Critical Contribution of Nuclear Factor Erythroid 2-related Factor 2 (NRF2) to Electrophile-induced Interleukin-11 Production*

Nuclear factor erythroid 2-related factor 2 (NRF2) is a transcription factor that plays a crucial role in protection of cells from electrophile-induced toxicity through up-regulating phase II detoxifying enzymes and phase III transporters. We previously reported that oxidative stress induces up-regulation of interleukin-11 (IL-11), a member of the IL-6 family that ameliorates acetaminophen-induced liver toxicity. However, a role for IL-11 in protection of cells from electrophile-induced toxicity remains unclear. Here we show that an environmental electrophile, 1,2-naphthoquinone (1,2-NQ), but not 15d-prostaglandin J2 (PGJ2) or tert-butylhydroxyquinone (tBHQ), induced IL-11 production. Consistent with a crucial role for prolonged ERK activation in H2O2-induced IL-11 production, 1,2-NQ, but not 15d-PGJ2 or tBHQ, elicited prolonged ERK activation. Conversely, inhibition of the ERK pathway by a MEK inhibitor completely blocked 1,2-NQ-induced IL-11 production at both protein and mRNA levels, further substantiating an intimate cross-talk between ERK activation and 1,2-NQ-induced IL-11 production. Promoter analysis of the Il11 gene revealed that two AP-1 sites were essential for 1,2-NQ-induced promoter activities. Among various members of the AP-1 family, Fra-1 was up-regulated by 1,2-NQ, and its up-regulation was blocked by a MEK inhibitor. Although NRF2 was not required for H2O2-induced IL11 up-regulation, NRF2 was essential for 1,2-NQ-induced IL11 up-regulation by increasing Fra-1 proteins possibly through promoting mRNA translation of FOSL1. Finally, intraperitoneal administration of 1,2-NQ induced body weight loss in wild-type mice, which was further exacerbated in Il11ra1−/− mice compared with Il11ra1+/− mice. Together, both Fra-1 and NRF2 play crucial roles in IL-11 production that protects cells from 1,2-NQ intestinal toxicity.

ubiquitination of NRF2 is attenuated, thereby rescuing newly synthesized NRF2 from proteasomal degradation and allowing translocation of NRF2 into the nucleus. NRF2 binds to a cognate DNA regulatory element termed the antioxidant response element and up-regulates its target genes. Of interest, 1,2-NQ also activates NRF2 through covalent modification of Keap1 (10). Collectively, it seems likely that there are different adaptive responses against environmental electrophiles, such as 1,2-NQ, via activating signal transduction pathways.
Interleukin-11 (IL-11) is a member of the IL-6 family cytokines and controls various cellular responses, including hematopoiesis, bone development, tissue repair, and carcinogenesis (11). IL-11 binds to the IL-11 receptor ␣1 (IL-11R␣1) and gp130 complex and activates the family of signal transducer and activator of transcription (STAT) proteins. We previously reported that IL-11 is produced by hepatocytes in an oxidative stress-dependent manner and ameliorates acetaminophen-induced liver injury (12). In addition, IL-11 treatment prevents cardiac dysfunction and attenuates oxidative stress after ischemia and reperfusion injury in heart (13). Together, these results suggest that IL-11 acts as a protective cytokine against oxidative stressinduced tissue injury. However, it is unclear whether IL-11 contributes to protection of cells against electrophile-induced toxicity. We previously reported that 1,2-NQ activates the MEK/ ERK pathway through modification and subsequent activation of epidermal growth factor receptor (14). Given that expression of Il11 is regulated by the MEK/ERK pathway (12), we surmised that 1,2-NQ might induce IL-11 production.
We found that 1,2-NQ, but not tBHQ or 15d-PGJ 2 , induced IL-11 expression at both mRNA and protein levels. We showed that 1,2-NQ-induced up-regulation of Il11 mRNA largely depended on phosphorylation of ERK and subsequent induction of Fra-1. Moreover, knockdown of NRF2 by siRNA completely abolished 1,2-NQ-induced IL11 induction. Conversely, overexpression of NRF2 induced Il11 reporter gene activation. Unexpectedly, we found that knockdown of NRF2 abolished expression of Fra-1 proteins before or after 1,2-NQ stimulation, whereas overexpression of NRF2 alone induced an increase in Fra-1 proteins. Finally, 1,2-NQ-induced body weight loss and intestinal toxicity were exacerbated in Il11ra1 Ϫ/Ϫ mice compared with control Il11ra1 ϩ/Ϫ mice. Together, these data suggest that NRF2 plays a crucial role in IL11 up-regulation through increasing Fra-1 proteins.

Results
An Electrophile Induces IL-11 Expression-To test whether electrophiles increase Il11 expression, we treated a murine colon cancer cell line, CT26, and a human hepatoma cell line, HepG2, with several electrophiles, including 15d-PGJ 2 , tBHQ, and 1,2-NQ. To be precise, tBHQ is not an electrophile but is easily converted to an electrophilic quinone through autoxidation. As expected, all three electrophiles induced accumulation of NRF2 ( Fig. 1A and supplemental Fig. 1) and significantly increased expression of an NRF2 target gene, Hmox1, in CT26 and HepG2 cells (Fig. 1, B and C). Intriguingly, 1,2-NQ, but not 15d-PGJ 2 or tBHQ, increased expression of Il11 mRNA in CT26 and HepG2 cells (Fig. 1, B and C). Moreover, we found that 1,2-NQ induced the production of IL-11 protein in CT26 and HepG2 cells (Fig. 1D).
Activation of the ERK Pathway Is Required for 1,2-NQ-induced Il11 Up-regulation-To investigate the molecular mechanisms underlying 1,2-NQ-induced Il11 mRNA expression, we sought to identify the signaling pathways leading to Il11 expression. We previously reported that H 2 O 2 -induced IL-11 production largely depends on the ERK pathway (12). We also reported that 1,2-NQ induces ERK activation by inactivating PTP1B through modification of a cysteine residue of PTP1B (7). As shown in Fig. 2A, 1,2-NQ induced phosphorylation of JNK, ERK, AKT, and to a lesser extent p38 in CT26 cells, whereas 1,2-NQ only induced phosphorylation of ERK in HepG2 cells. We next tested the effects of various MAP kinase inhibitors on 1,2-NQ-induced Il11 induction in these cells. Inhibition of the MEK/ERK pathway with a MEK inhibitor, U0126, almost completely abolished 1,2-NQ-induced Il11 up-regulation in both cells (Fig. 2B). Surprisingly, inhibition of the p38MAPK pathway by SB230580 strongly up-regulated expression of Il11 mRNA in HepG2, but not CT26 cells, although the mechanism was not investigated in this study. We also found that strong and sustained ERK activation was only observed in 1,2-NQstimulated, and not 15d-PGJ 2 -or tBHQ-stimulated, CT26 and HepG2 cells (Fig. 2C).
Two AP-1 Sites Are Essential for 1,2-NQ-induced Il11 Gene Promoter Activity-To further clarify the mechanisms underlying 1,2-NQ-dependent Il11 expression, we tested whether 1,2-NQ could activate a luciferase reporter under the control of the Ϫ1.3 kb fragment of murine Il11 promoter (Fig. 3A). Il11 gene promoter activity was increased by 1,2-NQ treatment, and this increase was almost completely abolished in the presence of U0126 (Fig. 3B). Given that the Il11 promoter contains two AP-1 binding sites and the MAPK pathway induces activation of AP-1 (15), we introduced mutation of each element alone or a combination of them (Fig. 3A). As expected, mutation of each AP-1 element alone partially reduced, and both mutations of two AP-1 sites completely abolished, a 1,2-NQ-induced increase in promoter activities (Fig. 3C). Moreover, 1,2-NQ could activate a luciferase reporter containing the Ϫ0.8 kb fragment of the murine Il11 promoter containing two wild-type AP-1 sites but not two AP-1 mutants (Fig. 3D). These data suggest that two AP-1 sites are required for 1,2-NQ-induced Il11 induction.
Fra-1 Is Required for Induction of Il11-The AP-1 is a dimeric transcriptional factor complex that is composed of Fos and Jun family proteins (15). Moreover, the MEK/ERK pathway regulates Fos and Jun family protein levels. We examined whether 1,2-NQ treatment might affect protein levels of members of the Fos and Jun family in CT-26 and HepG2 cells in the absence or presence of U0126. Whereas c-Fos, FosB, c-Jun, and JunD protein levels remained constant after 1,2-NQ treatment, expression of Fra-1 was increased by treatment with 1,2-NQ, and this increase was suppressed by U0126 pretreatment (Fig. 4A and supplemental Fig. 2). In addition, quantitative PCR showed that 1,2-NQ treatment increased expression of Fosl1 encoding Fra-1 protein along with Il11 expression, and this increase was suppressed by U0126 treatment in CT26 and HepG2 cells (Fig. 4B).
These results suggest that 1,2-NQ up-regulated Fra-1 at both mRNA and protein levels in a MEK/ERK-dependent manner.
We next performed an electrophoretic mobility shift assay (EMSA) using nuclear extracts from 1,2-NQ-treated HepG2 cells and an oligonucleotide probe containing two AP-1 sites of the Il11 promoter. 1,2-NQ induced a retarded complex containing labeled AP-1 oligonucleotides that disappeared in the presence of U0126 (Fig. 4C). Moreover, we found that the complex was supershifted with the addition of antibodies against Fra-1, JunB, or JunD, suggesting that the 1,2-NQ-induced nuclear complex is at least composed of Fra-1, JunB, and JunD. To further investigate whether these transcriptional factors are involved in Il11 expression, we knocked down expressions of each gene by respective siRNA. Whereas knockdown of Junb or Jund by the respective siRNA did not suppress, but rather enhanced, 1,2-NQ-induced Il11 expression (supplemental Fig.  3), knockdown of FOSL1 by siRNA significantly suppressed 1,2-NQ-induced FOSL1 and IL11 mRNA expression (Fig. 4D). Furthermore, a chromatin immunoprecipitation (ChIP) assay revealed that Fra-1 was recruited to the Il11 promoter following 1,2-NQ as well as H 2 O 2 stimulation (Fig. 4E).
NRF2 Is Required for 1,2-NQ-induced IL11 Expression-A previous study has shown that the murine glutathione S-transferase Ya subunit gene promoter contains the AP-1-like sites that partially overlapped with the antioxidant response element, and Gst Ya expression is induced by PMA and electrophiles (16). Given that one of the two AP-1 sites in the Il11 promoter overlapped with the consensus sequence of NRF2 binding sites (17) (Fig. 5A), we tested whether NRF2 could activate a luciferase reporter under the control of the Il11 promoter. Overexpression of NRF2 activated reporter vectors containing Ϫ1.3 or Ϫ0.8 kb fragments of the Il11 promoter in a dose-dependent manner (Fig. 5B). Moreover, NRF2-dependent promoter activity was abolished in a reporter vector, in which mutations of two AP-1 sites were introduced (Fig. 5B). We finally knocked down expression of NRF2 by siRNA in HepG2  5D). These results suggest that NRF2 is essential for 1,2-NQ but not H 2 O 2 -induced Il11 expression.
NRF2 Is Not Recruited to the IL11 Promoter and Does Not Interact with Fra-1-The fact that NRF2 activated transcription of IL11 mRNA prompted us to tested whether NRF2 was recruited to the IL11 promoter after 1,2-NQ treatment by a ChIP assay. Anti-NRF2 antibody efficiently precipitated the promoter, but not an unrelated region of NAD(P)H:quinone oxidoreductase 1 (NQO1), a canonical target gene of NRF2, after 1,2-NQ stimulation (Fig. 6A). However, under these experimental conditions, the promoter of IL11 was not precipitated with anti-NRF2 antibody after 1,2-NQ stimulation (Fig. 6A).
We next tested the possibility that NRF2 might be recruited to the IL11 promoter via interaction with Fra-1. To test this possibility, we transiently transfected an expression vector for Myc-tagged NRF2 along with FLAG-tagged Fra-1 into HEK293T cells. Moreover, to test a possibility that phosphorylation of Fra-1 might affect its binding to NRF2, we co-transfected expression vectors for Myc-tagged NRF2, FLAG-tagged Fra-1, HA-tagged constitutively active MEKK1 (MEKK1⌬N), and HA-tagged ERK2. Consistent with our previous study (12), HA-tagged ERK2 was efficiently immunoprecipitated with FLAG-tagged Fra-1 (Fig. 6B). In sharp contrast, anti-FLAG antibody could not immunoprecipitate Myc-tagged NRF2, suggesting that NRF2 could not interact with either phosphorylated or unphosphorylated Fra-1 at least under our experimental conditions. These results suggest that NRF2 does not appear to be directly or indirectly recruited to the IL11 promoter.
Knockdown of NRF2 Abolishes Expression of Fra-1 Proteins but Not FOSL1 mRNAs-The fact that NRF2 was not recruited to the IL11 promoter following 1,2-NQ stimulation or did not interact with Fra-1 ( Fig. 6) prompted us to test whether NRF2 might regulate expression of Fra-1 at mRNA or protein levels. Intriguingly, a very recent study has shown that NRF2 promotes mRNA translation of a particular set of genes (18). Although knockdown of NRF2 by siRNA did not affect FOSL1 mRNA expression (Fig. 7A), knockdown of NRF2 abolished Fra-1 expression at protein levels ( Fig. 7B). In sharp contrast, expressions of c-Fos, STAT3, pERK, or total ERK were not decreased in NRF2-knockdown cells. This suggests that NRF2 might increase Fra-1 proteins through preventing degradation of Fra-1 by the ubiquitin-proteasome pathway or enhance mRNA translation of FOSL1.
To discriminate between these two possibilities, we next tested the effect of knockdown of NRF2 on expression of Fra-1 proteins in the absence or presence of a proteasome inhibitor, MG132. Because unphosphorylated Fra-1 is constitutively degraded by the ubiquitin-proteasome pathway (12), hypophosphorylated Fra-1 accumulated in the presence of MG132 (Fig. 7B). Notably, even in the presence of MG132, Fra-1 was decreased in NRF2-knockdown cells, suggesting that NRF2 did not block degradation of Fra-1 by the ubiquitin-proteasome pathway. Conversely, overexpression of NRF2 increased Fra-1, but not c-Fos, STAT3, pERK, or ERK proteins (Fig. 7C). Together, NRF2 increases expression of Fra-1, possibly through up-regulating its translation, thereby inducing IL11 expression.

Administration of 1,2-NQ Induces Il11 Expression and Enhances Proliferation of Intestinal Epithelial Cells (IECs) in the
Cecum-To investigate the biological consequences of 1,2-NQdependent IL-11 production in vivo, wild-type mice were injected intraperitoneally with 1,2-NQ. Compared with oiltreated mice, 1,2-NQ-treated mice exhibited body weight loss 72 h after injection (Fig. 8A). To our surprise, histology of small intestine and colon appeared to be normal in 1,2-NQ-treated mice (data not shown). In sharp contrast, 1,2-NQ administration induced dilatation of the cecum and increased numbers of apoptotic IECs (Fig. 8, B-D). Moreover, numbers of Ki67-and cyclin D1-positive proliferating IECs were significantly increased in the cecum of mice compared with untreated mice (Fig. 8, E and F). Under these experimental conditions, we examined whether 1,2-NQ induced expression of Il11 and Hmox1 mRNAs in the cecum. Consistent with in vitro data, administration of 1,2-NQ into wild-type mice resulted in upregulation of Il11 and Hmox1 mRNAs (Fig. 8G). These results suggest that 1,2-NQ administrations may enhance proliferation of IECs through up-regulation of IL-11.

Discussion
In the present study, we showed that 1,2-NQ induced IL-11 expression in an MEK/ERK-dependent manner. Both Fra-1 and NRF2 were essential for 1,2-NQ-induced IL-11 production, although the mechanisms underlying Fra-1-and NRF2-dependent up-regulation of IL11 mRNA appeared to be different. Upon stimulation of cells with 1,2-NQ, Fra-1 was recruited to the Il11 promoter and acted as a transcription factor. In sharp contrast, NRF2 was not recruited to the IL11 promoter but increased translation of the FOSL1 transcript, which subsequently increased expression of Fra-1 proteins. Moreover, given that electrophile-induced intestinal toxicity was exacerbated in Il11ra1 Ϫ/Ϫ mice compared with Il11ra1 ϩ/Ϫ mice, IL-11 protects cells against electrophile-induced as well as oxidative stress-induced tissue injury.
The mechanisms underlying prolonged ERK activation by 1,2-NQ, but not other agents tested here, are not fully investigated in the present study. In this respect, quinones exhibit two chemical properties, including those of an electrophile and electron transfer agent. Thus, quinones transfer electrons from a reducing agent, such as NADPH, to oxygen, resulting in generation of superoxides that are finally converted to hydrogen peroxide. Consistently, a previous study has reported that 1,2-NQ induces hydrogen peroxide-dependent oxidation of protein thiols corresponding to sulfenic acid in the cells (19). Given that hydrogen peroxide is responsible for prolonged ERK activation (12), 1,2-NQ-dependent production of hydrogen peroxide, in concert with activation of EGFR, might be responsible for 1,2-NQ-induced ERK activation. Further study is required to address this issue.
One of the most important findings of this study is that NRF2 contributes to 1,2-NQ-induced IL-11 production. An essential role for NRF2 in 1,2-NQ-induced IL-11 production is supported by the following findings. First, 1,2-NQ induced accumulation of NRF2. Second, overexpression of NRF2 activated the IL11 gene promoter activities via AP-1 sites, which contain a putative NRF2 consensus motif. Third, knockdown of NRF2 by siRNAs blocked 1,2-NQ-induced IL11 up-regulation. However, we could not detect either the recruitment of NRF2 to the IL11 promoter by a ChIP assay or interaction of NRF2 with Fra-1. To our surprise, we found that knockdown of NRF2 abolished expression of Fra-1 proteins but not FOSL1 mRNA levels.
Moreover, a proteasome inhibitor did not suppress a decrease in Fra-1 in NRF2-knockdown cells, suggesting that NRF2 did not appear to up-regulate Fra-1 through preventing the degradation of Fra-1 by the ubiquitin-proteasome. A recent study has shown that NRF2 promotes mRNA translation of several genes through preventing oxidative stress-dependent blockade of assembly of the translation initiation complex in pancreatic cancer (18). Thus, it is reasonable to hypothesize that NRF2 promotes mRNA translation of FOSL1, thereby increasing protein expression of Fra-1. Further study will be required to elucidate the detailed mechanism whereby NRF2 regulates Fra-1 expression.
Constitutive activation of the NRF2 pathway and elevated expression of IL-11 are tightly associated with the development of cancer (11,20), IL-11 might contribute, at least in part, to NRF2-dependent oncogenesis in a context-dependent manner. Intriguingly, knockdown of NRF2 could not suppress H 2 O 2induced IL11 up-regulation. Although Fra-1 is essential for both 1,2-NQ-and H 2 O 2 -induced IL-11 production (12) (in this study), different transcription factors induced by H 2 O 2 or NRF2 in concert with Fra-1 might mediate IL-11 production in a stimulus-dependent manner.
Intraperitoneal injection of 1,2-NQ resulted in dilatation of the cecum along with severe weight loss of wild-type mice. We found that the number of apoptotic IECs was increased in the cecum of 1,2-NQ-treated wild-type mice compared with untreated mice. Under normal conditions, IECs move toward the tip of the villi and undergo apoptosis, followed by shedding into the lumen of the intestine. Numbers of Ki67-and cyclin D1-positive proliferating IECs were significantly increased in 1,2-NQ-treated wild-type mice compared with untreated mice, suggesting that an increase in apoptotic IECs might not be a primary event but rather a secondary event induced by an increased turnover of IECs after 1,2-NQ treatment. Consistent with this idea, numbers of both apoptotic and proliferating IECs were reduced in Il11ra1 Ϫ/Ϫ mice compared with Il11ra1 ϩ/Ϫ mice. Our preliminary experiments showed that injection of recombinant IL-11 up-regulated expression of Reg3b and Reg3g in the intestines (data not shown). Given that Reg3␤ and Reg3␥ act as antimicrobial proteins and promote tissue repair (21,22), IL-11 might attenuate 1,2-NQ-induced intestinal toxicity through up-regulation of Reg3b and Reg3g. It is currently unknown why intestinal toxicity was relatively restricted to the cecum and not the small intestine or colon. Further study is required to address this issue.  Mice-Il11ra1 Ϫ/Ϫ mice were provided by L. Robb and described previously (23), and they were back-crossed to C57BL/6 mice for at least 7 generations. C57BL/6 mice were purchased from Japan-SLC. All experiments were performed according to the guidelines approved by the Institutional Animal Experiment Committee of Juntendo University School of Medicine and Toho University School of Medicine.
EMSA-EMSA was performed as described previously (24). Briefly, nuclear extracts were prepared from unstimulated or 1,2-NQ-stimulated HepG2 cells in the absence or presence of U0126. The two AP-1-containing oligonucleotides (5Ј-AGGG-AGGGTGAGTCAGGATGTGTCAGGCC-3Ј and 5Ј-AGGG-CGGCCTGACACATCCTGACTCACCCT-3Ј) were labeled with T4 polynucleotide kinase in the presence of [␥-32 P]ATP (PerkinElmer Life Sciences). Then nuclear extracts were incubated with the labeled oligonucleotides. The composition of the AP-1 complex was examined by supershift analysis with the indicated antibodies.
HepG2 cells were transfected with the indicated reporter vector along with pRL-TK vector (Promega) using Lipofectamine 2000. At 24 h after transfection, fresh growth medium was added to the transfection reaction, cells were stimulated with 1,2-NQ for 18 h, and luciferase activities were measured using the Pica Gene Dual-Luciferase kit (Toyo Ink) on a Luminometer (Berthold).
ChIP Assay-The ChIP assay was performed as described previously with slight modification (25). Briefly, HepG2 cells were unstimulated or stimulated with the indicated concentrations of 1,2-NQ for 2 h. Cells were fixed with 1% formalin for 30 min and then harvested and lysed with a ChIP assay buffer. After brief sonication by using a Bioruptor II (BM Equipment), the lysates were immunoprecipitated with control Ig or anti-NRF2 antibody. After extensive washing, immunoprecipitated DNA fragments were released and subjected to qPCR. The primers to amplify the promoter regions of IL11 (Ϫ177/Ϫ9) and NQO1 (Ϫ429/Ϫ302) and an unrelated region of the NQO1 (ϩ8068/ϩ8163) gene were as follows: IL11 (Ϫ162/Ϫ80), 5Ј-GAGCGCGGCGGCGTGAGCCCT-3Ј and 5Ј-GACACA-TCCTGACTCACCCT-3Ј; NQO1 (Ϫ429/Ϫ302), 5Ј-CATGT-CTCCCCAGGACTCTC-3Ј and 5Ј-TTTTAGCCTTGGC-ACGAAAT-3Ј; NQO1 (ϩ8068/ϩ8163), 5Ј-CGTGTGTGCTT-TGTGTGTGT-3Ј and 5Ј-GCCTCCTTCATGGCATAGTT-3Ј. The amounts of a target DNA in immunoprecipitates with control rabbit Ig (Thermo Fisher Scientific) or anti-NRF2 antibody were quantified by qPCR using 7500 SDS software (Applied Biosystems). In brief, the ratios of the amounts of a target DNA fragment in each immunoprecipitate to those in the DNAs before immunoprecipitation (input DNA) were calculated from each cycle threshold value.
For transient expression of NRF2, HepG2 cells were transfected with Myc-tagged NRF2 as described above. After an 18-h transfection, cells were lysed in a radioimmune precipitation assay buffer, and cell lysates were subjected to SDS-PAGE analysis.
Histological and Immunohistochemical Analyses-Small intestine, cecum, and colon were fixed in 10% formalin and embedded in paraffin blocks. Paraffin-embedded intestinal sections were used for H&E staining. Paraffin-embedded sections were stained with anti-Ki67, anti-cyclin D1, and anti-cleaved caspase 3 (CC3) antibodies and visualized with biotin-conjugated donkey anti-rabbit IgG antibody and streptavidin-conjugated HRP. Pictures were obtained with an all-in-one microscope (BZ-X700, Keyence) and analyzed with Axio version 3.0 (Zeiss).
Statistical Analyses-Statistical analysis was performed by unpaired Student's t test. In some experiments that involved more than two conditions, Tukey's analysis of variance test was performed. p values of Ͻ0.05 were considered to be significant.