|Home | About | Journals | Submit | Contact Us | Français|
The Keap1-Nrf2-ARE signaling pathway evokes an adaptive response for cell survival following endogenous (for example, inflammation) and exogenous (for example, carcinogens) stresses. Keap1 inhibits the transcriptional activation activity of Nrf2 in unstressed cells by facilitating its degradation. Through transcriptional analyses in Keap1- or Nrf2-disrupted mice, we identified interactions with the Notch1 signaling pathway. We found a functional antioxidant response element (ARE) recognized by Nrf2 in the promoter of Notch1. Notch1 regulates processes such as proliferation and cell fate decisions. We report a functional role for this cross talk between the two pathways and show that disruption of Nrf2 impeded liver regeneration following partial hepatectomy and was rescued by re-establishment of Notch1 signaling.
Nrf2 (p45 nuclear factor erythroid-derived 2-related factor 2) is a prosurvival transcription factor that plays a pivotal role in maintaining cellular homeostasis following electrophile and oxidative stresses (1). Coordinated changes in response to Nrf2 activation result in enhanced expression of genes that contribute to (i) free radical metabolism, (ii) electrophile detoxication, (iii) glutathione biosynthesis, (iv) generation of reducing equivalents, (v) formation of direct antioxidants, (vi) toxin efflux transporters, (vii) the proteasomal system, as well as (viii) modulation of inflammatory pathways. Many of these genes contain cis-elements in their regulatory domains known as antioxidant response elements (AREs) (2). The ARE interacts with the cap ‘n’ collar (Cnc) family of basic leucine zipper (bZip) transcription factors, including Nrf2, and their dimerization partners that include other bZip transcription factors, such as the small Mafs (1). Experiments with knockout mice for various Cnc genes revealed that these transcription factors contribute to both constitutive and inducible expression of these cytoprotective genes (3).
Analysis of Nrf2-null mice revealed that Nrf2 is the primary factor in regulation of gene expression through the ARE in vivo. The knockout mice survive and produce viable offspring, but they are more sensitive towards toxicities associated with various electrophiles and reactive oxygen species than are wild-type mice (3, 4). In early development, Nrf2-null mice appear normal. However, it has not been determined whether development in the knockout mice is altered under stressed conditions. By contrast, mice null for Nrf1 (encoding a related Cnc bZip transcription factor) show a drastic phenotype in early development and are embryonic-lethal at mid-to-late gestation due to anemia caused by a non-cell autonomous defect in erythropoiesis in the fetal liver (5). This observation suggests a fundamental role for Nrf1 in development. Analysis of the liver-specific Nrf1-conditional knockout mouse indicates that Nrf1 contributes to postnatal ARE-dependent gene regulation as well as having a prenatal role (6). Because the Nrf1::Nrf2 double knockout mouse has a shorter life span in utero than the Nrf1 single knockout, Nrf2 has been proposed to contribute or partially compensate as a transcription factor in early development (7). Nrf2 influences tissue injury and repair. The increased abundance of Nrf2 in hepatocyte-specific Keap1-disrupted mice results in resistance to chemical- and inflammation- mediated hepatotoxicity (8, 9). By contrast, Nrf2-null mice, which are more sensitive to hepatotoxins (10), exhibit significantly delayed liver regeneration following partial hepatectomy (11).
Differential microarray analyses have been conducted in several tissues of wild-type and Nrf2-null mice (12, 13) to help define cytoprotective pathways and to characterize the role of the Keap1-Nrf2-ARE pathway in the pharmacodynamic action of several classes of anticarcinogens (14, 15). However, these analyses reflect expression patterns from heterogeneous cell populations within the analyzed tissues. To overcome this limitation, immortalized mouse embryonic fibroblasts (MEFs) were established from Keap1−/−, Nrf2−/−, Keap1−/−::Nrf2−/−, and wild-type mice (16). These MEFs showed striking differential sensitivity to redox cycling quinones (17) and photooxidative damage from combined exposure to retinaldehyde and UVA light (18). Strong differential expression of Notch1 and some of its downstream effectors was observed, raising the interesting possibility that the Notch1 signaling cascade may be regulated at the transcriptional level through interactions of Nrf2 with ARE. Levels of Notch1 mRNA were substantially reduced in Nrf2-null MEF compared to wild-type; transcripts for Notch1 target genes, such as Hes-1, Herp1, Herp2 (19), and Nrarp (20) were also reduced significantly.
The Notch family of transmembrane receptors participates in a signaling pathway controlling a broad spectrum of metazoan cell fates and developmental processes through local cell-cell interactions (21). Alteration of signaling through the Notch family of receptors can markedly affect differentiation, proliferation, and apoptotic events. Genetic ablation studies indicate that Notch1 is crucial for early development and re-growth of several tissues (22, 23). Activation of the Notch pathway inhibits differentiation in different developmental contexts and has been associated with the amplification of some somatic stem cells— not only the neural (24) and hematopoietic stem cells (25), but also hepatocyte (26, 27) and intestinal epithelial stem cells (28, 29). Considering the importance of the Notch1 signal cascade in developmental biology, the microarray observations indicated the possibility that Nrf2 could be a key molecule affecting both embryonic and adult tissue stem cell renewal as well as cell fates. This study characterizes the effects of Nrf2 genotype on the expression of Notch1 and its effector genes and the importance of Nrf2-Notch1 crosstalk in liver regeneration.
Microarray analyses compared gene expression patterns in wild-type or Nrf2-disrupted mice in order to identify Nrf2-regulated genes (13) and provided hints for interactions between Nrf2 and other signaling pathways such as the aryl hydrocarbon receptor (AhR) (30) and nuclear factor-κB (NF-κB) pathways (31). Further analysis of one of these data sets (30) revealed that Notch1 transcript levels were 14-fold lower in Nrf2-null compared to wild-type MEFs. Furthermore, the expression of Hes-1 and p21, two downstream genes in the Notch1 pathway (19, 32), were also reduced substantially in Nrf2-null cells. To confirm the microarray data indicating that Nrf2 promotes the Notch1 signaling pathway, we analyzed the expression of additional downstream effector genes by semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) in wild-type and Nrf2-null MEFs. We found that the abundance of transcripts for the Notch1 target genes Hes-1, Herp1, Herp2, Nrarp, and p21 were also reduced in Nrf2-null MEFs, corresponding to the reduced Notch1 expression. By contrast, no effect on the abundance of transcript for the Notch1 ligand, Jag1, was observed (Fig. 1A). Quantitative real-time RT-PCR analysis confirmed the reduction in Notch1 expression in Nrf2-null mice (Fig. 1B). We also found that expression of a prototypic Nrf2-regulated gene, Gsta1, was substantially lower in the Nrf2-deficient fibroblasts. Expression of Gsta1 was induced by treatment with 2.5 μM sulforaphane (an activator of Nrf2 signaling) (14) in wild-type cells only (Fig. 1B). As with Gsta1, induction of Notch1 by sulforaphane was also lost in the absence of functional Nrf2 (Fig. 1B). These results suggest a role for Nrf2 in Notch1 expression.
To elucidate whether the effect of Nrf2 genotype on Notch1 expression observed in MEFs also occurred in vivo, transcript abundance was assessed in adult livers. Notch1 transcripts were consistently reduced ~ 40% in livers of adult Nrf2-null mice compared to those in wild-type mouse livers (Fig. 1C and supplemental Fig. 1B). Keap1-null mice, which exhibit constitutive activation of the Nrf2 pathway, die by 3 weeks of age from malnutrition through formation of esophageal stricture (33). This knockout mouse can be rescued by reducing the function of Nrf2 through genetic engineering. To analyze the effect of Nrf2 on Notch1 expression in adult mice, Notch1 transcripts were measured in livers of 8-week old male mice of three genotypes: Keap1−/−::Nrf2+/−, Keap1+/−::Nrf2+/−, and Keap1+/−::Nrf2−/−. Although the mice of these three genotypes survive, the amount of functional Nrf2 differs among the genotypes. In the Keap1-deficient mice, which partially or completely lack this repressor of Nrf2, Nrf2 activity is higher in Keap1−/−::Nrf2+/− than in Keap1+/−::Nrf2+/− mice (33, 34). Nrf2 is absent in Keap1+/−::Nrf2−/− mice. The order of abundance of hepatic Nrf2 is Keap1−/−::Nrf2+/− > Keap1+/−::Nrf2+/− > Keap1+/−::Nrf2−/−. Notch1 expression exhibits the same qualitative pattern by genotype, as does expression of two other Nrf2 target genes Nqo1 and Gsta1 (Fig. 1C and supplemental Fig. 1C). Thus, the magnitude of Nrf2 signaling appears to affect the extent of Notch1 expression.
We investigated the possibility that Notch1 is a direct target gene of the Nrf2 transcription factor. Sequence analysis (NCBI, mouse genome sequence viewer and (35)) revealed that the proximal promoter region of Notch1 bears four consensus AREs (shown as 1 to 4 in Fig. 2A). Forced expression of Nrf2 enhanced the activity of the −1640 Notch1 luciferase reporter gene (Fig. 2B) in MEF cells. Also, in accord with the observations in MEF cells, treatment with sulforaphane significant increased Notch1 reporter activity, both in the absence and presence of exogenous Nrf2. Thus, Nrf2 appears to directly affect Notch1 expression in MEFs and this is likely mediated by binding of Nrf2 to regulatory domain(s) in the −1640 proximal promoter of Notch1.
To elucidate whether and where any functional enhancer regions of the Notch1 are located, serial deletion reporter constructs bearing the Notch1 5μ-flanking region were transfected into both wild-type MEF and the P19 embryonal carcinoma cell line, in which Notch1 is abundantly expressed within differentiation (36). We identified the region from −315 to −189 as the positive regulatory region in both cell types (Fig. 2C). One consensus ARE core sequence (1-ARE), TGABNNNGC (B=C, G, T, N=A, T, G, C) (1), was present in the region −204 to −196. To elucidate whether the 1-ARE was functional, we created p–206 and point mutant reporter genes (Fig. 2D) and transfected them into MEFs. Co-expression of Nrf2 with p-206 led to a 4-fold enhancement in reporter activity compared to that in vector-transfected cells (Fig. 2D). The responsiveness of a p–206 mutant ARE (mARE) luciferase reporter to Nrf2 was reduced compared to the wild-type p-206 reporter. These results suggest that the 1-ARE in the Notch1 gene regulatory region is a functional cis-element in MEFs.
We performed a series of electrophoretic mobility shift assays (EMSAs) to determine whether the positive regulatory element directly binds transcriptional factors. There are three major Cnc candidates for binding to AREs and stimulating gene expression: Nrf1, Nrf2, and Nrf3 (37). Nrf3 was considered the least likely molecule because of its low expression among the three candidates in MEFs. Neither Nrf1 nor Nrf3 gene expression was altered in the wild-type compared to Nrf2−/− MEFs (supplemental Fig. 1A). We considered Nrf2 the strongest candidate on the basis of the comparison of Notch1 expression in Nrf2-null versus wild-type MEFs. To determine whether Nrf2 formed a complex with 1-ARE from the Notch1 gene regulatory region, we prepared recombinant Nrf2 and its binding partner, small MafK. Neither Nrf2 alone nor MafK alone bound the Notch11-ARE (Fig. 2E, left). Formation of the heterodimeric complex (Nrf2-MafK) bound to the Notch1 1-ARE only occurred when both proteins were added (Fig. 2E, right). The intensities of the bands representing complex formation using both recombinant proteins were reduced by the addition of a cold competitor oligonucleotide, the ARE from Gsta1, but not by a point mutant oligonucleotide of the Notch1 1-ARE. Thus, Notch1 1-ARE has cis-element potential for Nrf2-MafK heterodimer binding.
The 1-ARE formed complexes when incubated with nuclear extracts from MEFs (Fig. 2F). The band indicating the binding complex disappeared following addition of the Gsta1-ARE as a competitor, but not in the presence of excess mutant Notch1 1-ARE (Fig. 2F). An EMSA conducted with an antibody to Nrf2 showed supershifted bands from the complex; however, we did not observe supershift or disappearance of the bands in the presence of an antibody to Nrf1 (Fig. 2G). To determine if basal Notch1 expression is stimulated by Nrf2 through 1-ARE in MEFs, we performed a chromatin immunoprecipitation (ChIP) assay with wild-type and Nrf2−/− MEFs. Under stringent conditions in which the positive control Gsta1-ARE locus (38) was barely detectable, the antibody to Nrf2 precipitated the proximal region of the Notch1 promoter, including 1-ARE, in nuclear extracts from wild-type, but not Nrf2−/−, MEFs. Collectively, the promoter reporter, EMSA, and ChIP results indicate that Nrf2, likely in complex with other bZip transcription factors such as the small Mafs, form a complex with the 1-ARE of the Notch1 gene regulatory region.
In the Keap1−/− liver, Nrf2 accumulates in the nucleus due to impaired Keap1-mediated Nrf2 degradation. We found that the amount of Notch1 transcripts was increased in Keap1−/− P1 liver (supplemental Fig. 1D). However, the constitutively active status of Nrf2 may produce indirect effects on gene expression. Indeed, these Keap1−/− mice do not survive more than 3 weeks (33). To avoid such constitutive Nrf2 activation and toxicity, we isolated conditional Nrf2-activating MEFs (RCKF) from RosaCreER/CreER::Keap1flox/flox mice as well as control MEFs (RC) from RosaCreER/CreER mice. Keap1 was disrupted in RCKF cells exposed to 4-hydroxytamoxifen (4-OHT), but not when exposed to vehicle (Fig. 3A). In control RC MEFs, 4-OHT did not induce Notch1 expression or that of Nqo1, a representative Nrf2 target gene (Fig. 3B). Thus, 4-OHT does not itself influence Nrf2-related gene expression. Expression of both Nqo1 and Notch1 was induced in RCKFs treated with 4-OHT (Fig. 3B), and this was accompanied by the accumulation of Nrf2 (Fig. 3C), and disruption of Keap1 (Fig. 3A). Thus, alteration of the abundance of cellular Nrf2 directly influenced the abundance of Notch1 transcripts.
We developed a co-culture model to determine whether downstream Notch1 signaling was affected by Nrf2 genotype. Nrf2-disrupted or wild-type MEFs were each co-cultured with clones of human embryonic kidney 293 (HEK293) cells stably over-expressing Notch1 ligands (human DLL1 and human JAG1), which served as pathway activators (Fig. 4). We assessed the amount of the exogenous Notch1 ligands JAG1 and DLL1 in the HEK293 cells by immunoblot analysis (Fig. 4A). To evaluate the activity of the Notch1 pathway, we monitored the expression of the direct Notch1 target genes Hes-1, Herp1, Herp2, and Nrarp. When wild-type MEFs were co-cultured with J1 or J2 HEK293 cells, both of which stably express human Jag1 protein, we observed a significant increase in the expression of the Notch1 target genes compared to that in co-cultures with clone 3 of mock-transfected HEK293 cells. Similar fold induction of Hes-1, Herp1, Herp2, or Nrarp was seen when the J1 or J2 lines were co-cultured with Nrf2−/− MEFs, although absolute expression was considerably lower than in wild-type MEFs. With the DLL1-expressing, D2 stable clone, Hes-1 transcripts were clearly increased in co-cultures with wild-type MEFs, but showed a dampened response in co-cultures with Nrf2-null MEFs (Fig. 4C). Herp1, which is another direct target gene of Notch1, also exhibited increased transcript abundance in wild-type, but not Nrf2-null MEFs co-cultured with DLL1-overxpressing HEK293 cells. Thus, Nrf2-null MEFs show significantly decreased Notch responses.
Notch signaling is essential during embryogenesis of mice. Conditional gene targeting studies have revealed that Notch signaling is also important after birth, as well as in adult mice (39, 40). Notch1 is activated and plays an important role in cell proliferation during liver regeneration after partial hepatectomy. Liver injection of silencing RNA for Notch1 or Jag-1 prior to partial hepatectomy, significantly suppresses the proliferation of hepatocytes at days 2 to 4 of the regenerative response (26). Liver regeneration from partial hepatectomy is also delayed in Nrf2-disrupted mice compared to that in wild-type mice (41). We confirmed this delayed liver regeneration at 72 hr after partial hepatectomy in Nrf2-null mice (Fig. 5B). We hypothesized that this delay was associated with altered Notch1 expression and signaling in the Nrf2-disrupted mice. To verify this hypothesis, we performed two-thirds partial hepatectomy (both median and left lobes) in Nrf2-null mice and measured the transcription of the Notch1 target gene Hes1 to evaluate the impact on Notch1 signaling (Fig. 5A). At 30 min after partial hepatectomy, the abundance of transcripts for Nqo1, which reflects Nrf2 activity, and Hes1, which reflects Notch1 signaling, was similar in livers of mice of each genotype to the abundance prior to hepatectomy (supplemental Fig. 2). However, at 3 hr after partial hepatectomy, there were increases in transcripts encoding the Notch1 ligands (Jag1 and Dll1) in livers of both genotypes (Fig. 5A). The abundance of the Notch1 transcript in livers of Nrf2−/− mice was less than the abundance in livers from wild-type mice, and the Nrf2−/− mice failed to induce Notch1 6 hours following partial hepatectomy (Fig. 5B). Hes1 transcripts were more abundant in the livers of wild-type mice compared to the amount in livers of Nrf2−/− mice. Although the fold induction of Hes1 was similar between genotypes, absolute levels of Hes1 transcripts were lower following partial hepatectomy in Nrf2−/− mice (Fig. 5B). The amount of basal Nqo1 transcripts was also reduced in the Nrf2-null mice and the induction in response to partial hepatectomy was completely blocked. In the livers of sham-operated mice, the abundance of transcripts for the direct partner of Notch1 (Rbpj) for the activation of the target genes was not different in the livers of Nrf2−/− mice compared to that in the wild-type mouse livers, and the abundance of these transcripts decreased 6 hours after partial hepatectomy only in the Nrf2−/− mice (Fig. 5B). This result suggests that the different responses of Nqo1 (absence of induction) and Hes1 (reduced induction) to partial hepatectomy between wild-type and Nrf2-null mice are caused by differences in initial amounts of Nrf2 and Notch1 proteins in the liver. Absence of Nrf2 in the liver seemed to affect Notch1 signaling in early phase of regenerative process after 2/3 partial hepatectomy due to the lower basal transcript abundance of Notch1 in the Nrf2-null mice.
To restore Notch1 signaling in the hepatocytes in the livers of Nrf2−/− mice, triple compound Nrf2−/−::RosaNICD/−::AlbCre mice were established by crossing Nrf2-null mice with RosaNICD/− or AlbCre transgenic mice in a step-by-step manner. Only in this composite line of mice is the Notch intracellular domain (NICD), which is the transcriptionally active Notch1 cleavage fragment, and enhanced GFP (EGFP) in hepatocytes present due to the removal of the transcriptional stop signal sequence placed between both loxP sequences directly downstream of the Rosa 26 promoter region by the specific expression of Cre recombinase controlled by the albumin enhancer promoter (Fig. 6A). Homologous recombination in the Rosa 26 locus (Cre-active) leading to expression of NICD was only detected in the DNA isolated from the liver of Nrf2−/−::RosaNICD/−::AlbCre mice (Fig. 6B). Both NICD and EGFP were only present in the triple compound mice, detected by immunoblot analyses of whole liver extracts (Fig. 6C). The abundance of transcripts for the Notch1 direct target gene Hes1 were increased in the Nrf2−/−::RosaNICD/−::AlbCre mice compared to the other two lines of mice (Fig. 6D). Thus, this triple compound line exhibited enhanced Notch1 signaling in hepatocytes. Two-thirds partial hepatectomies were conducted using this line and two sets of control lines, Nrf2−/−:: RosaNICD/− and Nrf2−/−::AlbCre. There were no significant differences among three genotypes in the size of livers in animals undergoing sham partial hepatectomy (Fig. 6E). But, following partial hepatectomy, the Nrf2−/−::RosaNICD/−::AlbCre line showed enhanced regeneration of liver weight comparable to that seen in wild-type mice (Fig. 6E, compare with Fig. 5B). Neither of the control lines (Nrf2−/−::RosaNICD/− or Nrf2::AlbCre) influenced the delayed regeneration rate typical of the Nrf2 -null mice. These results demonstrate that constitutive expression of the gene encoding Notch1 and Notch1 signaling are supported by Nrf2 and contibute to tissue regeneration.
We provide several lines of evidence indicating that Nrf2 enhances the expression of the gene encoding Notch1 and the amplitude of its downstream signals (Fig. 1). First, a comparison of expression profiles in MEFs isolated from wild-type and Nrf2-disrupted mice showed that Notch1 expression was dampened in the Nrf2-null cells. Additionally, transcripts for downstream effectors of Notch1, Hes-1, Herps, Nrarp, and p21, were also diminished in the knockout cells. Second, the Notch1 target genes were induced by treatment of wild-type, but not Nrf2-knockout cells, with the chemopreventive agent sulforaphane, a known activator of Nrf2 signaling. Third, genetic manipulation of the abundance of Nrf2 in mouse liver-- by disruption of the genes encoding Nrf2 and its repressor, Keap1-- demonstrated a dose-dependent association between magnitude of Nrf2 signaling and Notch1 expression. Notch1 expression was decreased in the livers of Nrf2-null mice, but was enhanced in Keap1−/−:: Nrf2+/− mouse livers, which is similar to the expression of prototypic Nrf2-regulated genes Gsta1 and Nqo1 in these genotypic settings. Although the Keap1−/−:: Nrf2+/− mice exhibit a constitutive activation of Nrf2, these animals do not exhibit the pre-weaning lethal phenotype of Keap1-null mice.
The co-culture experiments with HEK293 cells overexpressing the Notch1 ligands JAG1 or DLL-1 indicated that decreased expression of Notch1 in Nrf2-null MEF directly impaired Notch1 signal transduction, as reflected in reduced expression of its target genes Hes-1, Herps, and Nrarp. Thus, Notch1 signaling is limited by the abundance of Notch1, which is transcriptionally controlled by Nrf2, not by the abundance of the Notch1 ligands. This ligand-Notch1 interaction may mimic events early in tissue regeneration after injury. Partial hepatectomy provides a model to examine signaling pathways involved in the response to hepatic injury. Prior studies have shown that Notch1 signaling increases rapidly following partial hepatectomy (26), and that liver regeneration is delayed in Nrf2-null mice (41), an observation that we confirm here. Köhler et al. (26) observed that nuclear translocation of NICD increased and peaked within 15 minutes after partial hepatectomy in rats, indicating activation of Notch signaling. In the same study, expression of the Notch-dependent target gene Hes1 increased within 30 to 60 minutes. We report that at 6 hr after partial hepatectomy no change in abundance of Notch1 and Nrf2 transcripts was observed, but Hes1 transcripts, as well as those encoding the Notch1 ligands Jag1 or Dll-1, were increased. Although we observed similar relative increases of Hes1 transcripts in the livers of Nrf2-null mice following partial hepatectomy, the absolute amount was only one-third of that in wild-type mice. This phenomenon was similar to the observations in the in vitro co-culture system using MEFs. Because the abundance of the transcripts of the Notch1 ligands did not appear to be influenced by Nrf2 genotype following partial hepatectomy, the differential upregulation of Hes1 by genotype may reflect abundance of Notch1 protein. Notch1 signaling, which is supported by Nrf2, appeared to influence early steps in liver regeneration and we tested this by creating Nrf2−/−::RosaNICD/−::AlbCre mice, in which have NICD is produced in specifically in hepatocytes in the Nrf- null background. Forced hepatic Notch1 signaling mediated by NICD rescues the phenotype of delayed liver regeneration observed in the Nrf2-null mice.
Other effects of Nrf2 on Notch1 signaling can be envisioned. Nrf2-null mice do not show any growth and development phenotypes under normal conditions, either in early development or postnatally (3, 42). By contrast, Notch1 disruption is lethal to mice at around embryonic day 10.5 (E10.5), indicating that this gene and its signals are essential in embryonic development stage before E10.5 (21). The Nrf2 transcript is detected at this time only in the central nervous system by RNA in situ hybridization and this is the earliest point of detection (43). Thus, Notch1 is likely present before Nrf2 (Gene Expression Data from MGI), suggesting that expression of Notch1 in early development does not depend on Nrf2. Alternatively, considering Nrf1 expression patterns during embryonic development and the embryonic lethal phenotype of Nrf1-disrupted mice, Nrf1 may substitute for Nrf2 to support Notch1 expression through the 1-ARE early in life. Further analyses are required to elucidate how ARE-containing genes are regulated by Nrf2, Nrf1, or other factors during embryogenesis.
Nrf2-mediated regulation of Notch1 expression may be most critical in postnatal stages, rather than during development when Notch1 is likely regulated by other factors, because Nrf2 disruption does not produce altered phenotypes during early development stages. Nrf2 is abundant in liver, kidney, lung, and the gastrointestinal-tract, tissues that also contain stem cells. Notch and Notch-related gene expression also occur in these tissues (26, 28, 29, 44, 45). In such tissues, damaged cells must be replaced by newly differentiated cells for routine tissue maintenance. Furthermore, many of these tissues face the highest burdens from environmental exposures to toxicants and resultant tissue injury. The Keap1-Nrf2 pathway regulates a well described adaptive response that reduces macromolecular damage, following exposure to conditions that increase the amounts of free radicals and electrophiles, through stimulation of genes encoding antioxidative and electrophile detoxication enzymes (4). Nrf2 signaling also stimulates pathways affecting the recognition, repair, or removal, or all of these, of damaged macromolecules, notably proteins and DNA (38). Our previous work indicates that repair of tissue damage represents a third component of the adaptive cytoprotective response controlled by Nrf2. Nrf2-null mice are considerably more sensitive to hyperoxic lung injury (46). Treatment of mice with the triterpenoid Nrf2 activator CDDO-Imidazolide during, but not before hyperoxic exposure, led to substantial protection against the biochemical and morphological sequelae of acute lung injury and appeared to enhance tissue remodeling (47). The current study indicates that Nrf2 facilitates tissue regeneration in the liver by regulating Notch1 expression. With these diverse actions influencing the cellular response to toxins or tissue damage, strategies using drugs or foods to enhance Nrf2 signaling and that of its interacting networks provide multifaceted opportunities for disease prevention.
Keap1+/− mice were back-crossed more than 10 times with C57BL/6J mice purchased from the Jackson Laboratory. Nrf2−/− mice, which were back-crossed more than 20 times into a C57BL/6J background, were a gift from Dr. Ken Itoh (Hirosaki University). Keap1-Nrf2 compound knockout mice were established by mating Keap1+/− with Nrf2+/− mice to generate Keap1+/−::Nrf2+/− mice. Each compound genotype was obtained from matings between Keap1+/−::Nrf2+/− mice (33). Keap1flox/+ mice were generated as described previously (8). Gt(ROSA)26Sortm1(Cre/Esr1)Nat mice ( RosaCreER/CreER) (48), Gt(ROSA)26Sortm1(Notch1)Dam/J (RosaNICD/−)(49) and B6.Cg-Tg(Alb-cre)21Mgn/J (AlbCre) (50) were obtained from the Jackson Laboratory. Keap1flox/flox :: RosaCreER/CreER mice were established by matings with Keap1flox/+ and RosaCreER/CreER mice in consecutive steps. Nrf2−/−:: RosaNICD/− ::AlbCre mice resulted from crossings with Nrf2−/−:: RosaNICD/NICD and Nrf2−/−::AlbCre mice, which were established by the consecutive mating of Nrf2+/−::AlbCre and Nrf2+/−:: RosaNICD/− mice. All mice were genotyped by previously reported methods (8, 33, 48) or by PCR described in the table S2. All animal breeding and handling was conducted in accordance with protocols approved by the Animal Care and Use Committee of the Johns Hopkins Medical Institutions.
Male mice (9- to 10-week old) were placed into a plexiglass chamber for induction of anesthesia with 2% isoflurane and 2 liter/min oxygen flow. After primary anesthetization, they were maintained under anesthesia by isoflurane inhalation through a suitable mouthpiece. The left and median lobes of the liver were pushed out from an incision just under the xiphoid process and were removed with a ligature (51). Following surgery, mice were returned to their cages and given free access to food and water. Sham-operated control animals were treated in an identical fashion with the omission of hepatectomy.
MEFs were established from the embryos of Nrf2-null or wild-type littermates. On day 14.5 of pregnancy, mice were sacrificed and embryos removed from the uterus. Extra-embryonic tissues were removed from individual fetuses, followed by rinsing in PBS to remove any blood. Embryos were decapitated and eviscerated using sterile forceps. A portion of the tail was used as a source of DNA for genotype and gender confirmation (52) by PCR. The carcasses were transferred to a sterile conical centrifuge tube containing PBS and rinsed by inverting twice. The carcasses were dissociated by finely mincing with a scalpel blade, then pressing through a sterile cell strainer with forceps. They were rinsed on the screen twice with a 0.05% trypsin solution, transferred to a small culture flask, and stirred gently in a humidified 5% CO2 incubator at 37°C for 30 minutes. Suspended cells were decanted and collected by centrifugation. Cell pellets were washed twice with 30 ml Iscove’s MEM (Invitrogen, USA) containing 10% fetal bovine serum (FBS) (Invitrogen, USA) (MEF growth medium). Cells were plated on dishes in MEF growth medium; 24 hr later the cells were rinsed with PBS/0.02% EDTA twice and the medium was replaced with fresh MEF growth medium. In approximately 2–4 days, the cultures were confluent and expanded for use in experiments and for making frozen stocks. MEFs derived from male embryos were used in all experiments. P19 cells (ATCC) were grown as monolayers and maintained in α-minimum essential medium (Sigma, USA), containing 10% fetal bovine serum, and incubated at 37°C in a humidified atmosphere of 5% CO2. To establish human Jagged1 or human Delta-like 1 expressing cells, pLZRShJagged1 or pLZRhDll1 (53), provided by Dr. Leonor Parreira (University of Lisbon), were transfected into HEK293 cells. Simultaneously, negative-control transfectant, which was transfected with LZRSpBMN-linker-IRES-EGFP as a mock vector, was also prepared. After 2 weeks of selection with 10 μg/ml puromycin, several single resistant colonies were picked and maintained in each dish using Iscove’s MEM containing 10% FBS with 8 μg/ml puromycin. RCKF and RC MEFs were isolated from embryos of RosaCreER/CreER:: Keap1flox/flox or RosaCreER/CreER pregnant mice (ED14.5), respectively. Each MEF culture was established from single embryos. Experiments were conducted within passage number 5. 4-Hydroxy-tamoxifen (4-OHT, Sigma, USA) was dissolved in 100% ethanol at 10 mM. RCFK and RC were analyzed after 3 days-repetitive exposure to conditioned media that included 1 μM 4-OHT.
Cells expressing the Notch1 ligand or negative control cells (5 × 105) were spread in 10-cm dishes 4 days before co-culture with Nrf2-null or wild-type MEFs. Nrf2-null or wild-type MEF were plated at 4 × 105 cells/dish. Notch ligand-expressing cells or control cells were collected with PBS/EDTA (final 0.02% EDTA) and washed once before seeding. The transfectants were split into dishes of each MEF genotype, containing complete medium and incubated at 37°C, 5% CO2. Total RNA was isolated and analyzed by RT-PCR. It was confirmed that the primer set only amplified products from MEF under our PCR conditions.
Proper size-cut tissues were homogenized in RIPA-I buffer, which contained a protease inhibitor cocktail (Roche, USA). Cultured cells were harvested with RIPA-I buffer directly. An equal volume of 2xSDS sample buffer was added and the samples were were denatured by boiling for 5 min. Samples were applied to SDS-PAGE and transferred to an Immobilon PVDF membrane (Millipore, USA). The membranes were blocked with Tris-bufferred saline with 0.05% Tween 20 and 5% skim milk and then treated with a primary antibody. The preparative membranes were reacted with appropriate secondary antibodies conjugated to horseradish peroxidase (Invitrogen Zymed, USA). The immunocomplexes were visualized with ECL (GE-Healthcare, USA). The specific antibodies for Lamin B1 (sc-6216), EGFP (sc-8334), Jagged-1 (sc-8303), and Delta-1 (sc-9102) were obtained from from Santa Cruz Biotechnology. Antibodies for Cre recombinase (ab24608) and NICD (07-1232) were purchased from Abcam and Millipore, respectively. The monoclonal antibody against Nrf2 used for immunoblotting was a gift from Dr. Ken Itoh (Hirosaki University).
Total RNA was purified using the RNeasy Mini kit (Qiagen, USA) after isolation using TRIzol (Invitrogen, USA) according to the manufacturer’s instructions. Murine Genome MOE 430A GeneChip arrays (Affymetrix, USA), which contain probes for detecting ~14,500 well-characterized genes and 4371 expressed sequence tags were used to probe the gene expression in MEFs as described previously (12). Scanned output files were analyzed using Affymetrix GCOS (GeneChip Operating Software) and independently normalized to an average intensity of 500 before comparison.
The mouse Notch1 gene regulatory region (−2 kb) from the major transcripts including the initiation codon sequence was isolated from C57BL/6J mouse liver genomic DNA by PCR. The region was directly cloned between Bgl II and Nco I sites in pGL3 basic (Promega, USA) and confirmed by sequence analysis. In this construction, the ATG of Notch1 was replaced with the ATG of luciferase cDNA. Serial promoter deletion fragments were prepared by the S1-ExoIII nuclease reaction method followed by Nco I digestion. Then, the fragments were ligated to the site between Ecl 136II and NcoI sites of pGL3 basic. Serial point mutant and p-206 constructs were produced by PCR primer site-directed mutagenesis. Primers are shown in table S1. All constructs were confirmed by sequencing. Because pRLTK (Promega, USA) bears an ARE sequence in the thymidine kinase-promoter region, this ARE sequence was deleted in order to use this vector for normalizing transfection efficiency. The improved normalizing vector, pRLTK-ΔARE, was constructed by deleting the Sma I and Pvu II fragment from pRLTK. Constructs used in transfection experiments were purified with Qiagen plasmid kits (Qiagen, USA) and detoxinized with an endotoxin removal kit (Mirus Bio Corp., USA).
MEFs or P19 cells were plated at a density of 2 × 105 cells/60-mm dish 24 hr before DNA transfection. An equimolar concentration of the DNA of each construct (10 μg = 4.3 pmol of each reporter gene and 0.5 μg of pRLTK-ΔARE) was introduced into the cells by calcium phosphate co-precipitation (54). Transfections were repeated three to six times with independent plasmid DNA preparations. pBR322 plasmid DNA was used as a carrier to adjust the total amount of DNA. Cells were harvested 48 hrs after transfection. Luciferase activity was measured according to the manufacturer’s instructions (Promega, USA) and normalized to Renilla luciferase activity derived from pRLTK-ΔARE. Sulforaphane (LKT Labs, MN) was used as an inducer of ARE-regulated genes (55). For trans-activation assays, 4 μg of the reporter gene was transfected together with 8 μg pCMVmNrf2 and 0.5 μg pRLTK-ΔARE normalizing vector.
Cells were seeded at 60% confluence the day before treatment with vehicle or sulforaphane. Total RNA was prepared from MEFs or P19 cells using Trizol (Invitrogen, USA) or Isogen (Wako, USA), and then, total RNA was purified using the RNeasy mini kit (Qiagen, USA). Mouse livers were perfused with ice-cold PBS prior to isolation of total RNA using the Versagene RNA purification system (Gentra Systems, USA). RNA quality was confirmed by electrophoresis before reverse-transcriptase (RT) reaction. Several genes were analyzed by SYBR green real time quantitative RT-PCR, and representative differentiation marker genes were analyzed by semi-quantitative RT-PCR. cDNA was synthesized using the iScript system (BioRad, USA). Real-time PCR was performed on a Bio-Rad My-IQ real-time PCR machine using Applied Biosystems SYBR green PCR master mix in triplicate 20 μl reaction volumes. The PCR efficiency was determined from a standard curve and used the Pfaffl method for calculation of fold changes (56). Melt curves and agarose gel electrophoresis were employed to ensure specificity of amplified product. Primers are shown in table SI. The intensity of bands was measured with ImageJ analysis software provided from NIH.
Nuclear extracts were prepared from MEF and P19 cells according to Dignum et al (57). Protein concentrations of each nuclear extract were determined by the BioRad protein assay using IgG as the standard. Probes were prepared by end labeling 20 ng of the primary strand oligonucleotide with [γ-32P] ATP by T4 polynucleotide kinase, which were annealed subsequently to the complementary oligonucleotide. For the standard electrophoretic mobility shift assay (EMSA), 25 μl reaction mixture containing approximately 2 x104 cpm of probe nucleotide was incubated with 20μg protein of the nuclear extract in a buffer consisting of 10 mM HEPES (pH 7.9), 25 mM KCl, 1 mM EDTA, 1 mM ZnCl2, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 4 μg poly(dI-dC), 0.05% Nonidet P-40, and 10% glycerol at 4°C for 30 min. For the competition experiments, each competitor oligonucleotide (table S1) was also added to the standard EMSA reaction mixture at a 25, 50, or indicated-fold molar excess to the probe, containing approximately 2 x104 cpm, and the final volume was adjusted to 25 or 30 μl. The reaction products were loaded onto a 5.0% or 8.0% poly-acrylamide gels and run at 20 mA for 1.5 h in 0.5 × TBE (Tris-borate-EDTA buffer, 45 mM Tris-borate and 1 mM EDTA) at 4°C. All EMSA experiments were repeated three times utilizing nuclear extracts from MEF and P19 cells treated or untreated with 2.5 μM sulforaphane. Gels were dried and exposed to X-ray film at −80°C for 12 hr. In the case where recombinant Nrf2 and small MafK protein was used, the EMSA was performed as described previously (3).
Formaldehyde cross-linking and chromatin fragmentation were performed as described previously (30). Diluted chromatin solution was incubated with an antibody to Nrf2 (H-300: sc-13032X, Santa Cruz Biotechnology, USA), nonspecific immunoglobulin G (normal Rabbit IgG sc-2027, Santa Cruz Biotechnology, USA), or solution without antibody for 18 h at 4°C with rotation. After washing and elution, precipitated DNA was dissolved in 60 μl of water. DNA solution (2 μl) as a template was used for PCR with the following primers: −187mNotch1-CP: 5μ-CTCCCTCCCGCGGCAGAGGCAC-3μ and −257mNotch1-CP: 5μ-CGCAGGAACCAGGGGCGGAGCC-3μ. The primer sequences for promoters of β-Actin, which does not contain an ARE, and Gsta1, which bears functional AREs, have been described previously (17).
All values are expressed as mean ± SD, except as noted otherwise. Statistical analysis was performed with unpaired Student’s t-test or one-way analysis of variance (ANOVA) for comparison of multiple groups. Differences between groups were considered statistically significant when P < 0.05.
Supplemental Fig. 1 Differential exprression of Notch1 in Keap1-Nrf2 compound knockout MEFs and livers. A. Semi-quantitative RT- PCR for Nrf1 and Nrf3 genes. The abundance of the transcript for Tata binding protein (TBP) served as a loading control. Data are quantified as the ratio of the expression of the indicated genes in Nrf2 null/wild type. Values are the means of 3 independent experiments ± S.E. Expression of both genes was not altered between Nrf2 genotypes in MEFs (t- test). B. Real time RT- PCR analyses of Notch1 and Gsta1 expression in the livers of 8-week old Nrf2-null and wild-type mice. n=3, *p<0.01 C. Real time RT- PCR analyses of Notch1 and Gsta1 expression in the livers of various 8-week old Keap1 and Nrf2 compound-disrupted mice. n=3, *p<0.01. K0N1, K1N1 and K1N0 indicate Keap1−/−::Nrf2+/−, Keap1+/−::Nrf2+/−, and Keap1+/−::Nrf2−/− mice, respectively. C. RT- PCR analyses of Notch1 and Nqo1 expression in the livers of postnatal day 1 Keap1−/− mice. Hprt served as the loading control.
Supplemental Fig. 2. Nqo1 and Hes1 expression at 30 min after two-third partial hepatectomies. Hes1 and Nqo1 expression were monitored by real-time reverse transcriptase PCR as an index of Notch1 signaling 0.5 hr after 2/3 partial hepatectomy. Albumin expression was used for normalization. The ratio of hepatic gene of interest to albumin gene expression in Nrf2+/+ mice at time 0 was set at a value of 1.0. Values are mean ± S.D., n= 3, * p < 0.05, one-way analysis of variance.
Table S1. Oligonucleotides used in experiments.
Table S2. PCR conditions for genotyping.
We thank Drs. H. Motohashi (Tohoku Univ.), K. Itoh (Hirosaki Univ.), S. Takahashi (Univ. of Tsukuba), M. Morita (The Salk Inst. for Biol. Studies), A. P. McMahon (Harvard Univ.), C. Murtaugh (University of Utah), I. Alcobia (Univ. of Lisboa) and L. Parreira (Univ. of Lisboa) for MafK expression vectors, anti-Nrf2 antibody, pCaggs-nlsCre, pCaggs-Cre-EGFP, pCAGGCre-ER™ expression vectors, genomic sequence information for RosaNICD mice, human Dll-1 and Jag1 expression plasmids, respectively. We also appreciate the technical assistance of Mr. Patrick Dolan, Mr. Richard Rabold, Ms. Patricia Egner, Mr. John Skoko, Drs. Miho Iijima and Hiromi Sesaki. Funding sources: This work was supported by the Maryland Cigarette Restitution Fund, NIH grants R01 CA94076, R01 HL081205, P30 ES03819, T32 ES07141, T32 CA009110 and JST-ERATO. Soona Shin is the recipient of a Samsung Scholarship (Samsung Foundation of Culture).
Competing interests: There may be MTAs related to the new mouse constructs, per the policy of Johns Hopkins University.
Author contributions: NW designed and performed the research, established the NICD and compound genetically engineered mice, analyzed the data and wrote the manuscript; SS conducted the ChIP assays and together with JW conducted the RT-PCR analyses; SLS established the conditional Keap1 disrupted MEFs; ESA performed the RT-PCR in MEF and assisted in writing the manuscript; MK, VM and SB conducted and analyzed the initial microarray studies; MY provided the Nrf2 and Keap1 null mice and helped evaluate the data; and TWK hepled initiate the project and assisted with research design and manuscript preparation.