Novel n-3 Fatty Acid Oxidation Products Activate Nrf2 by Destabilizing the Association between Keap1 and Cullin3*

Consumption of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) can mitigate the progression of diseases in which oxidative stress represents a common underlying biochemical process. Nrf2-regulated gene expression regulates detoxification of reactive oxygen species. EPA and DHA were subjected to an in vitro free radical oxidation process that models in vivo conditions. Oxidized n-3 fatty acids reacted directly with the negative regulator of Nrf2, Keap1, initiating Keap1 dissociation with Cullin3, thereby inducing Nrf2-directed gene expression. Liquid chromatography-tandem mass spectrometry analyses of oxidized EPA demonstrated the presence of novel cyclopentenone-containing molecules termed J3-isoprostanes in vitro and in vivo and were shown to induce Nrf2-directed gene expression. These experiments provide a biochemical basis for the hypothesis that formation of J-ring compounds generated from oxidation of EPA and DHA in vivo can reach concentrations high enough to induce Nrf2-based cellular defense systems.

oil. Epidemiological studies and randomized controlled trials have demonstrated that n-3 fatty acid supplementation can reduce sudden cardiac death, nonfatal stroke, and decrease progression of arteriosclerosis (1)(2)(3). Dietary supplementation with fish oil also shows promise for suppressing progression of neurodegenerative diseases (4), neuropsychiatric disorders (5), human immunodeficiency virus (6), and diseases of the retina (7). Thus, identifying the molecular mechanism(s) responsible for n-3 fatty acid mitigation of these diverse diseases represents an important and intriguing question.
Oxidative stress is a common factor in the etiology of the diseases impacted by DHA and EPA (1, 8 -17). These n-3 fatty acids are very susceptible to free radical oxidation, exceeding that of arachidonic acid (AA) (18 -20). Yet, recent studies have shown that EPA and DHA supplementation reduced urinary F 2 -isoprotane levels, a marker for oxidative stress, as well as enhanced cellular antioxidant defense systems (21)(22)(23)(24). Although a reduction of F 2 -isoprostane levels can be attributed, in part, to a decrease in membrane AA content (20,25), the relationship between EPA/ DHA oxidation versus EPA/DHA-mediated induction of antioxidant responses is not well understood.
Non-enzymatic free-radical peroxidation of AA results in the formation of multiple stereoisomers of prostaglandin H 2 -like bicyclic endoperoxides that can undergo thiol-mediated reduction to form prostaglandin-like compounds termed F 2 -isoprostanes (F 2 -IsoPs) or rearrangement and reduction to form E 2 -IsoPs, D 2 -IsoPs, and isothromboxanes. E 2 /D 2 -IsoPs can then dehydrate to highly reactive A 2 /J 2 -isoprostanes containing cyclopentenone rings. These compounds readily adduct thiol groups on proteins and have been shown to induce Nrf2/ ARE-directed gene expression (26).
NF-E2-related factor 2, Nrf2, is a master transcription factor shown to regulate expression of more than 200 genes, including those involved in Phase II detoxification and antioxidant gene expression (27). Regulation of Nrf2 activity represents a critical step in initiating a cellular antioxidant response to reactive oxygen species. Nrf2-dependent gene expression is regulated by a cis-element located in the proximal promoter region and is termed an antioxidant response element (ARE) (28). Nrf2 binds to AREs as a heterodimer and regulates ARE-directed gene expression (29). Nrf2 is negatively regulated by association with the ubiquitin ligase adaptor protein Keap1 (30). Binding of Keap1 to Nrf2 directs the ubiquitination and proteasome-dependent degradation of Nrf2. Inhibition of Nrf2 ubiquitination results in Nrf2 accumulation, increased ARE-directed gene expression, and an enhanced ability to respond to oxidant stress (31).
Non-enzymatic free-radical peroxidation of n-3 fatty acids are predicted to generate reactive species analogous to those formed from AA. It would be of significant interest to identify those analogous species and elucidate their bioactivity. We report here on novel studies that demonstrate the formation of cyclopentenone-containing J-ring isoprostanes following oxidation of n-3 fatty acids. Furthermore, these studies show that J 3 -IsoPs stabilize Nrf2 levels and induce Nrf2-directed gene expression. Whereas, AA-derived 15-deoxy-⌬12,14-prostaglandin J 2 was hypothesized to activate Nrf2-directed gene expression via dissociation of Nrf2 from adducted Keap1 (26), our data support an alternative mechanism: dissociation of Cul-lin3 from oxidized Keap1.
Cell Culture-Human HepG2 cells stably expressing GFP under control of a murine GSTA1 ARE/thymidine kinase reporter were a gracious gift from Dr. W. E. Fahl (ProCertus BioPharm). HepG2 and Cos-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.
Animal Care-Mice were fed an AIN-76A diet and water ad libitum and housed under controlled conditions (12-h light/ dark periods). All experimental protocols conducted on the mice were performed in accordance with the standards established by the U.S. Animal Welfare Acts, as set forth in National Institutes of Health guidelines and the Policy and Procedures and the Vanderbilt University Institutional Animal Care and Use Committee.
Preparation of Oxidized Fatty Acid-EPA and DHA were oxidized by either AAPH in phosphate-buffered saline buffer (20) or MeOAMVN in benzene (32). Both AAPH and MeOAMVN are free radical initiators. The concentrations denoted in Figs. 1-3 and 6 represent the initial free acid concentrations of EPA or DHA that were subjected to oxidation.
Preparation of Nuclear Extracts, Immunoprecipitation, and Immunoblot Analysis-Preparation of nuclear extracts from cells and immunoblotting of Nrf2, HO1, and GCLC were car-ried out as described (33). Immunoprecipitation was carried out as described previously (34). Cardiac tissue was obtained as described in a previous study (35). Slices of flash frozen cardiac tissue were sonicated at 4°C for 10 s in nuclear isolation buffer (33). Nuclei were pelleted following centrifugation at 12,000 ϫ g, washed twice in nuclear isolation buffer, and then solubilized in 5ϫ Laemmli buffer.
Preparation for Flow Cytometric Assay-Washed, trypsinized HepG2 ARE-GFP reporter cells were assayed using a BDLS RII by the Vanderbilt Institutional flow cytometry core.
Purification of Active Component in Oxidized EPA-EPA was oxidized by AAPH, and the oxidation mixture was extracted using a C-18 cartridge. The eluate was evaporated to dryness, and the residue was resuspended in isopropanol and purified by normal phase high-pressure liquid chromatography (HPLC) using a Phenomenex Luna 10 (10 mm ϫ 25 cm) Silica semipreparation column with a mobile phase of hexane/isopropanol/acetic acid (100/3/0.1, v/v) at 9 ml/min flow rate. The fraction was collected every minute. HepG2 cells were exposed to the HPLC fractions, and nuclear Nrf2 was measured by immunoblotting. The HPLC fraction that maximally induced nuclear Nrf2 accumulation was further analyzed by LC/MS/MS.
Purification and Analyses of A 3 /J 3 -IsoPs from Tissue-Liver tissues were collected from mice fed with 6% fish oil (Dyets, Inc.) for 8 weeks. Phospholipids in the livers were extracted by Folch solution and hydrolyzed by phospholipase A 2 to release free A 3 /J 3 -IsoPs. The A 3 /J 3 -IsoPs in the samples were then analyzed by LC/MS/MS.
Structural Characterization Using Tandem Mass Spectrometry-The in vitro and in vivo samples were analyzed using a ThermoFinnigan TSQ Quantum 1.0 SR 1 mass spectrometer in negative ion mode. For tandem MS experiments, collision energies were optimized between 10 and 30 eV under 1.5 millitorr of argon. Spectra that are shown were obtained at 15eV. Data acquisition and analysis were performed using Xcaliber software, version 1.3. For electrospray ionization-MS, the capillary temperature was 300°C. The vaporizer temperature is 350°C. The sheath and aux gas pressure were 31 and 17 p.s.i. The spray voltage was 3.0 kV, and the tube lens voltage was Ϫ102 V. Samples were introduced by reversed-phase HPLC with a Phenomenex Luna ODS 5 (2.1 mm ϫ 25 cm) column. The mobile phase was a gradient starting with 80% solvent A (2 mM NH 4 Ac:CH 3 CN/MEOH(95/5) ϭ 95:5) and 20% solvent B (MeOH:CH 3 CN ϭ 5:95), holding for 1 min and increasing to 50% B in 18 min and holding for 2 min. The flow rate was 0.2 ml/min.
Gas chromatography (GC)/negative ion chemical ionization MS was performed using an Agilent technologies 6890N network GC system. Methane was used as the carrier gas at a flow rate of 2 ml/min. Ion source temperature was 250°C, electron

n-3 Fatty Acid Oxidation and Nrf2
energy was 70 eV, and filament current was 0.25 mA. For analysis, samples were dissolved in 20 l of undecane.
Reactivity of Keap1 Sulfhydryls-The reactivity of sulfhydryl groups in Keap1 was determined in a competition reaction occurring between oxidized EPA and 5,5Ј-dithiobis(2-nitrobenzoic acid) (DTNB). Recombinant Keap1 (0.2 mg/ml) was reacted at 30°C with 100 M DTNB in argon-saturated 100 mM Tris, pH 8.0, supplemented with either carrier control solvent or 100 M oxidized EPA. The initial relative absorbance of the two reactions was set to zero and then monitored simultaneously at A 412 using a dual-beam Varian DMS 80 spectrophotometer. Competition between oxidized EPA and DTNB for Keap1 sulfhydryls was recorded as a decrease in absorbance at A 412 , as a function of time relative to the absorbance produced in the reference cell by reaction of DTNB and Keap1 in the absence of oxidized EPA. Control experiments demonstrated that oxidized EPA did not react with DTNB. DTNB reactivity with Keap1, measured as an increase in absorbance at A 412 demonstrated that all 27 His 6 -Keap1 cysteine residues were available for titration.
CD Spectroscopy-CD spectra were acquired on a Jasco J-810 spectropolarimeter equipped with a Peltier-controlled temperature block using a 1-mm path length quartz cuvette. The data were collected at 1-nm increments from 200 to 260 nm using a response time of 2 s and a scan speed of 10 nm/min. Spectra represent an average 2 scans with the background corrected against a buffer blank or a buffer treatment compound. Data are reported as mean residue ellipticity, [](deg ϫ cm 2 ϫ dmol Ϫ1 ), which was calculated as [] ϭ 100 (signal)/(C ϫ n ϫ l), where deg represents degrees; signal represents raw output in millidegrees; C represents protein concentration in millimolar; n rep-resents number of amino acid residues; and l represents cell path length in centimeters.
hKeap1 was dissolved at a concentration of 0.1 mg/ml in 100 mM Tris buffer (pH 8.0). For the CD measurements, hKeapl was incubated with 10 M 15-J 2 -IsoP or 100 M oxidized EPA at 37°C for 1 h. Percentage of secondary structure was calculated using the K2d program (37). Structural changes in ␣-helix, ␤-sheet, and random coil were expressed as the increase/decrease of the percentage of structure for 15-J 2 -IsoP or oxidized EPAtreated hKeap1 compared with non-treated hKeap1.

RESULTS
Oxidation of n-3 polyunsaturated fatty acids is required for induction of ARE-directed GFP reporter gene expression. HepG2 cells stably transfected with a plasmid that expressed GFP under control of the murine Gsta1 ARE (38) were used to determine if n-3 fatty acids were capable of inducing Nrf2-directed gene expression. The data presented in Fig. 1 (A and B) show that a well characterized Nrf2 activator such as sulforaphane (31) produced a statistically significant increase in AREreporter directed GFP fluorescence (p Ͻ 0.05, analysis of variance). However, exposure to DHA or EPA free acids, did not affect GFP fluorescence compared with control (p Ͼ 0.05, analysis of variance).
Because EPA and DHA readily undergo auto-oxidation in vivo (39), it was of interest to determine if oxidized n-3 fatty acids exhibited the ability to induce Nrf2-regulated activity. Thus, EPA and DHA were oxidized in vitro by free radical initiation (see "Experimental Procedures"). HepG2 cells were exposed to the EPA or DHA oxidation products. A dose-dependent increase in ARE-reporter-directed GFP fluorescence was observed (Fig. 1, A and 1B, p Ͻ 0.05, analysis of variance). These data demonstrate that EPA and DHA oxidation products induce ARE-directed reporter gene activity.
Oxidized n-3 fatty acids stabilize Nrf2 and increase the endogenous expression of Nrf2-regulated genes. Nrf2 is a short-lived protein due to its association with Keap1, which directs the ubiquitination of Nrf2 and subsequent proteasomedependent degradation (40). Under basal conditions, very little Nrf2 accumulates in the nucleus (30). Agents that activate Nrf2 function do so by inhibiting Keap1-directed ubiquitination of Nrf2, allowing Nrf2-regulated gene expression to increase. Inhibition of ubiquitination and stabilization of Nrf2 can be followed by measuring nuclear accumulation of Nrf2 (31).
HepG2 cells were exposed to oxidized DHA (DHA ox ) for up to 30 h, and nuclear protein was isolated and immunoblotted for the presence of Nrf2 (Fig. 1C). Nuclear Nrf2 levels The free acids of DHA and EPA were added directly to cells. After 16 h of exposure, GFP fluorescence was measured. Cells exposed to 25 or 50 M sulforaphane were used as a positive control. The concentrations denoted refer to the initial starting concentrations that were subject to oxidation. HepG2 cells were also exposed to the oxidized n-3 fatty acids for the indicated times prior to immunoblotting (C and D).

n-3 Fatty Acid Oxidation and Nrf2
increased within 2 h of exposure to DHA ox . Similarly, HepG2 cells exposed to oxidized EPA (EPA ox ) showed significant increases in nuclear Nrf2 levels within 4 h, although the free acid form of EPA did not stabilize Nrf2 (Fig. 1D). Moreover, stabilization of Nrf2 was followed by the increased expression of two well characterized endogenous Nrf2-regulated genes, heme oxygenase-1 (HO1) (41) and Gclc (42) (Fig. 1, C  and E). These results are consistent with the data presented in Fig. 1 (A and B) and suggest that oxidation products of DHA and EPA result in Nrf2 stabilization, allowing increased Nrf2-regulated gene expression.
A siRNA approach was used to determine if oxidized n-3 fatty acid-mediated expression of HO1 was dependent upon Nrf2. NMuMG cells were transiently transfected with either control siRNA or siRNA directed against Nrf2. 48 h after transfection, cells were exposed to DHA ox for 4 h, and then nuclear extracts were prepared for immunoblotting. The immunoblot was probed with antibodies directed against Nrf2, HO1, and actin ( Fig. 2A). Exposure to DHA ox produced a robust increase in Nrf2 expression in cells transfected with control siRNA. Gene silencing with siRNA directed against Nrf2 suppressed DHA ox -mediated expression of Nrf2 by 50% (Fig. 2B). In cells transfected with control siRNA, DHA ox also produced a robust increase in the HO1 expression. The intensity of the HO1 immunoreactive band was measured and expressed relative to the expression of actin (Fig. 2B). Transfection of siRNA directed against Nrf2 suppressed DHA ox -mediated expression of HO1 by 50% (Fig. 2, A and B). These experiments demonstrate a direct correlation between Nrf2 and HO1 expression and support the contention that oxidized n-3 fatty acids-mediated expression of HO1 is Nrf2-dependent.

Oxidized n-3 Fatty Acids React with Keap1 Sulfhydryls, Altering Keap1 Secondary Structure That Is Followed by a Loss of
Association between Keap1 and Cullin3-It was previously concluded that adduction of Keap1 by 15-deoxy-⌬12,14-prostaglandin J2-initiated release of Nrf2, allowing for the latter's stabilization (26). To determine if a similar mechanism prevailed for n-3 omega fatty acids, experiments were undertaken to determine if EPA ox could reacted directly with Keap1 cysteine residues. The reaction of cyclopentenone isoprostanes with protein sulfhydryls can be monitored using DTNB titration (43). This assay was modified to increase sensitivity such that we could monitor the reactivity of a small number of sulfhydryls. The ability of EPA ox to compete with DTNB for reaction with recombinant Keap1 sulfhydryls was monitored as described under "Experimental Procedures," and the data are presented in Fig. 3A. The decrease in absorbance measured at A 412 can be attributed to EPA ox modification of 4 Keap1 thiol groups, rendering them unavailable

n-3 Fatty Acid Oxidation and Nrf2
for titration with DTNB. Taken together, the LC-MS-MS (see below) and DTNB titration data are consistent with the model proposed by (44) in which an inducer transiently reacts with Keap1 cysteine residues that ultimately undergo disulfide bond formation. CD was used to probe for secondary structural changes produced by EPA ox interaction with Keap1 residues (Fig. 3B). Based on work described below, we also interrogated secondary structural changes in Keap1 following reaction with 15-J 2 -IsoP. The analysis indicated that Keap1 underwent an isoprostane-mediated loss of ␣-helix content; coincident with an increase in ␤-sheet content and random coil structure. The EPA ox -mediated loss in ␣-helix conformation may be consequence of a change in the BTB domain ␣-helix region (45), because the kelch domain does not contain ␣-helix folding (46). Currently, there is no information concerning the structure of the intervening domain.
Keap1 BTB and intervening domains are essential for association with the E3 ligase scaffold protein Cullin3 (47). Loss of Keap1-Cullin3 association inhibits Nrf2 ubiquitination, thereby stabilizing Nrf2 and initiating Nrf2-directed gene expression (47,48). COS-7 cells, transiently co-transfected with plasmids expressing Keap1 and V5-tagged Cullin3, were exposed to solvent control or 50 M EPA ox for 2 h (Fig. 3C). Immunoprecipitation with V5 antibody followed by immunoblotting with antibody to Keap1 demonstrated that Keap1 and V5-Cullin3 associated in a complex under control conditions where Nrf2 immunoreactive protein was barely detected (Fig.  1D). Exposure to EPA ox , which resulted in Nrf2 stabilization (Fig. 1D), disrupted the association between Keap1 and V5-Cullin3 (Fig. 3C). Taken all together, the data presented in Fig. 3 indicate that EPA ox modification of Keap1 sulfhydryls was followed by a change in secondary structure that was accompanied by a loss of association with Cullin3 and provide a mechanism for Nrf2 stabilization observed in Fig. 1.
A Series of Novel Cyclopentenone-containing Molecules Termed J 3 -isoprostanes Generated from Non-enzymatic Peroxidation of EPA Are Potent Activators of Nrf2-Major free radical-catalyzed oxidation products of AA, including F 2 -IsoPs, D 2 /E 2 -IsoPs, and A 2 /J 2 -IsoPs, have been characterized previously (49). EPA and DHA also undergo analogous free radicalinduced oxidation, however significantly more regioisomers result (20,50). In the presence of the free radical initiators, AAPH or MeOAMVN, EPA and DHA are oxidized to bicyclic endoperoxides with dozens of regioisomers and diastereomers. These bicyclic endoperoxides can be reduced to F-ring isoprostanes by GSH (in vivo) (51) and PPh3 (in vitro) (52) or rearranged to D/E-ring isoprostanes in aqueous solution (53). To identify the active component(s) in oxidized n-3 fatty acids, we oxidized EPA in phosphate-buffered saline buffer, separated the oxidation mixture by normal phase HPLC, and collected 80 fractions. We then examined the ability of each HPLC fraction to induce nuclear accumulation of Nrf2, as measured by immunoblotting (data not shown). The oxidation products in the HPLC fraction that maximally induced Nrf2 accumulation in the nucleus were further identified by LC/MS/MS. The mass spectrometry analysis of this fraction showed the major ion in this fraction was m/z 331 (Fig. 4A). A series of peaks can be observed when m/z 331 was monitored by second ion mass spectrometry (Fig. 4B), suggesting the presence of different regioisomers in this fraction. We subsequently purified the A 3 /J 3 -IsoP regioisomers shown in Fig. 4B, exposed HepG2 cells to them for 16 h and measured Nrf2/ARE-directed GFP fluorescence. As shown in Fig. 5, A 3 /J 3 -IsoP regioisomers were able to produce a significant increase in GFP fluorescence compared with control (p Ͻ 0.05).
We then performed collision-induced dissociation (CID) analysis on the m/z 331 ion to get more structural information (Fig. 4C). The CID of the parental ion at 331 resulted in the formation of a number of relevant product ions that would be predicted to be common to all of the A 3 /J 3 -IsoP regioisomers (Fig. 4C) Although we have only demonstrated formation of A 3 /J 3 -IsoPs from oxidation of EPA, DHA should undergo a similar oxidation scheme. To support this contention, EPA and DHA were subjected to the oxidation process, and then PPh3 was added before applying the lipids to HepG2 cells. Addition of PPh3 to EPA ox and DHA ox resulted in the formation of F-ring final products rather than A or J cyclopentenones. As shown in supplemental Fig. S1, the product of DHA ox reduction by PPh3 was not able to activate the ARE-directed GFP reporter (p Ͼ 0.05, analysis of variance). Similarly, PPh3 reduction of EPA ox resulted in a loss of inducing activity (data not shown). These data are consistent with the hypothesis that A/J-ring compounds are required for the activation of ARE-directed reporter activity.
In addition, 15-F 2 -IsoP, 15-E 2 -IsoP, 15-A 2 -IsoP, and 15-J 2 -IsoP were used as surrogate compounds to further define the structure(s) responsible for activation of Nrf2. HepG2 cells were exposed to the surrogate compounds for 16 h, and Nrf2/ ARE-directed GFP fluorescence was measured (supplemental Fig. S2A). Exposure to 15-J 2 -IsoP produced a robust increase in GFP fluorescence compared with control (p Ͻ 0.05, Student's t test). Moreover, nuclear Nrf2 levels increased following exposure to 15-J 2 -IsoP for 4 h (supplemental Fig. S2B). These results are supported by the data shown in Fig. 3B and suggested that J-ring isoprostane structure is capable of Nrf2 activation.
A 3 /J 3 -isoprostanes Are Detected in Vivo-For these studies, mice were fed a diet supplemented with 0.56% EPA for 8 weeks. We did not observe any significant formation of IsoPs in the chow, which was changed daily, during the 8 weeks of feeding (data not shown). AA, EPA, F 2 -, and F 3 -isoprostane levels were quantitated per gram of cardiac tissue. In non-stressed conditions, cardiac tissue obtained from mice fed a control diet contained 0.35 (Ϯ0.07) pg of F 2 -IsoPs per nmol of AA. Nei-

n-3 Fatty Acid Oxidation and Nrf2
ther EPA nor F 3 -IsoPs were detected in cardiac tissue obtained from the control mice. In non-stressed conditions, cardiac tissue obtained from mice fed a diet supplemented with 0.56% EPA contained 25 (Ϯ5) pg of F 3 -IsoPs per nmol of EPA and 0.40 (Ϯ0.08) pg of F 2 -IsoPs per nmol of AA. These measurements demonstrate that EPA is significantly more susceptible to oxidation than AA and provide support for the concept that oxidation of EPA would result in significant generation of oxidant-mediated A 3 /J 3 -IsoPs.
Analogous to LC/ESI/MS/MS studies performed in vitro, experiments were then carried out to obtain evidence that A 3 /J 3 -IsoPs are detected in vivo. Liver and heart were extracted and utilized for A 3 /J 3 -IsoPs analysis by LC/MS/ MS. In the CID analysis of the liver, all of the product ions that would be predicted to be common to all of the A 3 /J 3 -IsoP regioisomers were detected (Fig. 4D) Fig. 4D, the CID pattern is very similar to that obtained from the in vitro studies, although

n-3 Fatty Acid Oxidation and Nrf2
the relative abundance of the different product ions are slightly different (compare with Fig. 4C). A similar result was obtained from heart tissue (data not shown). Taken together, these experiments provide evidence that A 3 /J 3 -IsoPs are present in vivo.
n-3 Fatty Acids Stabilize Nrf2 in Vivo-As shown in Fig. 6, a bolus intraperitoneal injection of DHA ox resulted in stabilization of Nrf2 in heart tissue. Similar results were observed if animals were injected with free acid DHA, consistent with the occurrence of in vivo oxidation (data not shown).

DISCUSSION
Oxidative stress represents a common underlying biochemical process that contributes to the progression of many diseases mitigated by dietary supplementation with DHA and EPA (1, 8 -14). Although EPA and DHA are more easily oxidized compared with arachidonic acid, DHA/EPA supplementation has been associated with the induction of cellular antioxidant responses (57)(58)(59). However, the biochemical pathways responsible for these effects are currently not well understood.
The generation of isoprostanes reflecting the non-enzymatic process of lipid peroxidation of arachidonic acid, EPA, or DHA is a major consequence of oxidative stress, as exemplified by formation of F 2 -, F 3 -, and F 4 -isoprostanes (19,20,60). Oxidative stress-mediated formation of isoprostanes represents an important biochemical parameter, because the different isoprostanes can exhibit various bioactive functions: F 2 -IsoPs can increase expression of the cytokine interleukin-8 (61) and stimulate the thromboxane receptor (62), whereas 15-J 2 -IsoP can activate the peroxisome proliferator-activated receptor ␥ nuclear receptor (63), as well as ARE-directed Phase II gene expression (26).
The transcription factor NF-E2-related factor 2 (Nrf2) plays an important role in regulating the expression of genes that encode proteins responsible for decreasing oxidative stress and inflammation in vivo (64 -67). Nrf2 is negatively regulated by association with Keap1, a Cullin3 ubiquitin ligase adaptor protein (47,48). Keap1 function, in turn, requires association with Cullin3, a scaffold protein that positions Keap1 and its substrate in proximity to the E3 ligase Rbx1, allowing the substrate to be polyubiquitinated. Cullin3 is regulated by neddylation and deneddylation (34,68). Human Keap1 contains 27 cysteine residues hypothesized to function as sensors of oxidative stress (31,69). However, adduction of Keap1 cysteine residues does not induce release of Nrf2 (70). Nor does every reagent that adducts Keap1 induce Nrf2-mediated gene expression (33). Thus, the rate-limiting steps responsible for activation of Nrf2 are not well understood.
Levonen, et al. (26) found that 15-deoxy-⌬12,14-prostaglandin J 2 adducted Keap1and hypothesized that activation of Nrf2 by15-deoxy-⌬12,14-prostaglandin J 2 was a consequence of dissociation of Nrf2 from Keap1. However, subsequent work has shown that adduction of Keap1 by 15-deoxy-⌬12,14-prostaglandin J 2 does not cause disassociation from Nrf2 (70). Thus, the mechanism by which cyclopentenone-containing IsoPs activate Nrf2 has remained obscure. Our studies demonstrate a requirement for free radical-mediated oxidation of DHA and EPA for the process of Nrf2 stabilization, ARE-directed reporter gene expression, as well as for expression of endogenous HO1 and GCLC. We have presented novel data demonstrating that free radical-mediated oxidation products reacted with Keap1 sulfhydryls, altering Keap1 structure, a conformational change that was associated with loss of binding to Cul-lin3. These data identify the association of Keap1 with Cullin3 as a key regulatory step in Nrf2 biology with respect to activation by J 3 -IsoPs and, by extension, other cyclopentenone-containing IsoPs.
LC-MS-MS analysis of the oxidized EPA identified novel cyclopentenone-containing J 3 -IsoPs that induced Nrf2-mediated gene expression. Our analysis provided evidence that these novel IsoPs are present in vivo. Moreover, our studies suggest that the bioactivity of J 3 -IsoPs may come from the mixture of different of regioisomers and suggest that J-ring IsoPs are far more potent than A-ring IsoPs in activation of Nrf2. Taken all together the data suggest that J-ring structure is crucial for Nrf2 activation.
In vivo DHA and EPA are more susceptible to auto-oxidation than AA, because they possess more double bonds than AA. We found that the level of F-ring IsoPs generated from EPA and DHA is at least one order higher than that generated from AA after mice were supplemented with fish oil.  Immunoblot illustrating increased expression of endogenous Nrf2 in cardiac tissue following intraperitoneal injection of DHA ox . 6-week-old female Nrf2 wild-type C57BL/6J mice were injected intraperitoneally with either ETOH (lanes 1-4) or DHA ox (2 mg/animal, lanes 5-10). 6 h after injection, the mice were sacrificed. Hearts were removed and washed. Nuclei isolated from heart tissue and nuclear protein was immunoblotted for the presence of Nrf2 or actin.
Our data support the hypothesis that the formation of J-ring compounds generated from oxidation of EPA and DHA in vivo can reach concentrations high enough to induce Nrf2based antioxidant and Phase II detoxification defense systems.