 P 2 O 7 to quench ribonucleotide reductase activity. The samples were incubated at 4°C for 15 min, then diluted with 1 ml of H 2 O, and centrifuged to precipitate acid-insoluble material. The pellet was extracted with 100 l of 0.2 N NaOH and incubated at 37°C for 30 min. 75 l of each sample was suspended in 5 ml of Ecoscint A, and 2Ј-deoxy-[ 3 H]rCDP, the radioactive product, was counted using a Wallac Microbeta liquid scintillation counter. Results were normalized to the total number of cells.
Immunochemical Analysis 4-HNE:TrxR Protein Adducts-For direct modification of TrxR by 4-HNE, purified TrxR (0.51 g/ul) in 50 M K 2 PO 4 , pH 6.5, and 200 M EDTA, 2 M NADPH was incubated with 10 -30 M of 4-HNE or vehicle for 0.5 min. To facilitate spontaneous formation of 4-HNE, 10 -30 M 15(S)-HpETE was preincubated with NADPH in K 2 PO 4 and EDTA buffer at 37°C for 10 min. Isolated TrxR (0.51 g/ul) was then added to the mixture and incubated for 0.5 min before fractionation by 10% SDS-PAGE. In the 4-HNE-treated TrxR samples, 75 ng of protein was added to Laemmli buffer with and without ␤-mercaptoethanol and boiled at 100°C for 5 min. Three times as much protein (225 ng) was used in the 15(S)HpETE-treated TrxR samples. Gels were transferred to polyvinylidene difluoride membranes and probed for 4-HNE adducts with rabbit anti-4-HNE (1:1000) and goat anti-rabbit-HRP (1:5000). Antigen-antibody complexes were detected with ECL reagents.
Statistics-Statistical significance at a 95% confidence interval was determined by analysis of variance with the Krusky-Wallis non-parametric test and Dunn's post hoc test.

Effect of LOX Metabolites on Purified
TrxR-Metabolites of 15-LOX-1 inhibited isolated TrxR more potently than metabolites of 5-LOX. Table I lists the concentration for half-maximal inhibition (IC 50 ) by reactive carbonyls (enones); hydroperoxy lipids, and hydroxy lipids. Overall, the most potent inhibitor was 4-HNE, a spontaneous rearrangement product of the HpETEs and HpODEs. The most potent inhibitors among hydroperoxy and hydroxy lipids were 15(S)-HpETE and 15(S)-HETE, respectively. Hydroperoxy substituents conferred ϳ2-3-fold more potency than the corresponding hydroxy substituents. The steric configuration of the hydroxy substituent also influenced the IC 50 : 15(S)-HETE was 4-fold more po-tent than the 15(R)-stereoisomer. The LOX substrates, AA and LA did not inhibit TrxR at Ͼ100 M.
4-HNE formed a covalent adduct with TrxR in vitro, as shown by the appearance of a protein with a 4-HNE epitope co-migrating with TrxR under denaturing conditions (Fig. 1). Maximal formation of the 4-HNE:TrxR adduct occurred within ϳ0.5 min. Incubation of TrxR with 15(S)-HpETE generated lesser, but detectable amounts of this adduct, consistent with spontaneous generation of 4-HNE from lipid peroxides (9,11). Boiling samples in Laemmli buffer with ␤-mercaptoethanol eliminated the 4-HNE adduct, consistent with its formation by a Michael reaction between a redox-sensitive nucleophile on TrxR and the electrophilic ␤-carbon of 4-HNE.
Inhibition of Cellular TrxR Activity by 15-LOX-1 Induction and Catalysis-Ponasterone induced enzymatically competent 15-LOX-1 in a concentration and time-dependent manner (Fig.  2). Half-maximal formation of 15(S)-HETE occurred with 20 M AA (Fig. 2C). TrxR activity in Lox15-1 ϩ cells declined after induction of 15-LOX-1 (Fig. 3). In Lox15-1 ϩ cells treated with ponasterone and 20 M AA or LA, TrxR activity was 50% lower than the corresponding control cells without 15-LOX-1 induction (p Ͻ 0.05, n ϭ 8). The NADPH oxidation shown in Fig. 3 was due to TrxR activity, because NADP ϩ formation was indistinguishable from background in the absence of added Trx (Fig. 3, panel A).  Kinetic experiments also support the conclusion that 15-LOX-1 metabolism causes the loss of TrxR activity. In Lox15-1 ϩ cells treated with ponasterone, TrxR activity fell within 10 min after the addition of AA. The inhibition was saturable, and it persisted for at least 8 h, consistent with an irreversible, or slowly reversible mechanism (Fig. 3, B and C). Comparable experiments with Cox-2 ϩ cells indicated that the induction of COX-2 had no effect on cellular TrxR activity (Table II). Ponasterone induced COX-2 with a dose-response and time course similar to its induction of 15-LOX-1. Cox-2 ϩ cells converted the majority of AA to PGE 2 .
Cell Cycle Arrest, Ribonucleotide Reductase Inhibition, and Anchorage-independent Growth: Relation to 15-LOX-1 Induction and Catalysis-Lox15-1 ϩ cells were arrested in the G 1 phase of the cell cycle following the induction of 15-LOX-1 (Fig.  4A). Results were similar with Lox15-1 ϩ cells grown in serumfree medium (CD293), which excludes the possibility that fatty acids or other substances present in the serum caused cell cycle arrest (Fig. 4).
Cell cycle arrest in G 1 is consistent with inhibition of TrxR, which regulates cellular ribonucleotide reductase activity (3). Measurements affirmed that Lox15-1 ϩ cells treated with ponasterone and AA had a 32 Ϯ 4% decline in ribonucleotide reductase activity compared with control cells (Fig. 4C).
Anchorage-independent growth in soft agar is one index of the oncogenic potential of cells. Lox15-1 ϩ cells treated with ponasterone formed fewer colonies in soft agar compared with Lox15-1 ϩ cells treated with vehicle, or compared with Cox-2 ϩ cells treated with vehicle or ponasterone (Fig. 5). Cox-2 ϩ cells treated with ponasterone formed more colonies in soft agar compared with Cox-2 ϩ cells treated with vehicle. Thus, 15-LOX-1 induction opposes anchorage-independent growth; COX-2 induction favors it.
Apoptosis: Relationship to 15-LOX-1 Induction-To determine if induction of 15-LOX-1 enhanced apoptosis, we measured bcl-2 and BAX protein expression. Bcl-2 and BAX affect apoptosis in opposite ways (27,28). Bcl-2 is anti-apoptotic, while BAX is pro-apoptotic. The ratio of BAX to bcl-2 rose in the cells expressing 15-LOX-1, suggesting higher numbers of cells poised for apoptosis (Fig. 6A). The level of BAX expression increased progressively from 0 -96 h, while the level of bcl-2 expression remained constant in Lox15-1 ϩ cells treated with ponasterone (Fig. 6A). In contrast, in Cox-2 ϩ cells the ratio of bcl-2 to BAX rose, consistent with fewer cells undergoing apoptosis (Fig. 6B). Bcl-2 protein expression increased from 24 to 96 h, while BAX expression remained constant. Consistent with these observations, Lox15-1 ϩ cells treated with ponasterone displayed a 3-fold higher caspase-3 activity than cells treated with vehicle (data not shown).
Effect of 5-LOX Expression on Cellular TrxR Activity and Cell Cycle-Data in Table I indicate 5 and 7). At t ϭ 96 h cells were lysed, and TrxR activity was quantified as described. The background level of NADPH oxidation in the lysate was negligible when no Trx substrate was added (lane 1). TrxR activity declined in Lox15-1 ϩ cells treated with ponasterone to induce 15-LOX-1 (lanes 3-5). The decline was greatest in ponasterone-treated cells that metabolized AA (lane 4) or LA (lane 5). TrxR activity in ponasterone-treated cells was Ն50% lower than the corresponding control cells (p Ͻ 0.05, n ϭ 8, analysis of variance). Addition of AA or LA had no effect on TrxR activity in cells that were not treated with ponasterone and did not express 15-LOX-1 (lanes 6 and 7). These results are consistent with data in Table  I Table I showing that several of the AA metabolites of 15-LOX-1 inhibit TrxR, but AA itself does not.

11
). Third, 15-LOX-1 overexpression had a distinctive effect that differed from two other lipid dioxygenase enzymes that we investigated. Under equivalent conditions, the overexpression of COX-2 and 5-LOX did not inhibit TrxR or modulate processes governed by TrxR. In fact, induction of 15-LOX-1 and COX-2 had opposite effects on growth in soft agar, and BAX/ Bcl-2 ratios. Our finding that 5-LOX induction did not inhibit cellular TrxR activity or arrest the cell cycle in G 1, agrees with our data showing that 5-LOX metabolites were weaker inhibitors of isolated TrxR in vitro (Table I). 5-HpETE is less efficient at generating 4-HNE via Hock cleavage, compared with 15(S)-HpETE and 12(S)-HpETE isomers. Formation of 3-Znonenal from 5-HpETE involves hydrogen abstraction from C13, which occurs less readily than hydrogen abstraction from C10 (9). 2 Nevertheless, induction of 5-LOX might modulate TrxR activity under other experimental conditions or in different cell backgrounds. So far, we have not succeeded at isolating an HEK-293 cell line that stably expresses a ponasterone responsive 12-LOX gene. We speculate that such cells would resemble Lox15-1 ϩ cells more than Lox5 ϩ cells because 15-LOX-1 can generate some 12(S)-HpETE and other related compounds (26).
To address these issues, we created cells that contain a 15-LOX-1 gene under the control of a ponasterone-responsive, heterologous transcription complex. This system (13) enables the selective induction of 15-LOX-1, and controls for effects from all endogenous metabolites, including primary hydroperoxide metabolites in vitro. This approach also minimizes the possibility of erroneous conclusions about 15-LOX-1 that might actually be due to pleiotropic effects of reagents like sodium butyrate or NSAIDS. Our results show that 15-LOX-1 negatively regulates the selenoenzyme TrxR, and this has functional consequences for cell proliferation. The chemically stable HETE and HODE metabolites of 15-LOX-1 inhibit purified TrxR enzyme, but our data suggest that the metastable HpETEs and HpODEs, or their rearrangement product 4-HNE, also contribute to the inhibition of TrxR in cells. HETEs or HODEs bind weakly and reversibly with cellular receptors, including PPAR␦ (36) or PPAR␥ (31,32,37,44). If HETEs or HODEs alone were the inhibitory molecules, then TrxR activity should have recovered promptly. However, inhibition of TrxR persisted for 8 h (Fig. 3B), consistent with irreversible, autoinactivation that accompanies lipid peroxidase catalysis (45)(46)(47), or with irreversible inactivation by 4-HNE reported for other enzymes (48) and selenoenzymes (49).
In human cancers, expression of 15-LOX-1 correlates positively with the aggressiveness of prostate tumors (25), but not with colorectal tumors (35,32,50). The relationship between 15-LOX-1 and colorectal cancer is elusive. Ikawa et al. (50) found elevated expression of 15-LOX-1 in ϳ 60% (13/21) of colon tumors. In contrast, Shureiqi et al. (35) found lower levels of 13(S)-HODE in 15/18 (ϳ 83%) colon tumors compared with normal tissue. Chen et al. (32) found that the concentration of 15(S)-HETE in serum was ϳ60% lower in patients with colon cancer compared with normal patients. Our results are consistent with those reported by Shureiqi et al. (35) and Chen et al. (32), and with the hypothesis that loss of 15-LOX-1 expression favors malignant progression in colon tissue (17,34,39,35,32). For instance, loss of 15-LOX-1 would nullify its regulation of TrxR, the G 1 cell cycle checkpoint, and BAX expression. Our results are compatible with a model of colorectal cancer progression in which 15-LOX-1 opposes the actions of COX-2. In the basal, healthy state, 15-LOX-1 and the COX-1 isoenzyme coordinately govern proliferation and differentiation. In the diseased state, the acquisition of excess COX-2 isoenzyme and the loss of 15-LOX-1 propel oncogenesis. NSAIDs slow colon cancer progression because they inhibit COX-2 catalysis, eliminating an oncogenic stimulus. NSAIDS also induce 15-LOX-1, providing a growth suppressing effect.
TrxR regulates tumor suppressors like p53, as well as oncogenes like HIF1 (3,4,8). TrxR also regulates enzymes like ribonucleotide reductase that influence DNA repair, as well as the cell cycle. Thus, 15-LOX-1 could inhibit TrxR consistently in different cell types, but this could translate into disparate effects. Although speculative, this might explain why 15-LOX-1 restrains tumor progression in some cancers, while enabling it in others.