Nitro-fatty acids are electrophilic fatty acids produced in vivo from nitrogen peroxide that have many physiological activities. We recently demonstrated that nitro-fatty acids activate the Keap1-Nrf2 system, which protects cells from damage owing to electrophilic or oxidative stresses via transactivating an array of cytoprotective genes, although the molecular mechanism how they activate Nrf2 is unclear. A number of chemical compounds with different structures have been reported to activate the Keap1-Nrf2 system, which can be categorized into at least six classes based on their sensing pathways. In this study, we showed that nitro-oleic acid (OA-NO2), one of major nitro-fatty acids, activates Nrf2 in the same manner that of a cyclopentenone prostaglandin 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) using transgenic zebrafish that expresses green fluorescent protein (GFP) in response to Nrf2 activators. In transgenic embryos, GFP was induced in the whole body by treatment with OA-NO2, 15d-PGJ2 or diethylmaleate (DEM), but not with hydrogen peroxide (H2O2), when exogenous Nrf2 and Keap1 were co-overexpressed. Induction by OA-NO2 or 15d-PGJ2 but not DEM was observed, even when a C151S mutation was introduced in Keap1. Our results support the contention that OA-NO2 and 15d-PGJ2 share an analogous cysteine code as electrophiles and also have similar anti-inflammatory roles.
The Keap1-Nrf2 system plays a central role in the cellular defense against electrophilic and oxidative stresses by orchestrating gene expression of detoxifying and antioxidant enzymes (Kobayashi & Yamamoto 2006; Kensler & Wakabayashi 2010). Nrf2 is a transcription factor that heterodimerizes with small Maf proteins and binds to the antioxidant responsive element (ARE)/electrophile responsive element (EpRE) within the regulatory region of its target genes. Keap1 is a substrate-specific adaptor protein for a Cul3-dependent E3 ubiquitin ligase complex that homodimerizes and interacts with the ETGE and DLG motifs of Nrf2. Under basal conditions, Nrf2 is maintained at low levels because of Keap1-dependent ubiquitination and proteasomal degradation. Upon exposure to electrophiles or oxidative stress, Nrf2 is stabilized and accumulates in the nucleus, where it transactivates ARE/EpRE-regulated genes.
A variety of Nrf2 activators have been reported (Kobayashi & Yamamoto 2005). Some of these activators have protective activities against carcinogenesis, neuronal damage and inflammation that can be ingested as dietary agents for the prevention and therapy of age-related diseases such as cancer, cardiovascular diseases, chronic inflammation and neurodegenerative diseases (Calabrese et al. 2008). However, positive roles of Nrf2 in cancer tumorigenesis and chemoresistance have recently been uncovered (Hayes & McMahon 2009; Lau et al. 2008). It will be important to understand the differences between the Nrf2 activators, in particular the mechanisms of Keap1-Nrf2 activation, the activation of other systems and which Nrf2-target genes they induce.
The new field has focused on the identification of sensor molecules that activate Nrf2. These activators have a common ability to modify sulfhydryl groups by alkylation, oxidation or reduction. This observation suggests that cells possess a primary sensor for Nrf2 activators that is equipped with highly reactive cysteine residues. Cys-151 of mouse Keap1 is a target site for many Nrf2 activators such as diethylmaleate (DEM) (Kobayashi et al. 2009; Sekhar et al. 2010). We recently found that cyclopentenone prostaglandins, 15d-PGJ2 and prostaglandin A2 (PGA2), target residues other than Cys-151 in Keap1, perhaps Cys-273 (Kobayashi et al. 2009). Furthermore, sensor proteins for H2O2, cadmium chloride and gold-compound auranofin were not Keap1 or Nrf2 (Kobayashi et al. 2009). The target selections of a variety of Nrf2 activators should be investigated comprehensively.
The Keap1-Nrf2 system is conserved among vertebrates, including zebrafish (Kobayashi et al. 2002; Takagi et al. 2004; Li et al. 2008). Among the known endogenous targets of zebrafish Nrf2, the pi-class glutathione S-transferase 1 gene (gstp1) showed the strongest induction in both electrophile-treated larvae and Nrf2-overexpressing embryos. The gene regulatory region of gstp1 was examined by a green fluorescent protein (GFP) reporter gene analysis using microinjection into zebrafish embryos, and an ARE/EpRE-like sequence located 30 bp upstream of the transcription initiation site was shown to be necessary and sufficient for the induction by Nrf2 (Suzuki et al. 2005).
In this study, we tried to generate stable transgenic zebrafish lines that express GFP in response to Nrf2 activators using a 3.5-kb gene regulatory region of gstp1 to develop a rapid and easy method for screening and classifying Nrf2 activators. Two stable transgenic lines that express GFP in the larval olfactory regions in response to Nrf2 activators were isolated. No GFP induction was detected in transgenic embryos, but strong induction in response to DEM and 15d-PGJ2, but not H2O2, was observed when both Nrf2 and Keap1 were overexpressed. Using this system, we classified a newly identified Nrf2 activator, OA-NO2, into the same category as 15d-PGJ2.
Generation of stable transgenic lines that express GFP in response to Nrf2 activators
In transient assays, GFP expression from the p3.5gstp1GFP construct, which contains a 3.5-kb promoter region of the zebrafish gstp1 gene, is strongly transactivated by Nrf2 in zebrafish embryos (Fig. 1A; Suzuki et al. 2005). Stable transgenic lines carrying this construct were isolated by genotyping F1 embryos from p3.5gstp1GFP-injected founders. Although three stable lines were isolated, none of the three lines exhibited GFP expression upon treatment with DEM, thus suggesting a position effect of the transgene (data not shown; Table S1 in Supporting Information). To circumvent this problem, a highly efficient Tol2 transposon system (Kawakami et al. 2004) was used to generate additional stable lines. A pT3.5gstp1GFP construct was made by introducing Tol2 sequences into p3.5gstp1GFP and were co-injected into zebrafish embryos with mRNA encoding Tol2-specific transposase. Injected founders (n = 104) were raised and 12 transgenic lines, which showed GFP expression in F1 larvae, were isolated (Table S1 in Supporting Information). Out of 12 Tg(-3.5gstp1:GFP) lines, two lines exhibited GFP induction in response to DEM, and the rest displayed only basal GFP expression. In the larvae of these two lines (it416a and it416b), strong GFP induction was observed in the olfactory regions as for endogenous gstp1, whereas constitutive GFP expression was also detected in the lens, ears, fins, pericardium and lateral lines (Figs S1 and S2 in Supporting Information). We used the it416b line for further experiments because it exhibited stronger GFP induction than the it416a line.
In day 4 larvae, induction of endogenous gstp1 expression began approximately 3 h after DEM treatment and reached maximum expression levels approximately 6–9 h (data not shown). GFP induction in the olfactory regions of Tg(-3.5gstp1:GFP)it416b larvae was first detected at 6 h after DEM treatment and later became stronger (Fig. 1B; Movie S1 in Supporting Information). This induction was greatly reduced when nrf2 was knocked down with the nrf2-specific antisense morpholino oligonucleotide (nrf2MO) (Kobayashi et al. 2002), suggesting that GFP induction was mediated by Nrf2 (Fig. 1C).
Quantification of GFP induction
To quantify the level of the GFP expression, a fluorescent intensity in the olfactory regions was measured using an imaging analyzer equipped with the BZ-8000 microscope (Keyence) (Fig. 2A). The average fluorescent intensities of DEM-induced GFP expression in the olfactory regions were measured using 3, 10 or 30 individual larvae, and all of these showed similar values (Fig. 2B). We therefore used three individuals for quantification of the GFP induction in further analyses. According to this quantification, the reduction in DEM-induced GFP expression by the nrf2-knockdown was calculated as one-third (Fig. 2C).
We first quantified the time- and dose-dependencies of DEM-induced GFP expression. GFP expression in the olfactory regions of Tg(-3.5gstp1:GFP)it416b larvae treated with various DEM concentrations was measured at 3, 6, 10 and 14 h after treatment (Fig. 2D). GFP induction is detectable with treatment over 30 μm DEM, which is similar for detecting endogenous gstp1 induction by a real-time reverse transcription (RT)-PCR examination (Fig. 2E) or by a whole-mount in situ hybridization (data not shown). The onset of GFP induction (6 h) was later than that of endogenous gstp1 induction (3 h), most likely due to the time required for the accumulation of GFP protein. Altogether, these results suggest that the measurement sensitivity of DEM-induced GFP expression using the fluorescence intensities in the olfactory regions is comparable to that of endogenous gstp1 induction in the whole body using a real-time RT-PCR analysis.
Quantification of reduced GFP induction in the it567 mutant larvae
We have screened and isolated zebrafish mutants that have defects in gstp1 induction by Nrf2 activators (Kobayashi et al. 2009). One of these mutants, it567, has reduced induction of gstp1 in response to DEM. To visualize this reduction in living larvae, we crossed it567 fish with Tg(-3.5gstp1:GFP)it416b fish. As expected, DEM-induced GFP expression in the olfactory regions of it567 homozygous larvae was significantly reduced compared to that of sibling wild-type larvae (Fig. 3; Movie S2 in Supporting Information). The result indicates that homozygous mutant larvae are easily distinguishable from wild-type siblings by monitoring GFP induction upon treatment with DEM, although we have not identified the gene responsible for the it567 mutation.
GFP induction by a variety of Nrf2 activators
To elucidate whether Tg(-3.5gstp1:GFP)it416b can monitor the induction by a variety of Nrf2 activators, five Nrf2-activating compounds in different classes were examined (Kobayashi et al. 2009). Day 4 larvae of Tg(-3.5gstp1:GFP)it416b fish were treated with sulforaphane, 15d-PGJ2, H2O2, auranofin or CdCl2, and the GFP expression in the olfactory regions was monitored in three individuals (Fig. 4A). GFP induction was observed in larvae treated with all tested Nrf2 activators, although it was a bit weaker upon induction with H2O2 and CdCl2. The results suggest that Tg(-3.5gstp1:GFP)it416b can be used for the analysis of Nrf2 activator sensing mechanisms and also to screen for novel and unidentified activators.
We recently classified eleven Nrf2 activators based on the requirement of these activators for an Nrf2 response (Kobayashi et al. 2009). The classification was carried out using zebrafish embryos overexpressing Nrf2 and Keap1. The embryos were treated with each Nrf2 activator, and the induction of the endogenous gstp1 gene was analyzed by a whole-mount in situ hybridization or by an RT-PCR analysis. It will be time- and labor-saving if Tg(-3.5gstp1:GFP)it416b can be used for this analysis. To test this possibility, Nrf2 was overexpressed by an mRNA injection in embryos obtained from crossing heterozygous transgenic males and non-transgenic females and GFP expression was analyzed during gastrulation stages (Fig. S3 in Supporting Information). Robust GFP expression was observed in the embryos injected with Nrf2 mRNA, whereas no GFP expression was observed in uninjected embryos.
We then generated homozygous transgenic lines to make all the injected embryos positive for GFP. Embryos obtained from homozygous transgenic males and non-transgenic females were used for the further analyses, because maternal GFP expression was observed in embryos derived from transgenic females (data not shown). When wild-type or C151S mutant Keap1 was overexpressed with Nrf2, no GFP expression was observed, indicating that Keap1 suppressed GFP-inducing activity of Nrf2 (Fig. 4B). Treatment of DEM canceled the Nrf2-suppression by wild-type Keap1 but not by C151S Keap1, whereas 15d-PGJ2 released Nrf2 activity from both. H2O2 treatment had no effect at all. These results suggest that Keap1 is a sensor for DEM and 15d-PGJ2, but not for H2O2, and Cys-151 is required for DEM but not 15d-PGJ2 sensing. All these results were identical as in the case of the previous analyses conducted using endogenous gstp1 induction (Kobayashi et al. 2009).
OA-NO2 activates Nrf2 as 15d-PGJ2
OA-NO2 was next analyzed as an uncharacterized Nrf2 activator using Tg(-3.5gstp1:GFP)it416b. OA-NO2 is a nitro-fatty acid that has been recently shown to activate Nrf2 (Freeman et al. 2008). Nitro-fatty acids are electrophilic fatty acids produced in vivo from nitrogen peroxide that have many physiological activities, such as cGMP-dependent vessel relaxation, the inhibition of the inflammatory cell function, the induction of heme oxygenase 1 expression, the inhibition of NFκB, peroxisome proliferator-activated receptor (PPAR) activation and Nrf2 activation. GFP expression in the olfactory regions of Tg(-3.5gstp1:GFP)it416b larvae was analyzed after treatment with OA-NO2 at 4 days postfertilization (dpf) (Fig. 5A,B). A significant GFP induction was observed in the olfactory regions of OA-NO2-treated larvae and was decreased upon Nrf2 knockdown after injecting nrf2MO. GFP induction was also observed with OA-NO2 treatment of embryos overexpressing Nrf2 and Keap1 (Fig. 5C). Significantly, the Keap1 C151S mutation could not eliminate this induction, suggesting that Cys-151 is not a sensing site for OA-NO2 in the Keap1-Nrf2 system. Identical results were obtained when the induction of endogenous gstp1 was analyzed by RT-PCR (Fig. 5D). We also showed that a Cys-to-Ser mutation in Cys-288, another key reactive cysteine in Keap1, had no effect on the response to OA-NO2 (Fig. S4 in Supporting Information). It must be noted that these are same results as in the case of 15d-PGJ2 (Kobayashi et al. 2009).
To determine the sites in Keap1 modified by OA-NO2, mouse Keap1 protein treated with or without OA-NO2 was digested with trypsin and analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS). Peptide mass mapping by an MALDI-TOF MS analysis of the trypsin fragments of native mouse Keap1 made it possible to identify the peptides, thus accounting for approximately 80% of the protein sequence (see Table S2 in the Supporting Information). Compared with the calculated masses of the unmodified peptides, modified peptides P-1 to P-6 had an increased mass of +327.4 Da, thus corresponding to the addition of a single equivalent of OA-NO2 (Fig. 6). The sequences and masses of the peptides were as follows: P-1 (CHALTPR, m/z =1124.8), P-2 (SGVGVAVTMEPCR, m/z =1632.7), P-3 (LNSAECYYPER, m/z =1671.7), P-4 (LSQQLCDVTLQVK, m/z = 1802.1), P-5 (SGLAGCVVGGLLYAVGGR, m/z = 1976.2) and P-6 (QEEFFNLSHCQLATLISR, m/z =2463.3), indicating that mouse Keap1 is modified by OA-NO2 at Cys-77, Cys-226, Cys-273, Cys-368, Cys-489 and Cys-613. It should be noted that Cys-273 was included in the amino acids to which 15d-PGJ2 and PGA2 were shown to bind in our previous analyses (Fig. 6B; Kobayashi et al. 2009).
Another criterion of the classification is dependency to it275 gene. it275 is a zebrafish mutant, which has defects in the response to some, but not all, Nrf2 activators. For example, induction of endogenous gstp1 was decreased in it275 larvae in response to 15d-PGJ2, auranofin and CdCl2 and was similar with wild-type larvae in response to DEM, sulforaphane and H2O2. gstp1 expression was analyzed in it275 larvae or sibling wild-type larvae with or without OA-NO2 treatment at 4 dpf by an in situ hybridization (Fig. 7). OA-NO2-dependent gstp1 induction in gills or olfactory regions was dramatically reduced in it275 larvae compared to wild-type larvae. The reduction was not observed in larvae treated with DEM. This result suggests that Nrf2 activation mediated by OA-NO2 requires the it275 gene, which is also similar to 15d-PGJ2.
Taken together, OA-NO2 was classified as a class 4 Nrf2 activator, as are 15d-PGJ2 and PGA2 (Kobayashi et al. 2009). We thus hypothesize that nitro-fatty acids and cyclopentenone prostaglandins share strategies to activate the Keap1-Nrf2 system.
We previously demonstrated that nitrolinoleic acid (LNO2), a nitro-fatty acid similar to OA-NO2, also activates Nrf2 in cultured cells (Villacorta et al. 2007). The present study verifies that nitro-fatty acids are indeed effective Nrf2 activators in living animals. The covalent binding of OA-NO2 with reactive cysteines and histidines has been demonstrated previously in a variety of proteins (Batthyany et al. 2006). It is quite possible that OA-NO2 directly binds Keap1, most likely at Cys-273. Importantly, OA-NO2 appears to belong to the same Nrf2 activator class as 15d-PGJ2 as we have described herein. Strikingly, the chemical structures of nitro-fatty acids (OA-NO2 and LNO2) and cyclopentenone prostaglandins (15d-PGJ2 and PGA2) are similar (Fig. 8A). We speculate that unique structures common between these compounds are essential for targeting identical sites in the Keap1 protein. Besides activating Nrf2, nitro-fatty acids can activate PPARγ as an endogenous ligand with covalent binding to Cys-285 (Schopfer et al. 2005, 2010) and can inhibit NFκB-DNA binding through direct interaction with p65 (Cui et al. 2006). Interestingly, 15d-PGJ2 also exhibits these properties via direct binding to PPARγ at Cys-285 (Shiraki et al. 2006) and p65 of NFκB at Cys-38 (Straus et al. 2000). It is plausible that OA-NO2 can also bind to Cys-38 of NFκB p65. We hypothesize that electrophilic signaling compounds, such as nitro-fatty acids and cyclopentenone prostaglandins, can bind to a series of key target proteins (e.g., Keap1, PPARγ and NFκB) and thereby exert their physiological functions (for example, anti-inflammatory effects) by the combined actions of their target proteins (Fig. 8B) (the cysteine code hypothesis, Kobayashi et al. 2009).
The Tg(-3.5gstp1:GFP)it416b line established in the current study can monitor electrophile-induced gene expression in vivo and will be useful for many situations, for example in gene analyses, small-molecule screens and mutant fish screens. Generation of real-time monitoring systems of electrophile-induced Nrf2 activation has been studied for many years. Stable cell lines that can monitor Nrf2 activation were previously produced using multiple ARE/EpRE sequences as transcriptional drivers and GFP or luciferase as reporter genes (Zhu & Fahl 2001; Wang et al. 2006). The results obtained by use of these cell lines are reproducible and are more reliable than experiments relying on transient transfection. Transgenic mice, which have luciferase reporter genes fused to multiple ARE/EpRE sequences, were also generated (Yates et al. 2007; Fisher et al. 2007). These transgenic mice are useful because they can allow the monitoring of tissue- and developmental stage-specific profiles of Nrf2 activation. Real-time activation of Nrf2 in disease models can also be analyzed by crossing these mice with a variety of gene-disrupted mice. However, this model has shortcomings, in that administration of the substrate luciferin into mice is required, whose disposition and side effects might affect the proper analysis of results. The use of GFP instead of luciferase will be helpful to overcome this problem; however, it is difficult to analyze GFP expression in the internal regions of a mouse. Trials to generate stable transgenic zebrafish that harbor ARE/EpRE-driven GFP reporter gene have been carried out (Carvan et al. 2000; Kusik et al. 2008) but were unsuccessful. Tg(-3.5gstp1:GFP)it416b is the first ARE/EpRE-GFP reporter line that can be used to monitor Nrf2 activation at a practical and physiologically relevant level. Our success was attributed to a screen of effective lines from substantial numbers of stable transgenic lines. In that sense, the use of the Tol2 transposon system was an important strategy which greatly improved the transgene integration efficiency (Kawakami 2007).
At the onset of this study, we hoped to establish transgenic lines that can analyze tissue-specific differences in GFP induction upon induction by a variety of Nrf2 activators. Unfortunately, this method did not succeed. GFP induction in gills and livers, in which endogenous gstp1 was induced (in addition to the olfactory regions), was not observed in the present lines. We supposed that cis- and trans-activating factors required for gstp1 induction differ between tissues. Indeed, cell-specific differences in the induction levels of the ARE/EpRE reporter gene have been demonstrated using cultured cell lines (Wang et al. 2006). Gene regulatory elements required for the expression in gills and liver other than ARE/EpRE sequences may not be included in the 3.5-kb gstp1 region. The negative and positive roles of Nrf2 in human health have both been recently demonstrated (Hayes & McMahon 2009; Lau et al. 2008). It will be important to elucidate the molecular mechanisms of the cell-specific induction of Nrf2 target genes, particularly in cancer cells. Modified bacterial artificial chromosomes for transgenesis will be useful to identify cis-activating factors for cell specificity (Yang et al. 2009).
The zebrafish is an attractive model for analyzing gene expression profiles because it has transparent embryos and larvae. Moreover, a transparent adult line, termed casper, has been recently generated (White et al. 2008). Thus, utilizing a GFP reporter will facilitate the visualization of a variety of stress responses in living animals. Indeed, cellular responses to metals, such as cadmium, copper and arsenate, were monitored in GFP transgenic zebrafish using heat shock promoters (Blechinger et al. 2007; Wu et al. 2008; Seok et al. 2007). GFP induction by environmental estrogens and dioxin was also analyzed by GFP transgenic zebrafish using estrogen response elements + vitellogenin promoter (Chen et al. 2010) and cytochrome P4501A1 promoter (Mattingly et al. 2001), respectively, as transcriptional drivers. The zebrafish provides opportunities to accelerate the process of drug discovery (Zon & Peterson 2005). These GFP transgenic lines, including Tg(-3.5gspt1:GFP)it416b, will contribute to several aspects of the drug development process, including target identification by genetic and morpholino oligonucleotide screens, lead discovery by small-molecule screens and the testing of drug efficacy and toxicity. The application of Tg(-3.5gspt1:GFP)it416b to screen novel Nrf2 activators and inhibitors will be therefore be pursued in future studies.
Embryos and larvae were obtained by natural mating. The wild-type strain AB and the mutant strain it567 (Kobayashi et al. 2009) were used. The stable transgenic lines Tg(-3.5gstp1:GFP)it416a and Tg(-3.5gstp1:GFP)it416b were maintained as homozygous lines. Homozygous transgenic lines were crossed with the AB strain to obtain embryos and larvae for experiments. The genotyping of transgenic lines was carried out by PCR using genomic DNA as templates and the primers 5′- CCTCGTGACCACCCTGAC-3′ and 5′- TGGCGGATCTTGAAGTTCAC-3′.
For the induction studies, zebrafish were placed in culture dishes containing 100 μm DEM (Wako, Osaka, Japan), 40 μm sulforaphane (LKT laboratories, St. Paul, MN), 2.5 μm 15d-PGJ2 (Cayman Chemical, Ann Arbor, MI), 1 μm auranofin (Enzo Life Sciences, Farmingdale, NY), 1 mm H2O2 (Wako), 20 μm CdCl2 (Wako) and 1.5 or 2.5 μm OA-NO2. OA-NO2 was prepared as previously described (Baker et al. 2005).
For the overexpression of Nrf2 and Keap1 proteins, synthetic capped mRNA was generated with an SP6 mMESSAGE mMACHINE in vitro transcription kit (Ambion, Austin, TX) using linearized DNA of the pCS2 derivatives described above and injected into yolk at the one-cell stage using an IM300 microinjector (Narishige, Tokyo, Japan). For knocking-down zebrafish Nrf2 (nrf2), antisense morpholino oligonucleotides (Gene Tools, Philomath, OR) specific for nrf2 (nrf2MO) (Kobayashi et al. 2002) were injected into yolk at the one-cell stage. For generating stable transgenic lines, 25 pg of pT3.5gstp1GFP or p3.5gstpGFP was co-injected into a blastomere of early one-cell stage embryos with 25 pg of Tol2-transposase mRNA synthesized from NotI-linearized pCS-TP as previously described (Takeuchi et al. 2010).
GFP expression during various embryonic stages was examined under a GFP2 (480 nm excitation, 510 nm barrier) filter of a MZFLIII microscope (Leica, Wetzler, Germany) equipped with a 600CL-CU digital camera (Pixera Los Galos, CA). GFP expression during larval stages was analyzed under a GFP-BP BZ filter (470 nm excitation, 535 nm emission) of a Biozero BZ-8000 microscope (Keyence, Osaka, Japan). For taking time-lapse photographs, larvae were anesthetized with 0.016% 3-aminobenzoic acid ethyl ester (Sigma-Aldrich, St. Louis, MO) and placed on a hole slide glass containing 3% methylcellulose (Sigma Aldrich) for immobilization. To quantify GFP expression, fluorescence intensity in the olfactory regions was measured by an image processing system equipped with the BZ-8000 microscope. The gene expressions of endogenous gstp1 were examined by an in situ hybridization analysis, a RT-PCR analysis or a real-time RT-PCR analysis, as previously described (Li et al. 2008; Kobayashi et al. 2001).
MALDI-TOF MS analysis
The recombinant mouse Keap1 protein expressed by Escherichia coli was incubated with or without 100 μm OA-NO2 for 60 min at 25 °C in buffer containing 20 mm Tris–HCl, pH 8.5. The trypsin-digested mouse Keap1 was mixed with dithiothreitol and trifluoroacetic acid. To improve the ionization efficiency of MS, samples were purified with Zip-tip μC18 (Millipore, Bedford, MA) before MS analysis. Peptides were mixed with α-cyano-4-hydroxycinnamic acid (2.5 mg/ml) containing 50% acetonitrile and 0.1% trifluoroacetic acid and dried on stainless steel targets at room temperature. The analyses were carried out using an AXIMA-TOF2 (Shimadzu, Tokyo, Japan) with a nitrogen laser. All analyses took place in the positive ion mode, and the instrument was calibrated immediately before each series of studies.
We thank Sakura Motion Picture Co. Ltd for their assistance with the movies. We also thank T. Kinoshita, H. Niu, T. Shimokoube and Y. Terashita for help in fish maintenance, and T. Suzuki, M. Takeuchi and K. Mukaigasa for help and discussion. We also thank T. Kudoh for critical reading of the manuscript. This work was supported by Grants-in-Aid from the Japan Science and Technology Corporation (ERATO) (to M.Y.), and the Ministry of Education, Culture, Sports, Science and Technology of Japan (to M.K. and M.Y.).