The consensus cis-regulatory AP-1 (activator protein-1)-like AREs (antioxidant-response elements) and/or EpREs (electrophile-response elements) allow for differential recruitment of Nrf1 [NF-E2 (nuclear factor-erythroid 2)-related factor 1], Nrf2 and Nrf3, together with each of their heterodimeric partners (e.g. sMaf, c-Jun, JunD or c-Fos), to regulate different sets of cognate genes. Among them, NF-E2 p45 and Nrf3 are subject to tissue-specific expression in haemopoietic and placental cell lineages respectively. By contrast, Nrf1 and Nrf2 are two important transcription factors expressed ubiquitously in various vertebrate tissues and hence may elicit putative combinational or competitive functions. Nevertheless, they have de facto distinct biological activities because knockout of their genes in mice leads to distinguishable phenotypes. Of note, Nrf2 is dispensable during development and growth, albeit it is accepted as a master regulator of antioxidant, detoxification and cytoprotective genes against cellular stress. Relative to the water-soluble Nrf2, less attention has hitherto been drawn to the membrane-bound Nrf1, even though it has been shown to be indispensable for embryonic development and organ integrity. The biological discrepancy between Nrf1 and Nrf2 is determined by differences in both their primary structures and topovectorial subcellular locations, in which they are subjected to distinct post-translational processing so as to mediate differential expression of ARE-driven cytoprotective genes. In the present review, we focus on the molecular and cellular basis for Nrf1 and its isoforms, which together exert its essential functions for maintaining cellular homoeostasis, normal organ development and growth during life processes. Conversely, dysfunction of Nrf1 results in spontaneous development of non-alcoholic steatohepatitis, hepatoma, diabetes and neurodegenerative diseases in animal models.

INTRODUCTION

All organisms living in the oxygenated environment (possibly with toxic pollutants) have evolutionarily established effective antioxidant, detoxification and cytoprotective defences, in order to maintain cell homoeostasis and organ integrity during development and life processes. Under normal physiological conditions, ROS (reactive oxygen species) and derivatives serve as sensors and/or messengers to trigger a vast variety of cellular functional responses through discrete signalling networks [1,2]. Consequently, these networks control transactivation activity of many distinct transcription factors, such as the CNC (cap'n’collar) bZIP (basic-region leucine zipper) family that comprises the vertebrate activator NF-E2 (nuclear factor-erythroid 2) p45 and Nrf1 (NF-E2-related factor 1) [including its long TCF11 (transcription factor 11) and its short Nrf1β/LCR-F1 (locus control region factor 1)], Nrf2 and Nrf3, along with the repressors Bach1 (BTB and CNC homology 1) and Bach2, and the Caenorhabditis elegans protein Skn-1 (skinhead-1), besides the founding member Drosophila Cnc protein, as well as AP1 (activator protein 1), HIF-1 (hypoxia-inducible factor 1), NF-κB (nuclear factor κB) and the tumour suppressor p53 [35]. As a result, these transcription factors regulate expression of their target genes through specific binding to the cognate cis-acting elements in the promoter regions, important among which are designated AREs (antioxidant-response element) and/or EpREs (electrophile-response element) [69], with homologues depicted in Table 1. The ARE battery mediates a large number of cytoprotective genes encoding: (i) antioxidant defence enzymes, such as NQO1 [NAD(P):quinone oxidoreductase 1), AKRs (aldoketoreductases) and HO-1 (haem oxygenase 1); (ii) phase 2 detoxification enzymes [e.g. GSTs (glutathione S-transferases) and UGTs (UDP-glucuronosyl tranferases)]; (iii) enzymes in glutathione synthesis and modulating enzymes [e.g. GCLC (glutamate–cysteine ligase catalytic) and GCLM (glutamate–cysteine ligase modifier) subunits, GSS (glutathione synthetase), GPx (glutathione peroxidase), GR (glutathione reductase), TPx (thioredoxin peroxidase) and TR (thioredoxin reductase)]; (iv) enzymes for NADPH generation (e.g. malic enzyme); (v) other oxidoreductases (e.g. NADPH cytochrome b5 oxidoreductases); and (vi) phase 3 enzymes involved in cellular efflux, such as MRP (multidrug-resistance-associated protein)-1 and MRP-2 [10,11].

Table 1
Comparison of the ARE sequence with its homologues, together with other cis-regulatory consensus sequences

Abbreviations: AARE, amino acid-response element; ARE, antioxidant-response element; AP1, activator protein 1; CRE, cAMP-response element; C-MARE, CRE-type Maf-recognition element; CNC, cap‘n'collar; C/EBP, CCAAT/enhancer-binding protein; EpRE, electrophile-response element; HRE, hypoxia-response element; NF-E2, nuclear factor-erythroid 2; NF-κB, nuclear factor κB; Skn-1, skinhead-1; SRE, sterol-regulatory element; TRE, TPA (PMA)-response element (also called AP1-binding site); T-MARE, TRE-type Maf-recognition element; UPRE, unfolded protein response element; XRE, xenobiotic-response element. Alternative single nucleotide code: M=A or C; R=A or G; Y=C or T; B=C, G or T; W=A or T. The extended ARE/EpRE sequences were reported by three independent groups as indicated in the table bottom [84,184,188].

Abbreviations: AARE, amino acid-response element; ARE, antioxidant-response element; AP1, activator protein 1; CRE, cAMP-response element; C-MARE, CRE-type Maf-recognition element; CNC, cap‘n'collar; C/EBP, CCAAT/enhancer-binding protein; EpRE, electrophile-response element; HRE, hypoxia-response element; NF-E2, nuclear factor-erythroid 2; NF-κB, nuclear factor κB; Skn-1, skinhead-1; SRE, sterol-regulatory element; TRE, TPA (PMA)-response element (also called AP1-binding site); T-MARE, TRE-type Maf-recognition element; UPRE, unfolded protein response element; XRE, xenobiotic-response element. Alternative single nucleotide code: M=A or C; R=A or G; Y=C or T; B=C, G or T; W=A or T. The extended ARE/EpRE sequences were reported by three independent groups as indicated in the table bottom [84,184,188].
Abbreviations: AARE, amino acid-response element; ARE, antioxidant-response element; AP1, activator protein 1; CRE, cAMP-response element; C-MARE, CRE-type Maf-recognition element; CNC, cap‘n'collar; C/EBP, CCAAT/enhancer-binding protein; EpRE, electrophile-response element; HRE, hypoxia-response element; NF-E2, nuclear factor-erythroid 2; NF-κB, nuclear factor κB; Skn-1, skinhead-1; SRE, sterol-regulatory element; TRE, TPA (PMA)-response element (also called AP1-binding site); T-MARE, TRE-type Maf-recognition element; UPRE, unfolded protein response element; XRE, xenobiotic-response element. Alternative single nucleotide code: M=A or C; R=A or G; Y=C or T; B=C, G or T; W=A or T. The extended ARE/EpRE sequences were reported by three independent groups as indicated in the table bottom [84,184,188].

In response to oxidative stress, transcriptional expression of ARE-driven genes is regulated primarily by CNC–bZIP family factors, aiming to maintain an appropriate redox homoeostasis. The adaptive defence response also enables transactivation of other ARE-battery genes, such as those encoding DNA-repair enzymes, cofactor-generating enzymes, molecular chaperones, proteasome subunits and anti-inflammatory response proteins (e.g. leucokotriene B4 dehydrogenase and ferritin) [1215]. However, a long-term redox stress, together with its secondary reactive species and radical derivatives, could over-stimulate expression of such key gene regulators that control the cell cycle, metabolism, immune responses, neurodegeneration, apoptosis, necrosis, inflammation and other biological processes. The pathological over-stimulants have been shown to contribute to carcinogenesis [1619], and aging-related neurodegenerative [20,21], inflammatory, cardiovascular and autoimmune diseases [5].

Among mammalian CNC–bZIP proteins, Nrf1 and Nrf2 are two principal factors to regulate ARE-driven cytoprotective genes against cellular stress [2224]. To date, accumulated evidence has revealed that Nrf1 fulfils a unique function that is distinctive from Nrf2, in maintaining cellular homoeostasis and organ integrity, although Nrf2 is considered to be a master regulator of adaptive responses to oxidative stressors and electrophiles [9,25]. Gene-targeting experiments demonstrate that Nrf1, but not Nrf2, is indispensable for normal development and healthy growth. Global KO (knockout) of Nrf2 (also called nfe2l2) in mice still yields viable animals [26], and such Nrf2−/− mice do not develop any spontaneous cancer and diabetes, although they are more susceptible than wild-type mice to carcinogens and oxidative stress [27]. By contrast, relatively less is known about Nrf1, although global KO of Nrf1 (also called nfe2l1) in mice causes embryonic lethality and severe oxidative stress [2831], and the conditional KO of the gene within specific organs results in different pathological phenotypes, such as NASH (non-alcoholic steatohepatitis) and hepatoma [32,33].

For reasons that are largely historical or involve difficulty of experimentation, the vast majority of researchers in the field have been almost entirely fixated on Nrf2 (to the tune of hundreds of publications per year), whereas Nrf1 appears to be comparatively ignored (with only a small number of publications, as shown in Figure 1), although the latter's CNC–bZIP factor has been shown to be actually essential for regulating critical cellular homoeostatic and developmental processes [22,24]. Nonetheless, the timeline of the major discoveries that have been made in the last two decades (Figure 2) predicts a promising tendency for Nrf1 to be coming more into its own significance lately, which will attract much attention on the factor, particularly with the demonstration that it is a critical regulator of transcriptional expression of 26S proteasomal subunits and the AAA (ATPase associated with various cellular activities) protein p97/VCP (valosin-containing protein) through a positive-feedback circuit, in response to challenges (i.e. with proteasomal inhibitors as chemotherapeutic drugs) [3436]. More interestingly much of the intracellular Nrf1 (but not Nrf2), with several domains being more conserved than its ancestral proteins CNC and Skn-1 (as illustrated in Figure 3) [37], is located in the ER (endoplasmic reticulum) and the NE (nuclear envelope) membranes [3841]. This raises an intriguing question of how Nrf1 is regulated in distinct topovectorial processes such that its main functions are efficiently exerted to maintain cellular homoeostasis and organ integrity.

Incidence of publications on transcription factors Nrf1 and Nrf2 from 1993 to 2014

Figure 1
Incidence of publications on transcription factors Nrf1 and Nrf2 from 1993 to 2014

In the last two decades, all publications on the main topics of Nrf1 (including Nfe2l1, TCF11 and LCR-F1) and Nrf2 have been collected from the scientific literature available from PubMed. The number of relevant publications per year has been calculated and is shown graphically. Note that all other publications on abbreviations confusingly similar to Nrf1 or Nrf2 have been excluded. Remarkable differences in the salient features of Nrf1 from Nrf2 are highlighted in Box 2.

Figure 1
Incidence of publications on transcription factors Nrf1 and Nrf2 from 1993 to 2014

In the last two decades, all publications on the main topics of Nrf1 (including Nfe2l1, TCF11 and LCR-F1) and Nrf2 have been collected from the scientific literature available from PubMed. The number of relevant publications per year has been calculated and is shown graphically. Note that all other publications on abbreviations confusingly similar to Nrf1 or Nrf2 have been excluded. Remarkable differences in the salient features of Nrf1 from Nrf2 are highlighted in Box 2.

Timeline of research progression in Nrf1 and its isoforms from 1993 to 2014

Figure 2
Timeline of research progression in Nrf1 and its isoforms from 1993 to 2014

The timeline represents a history of scientific research on distinct Nrf1 isoforms (together with a few cognate dimeric partners) in the last two decades. Main discoveries about Nrf1 are indicated perpendicular to the timeline. Further details are provided in the text.

Figure 2
Timeline of research progression in Nrf1 and its isoforms from 1993 to 2014

The timeline represents a history of scientific research on distinct Nrf1 isoforms (together with a few cognate dimeric partners) in the last two decades. Main discoveries about Nrf1 are indicated perpendicular to the timeline. Further details are provided in the text.

Structural domains of distinct Nrf1 isoforms and other CNC–bZIP factors

Figure 3
Structural domains of distinct Nrf1 isoforms and other CNC–bZIP factors

Within the NTD, the NHB1 (amino acids 11–30) signal sequence dictates Nrf1 to target to the ER and enables it to anchor within the membrane [40], whereas NHB2 (amino acids 81–106, along with a space region between NHB1 and NHB2) controls potential proteolytic processing of Nrf1 to generate N-terminally truncated isoforms [59]. NHB1 and/or NHB2 are highly conserved with other CNC–bZIP factors such as Nrf3, CncC and Skn-1, and thus they, together with Nrf1 and its isoforms (TCF11, Nrf1D and Nrf1ΔS), are grouped into an NHB1–CNC subfamily of membrane-bound transcription factors [37]. The Neh1L domain is composed of both CNC and bZIP regions [38,79], which enables heterodimerization with a sMaf or other bZIP proteins before DNA binding to NF-E2/AP1-like ARE sequences in the promoter regions of target genes. The TADs of Nrf1 comprise AD1, NST, AD2 and SR regions. Within AD1, Neh2L contains the DIDLID/DLG element and ETGE motif; both are conserved with equivalents of Nrf2 and CncC and act as two Keap1-binding sites to negatively regulate Nrf2 and CncC rather than Nrf1. In addition to Neh2L, the Neh5L, but not Neh4L, subdomain is included within AD1 of Nrf1, when compared with its longer TCF11. The N-linked glycosylation of the NST domain causes Nrf1 and TCF11 to migrate electrophoretically at estimated masses of ∼120 kDa and ∼140 kDa respectively, whereas their non-glycosylated/deglycosylated proteins exhibit fast electrophoretic mobilities at ∼95 kDa and ∼110 kDa. They may be processed further through selective proteolysis to give rise to multiple cleaved forms of between 85 kDa and 25 kDa, of which the putative activated 85-kDa protein could be also translated through an alternative transcript-encompassed initiation signal (in Nrf1 Δ767 clones designated originally [62,63]) so that it lacks an ER-anchoring NTD (hence called Nrf1ΔN), and may be processed further into an unstable polypeptide of 65 or 55 kDa. The 55-kDa form, originally designated LCR-F1 (also called Nrf1β), is produced by in-frame translation and/or selective proteolysis, and may be rapidly degraded to yield short dominant-negative isoforms of 46 (called Nrf1β2), 36 (Nrf1γ) and 25 (Nrf1δ) kDa. For convenience, Nrf1 variant ΔD and Δ10 clones [62,63] are renamed Nrf1D or Nrf1ΔS respectively. The Nrf1ΔS mutant lacks both SR and Neh6L domains by alternative splicing, whereas the Nrf1D mutant lacks an originally C-terminal portion of ZIP and Neh3L existing in the prototypic Nrf1 protein, which is replaced by an additional 80-amino-acid region containing a predicted hydrophobic TM domain. For detailed definitions of these domains and motifs, see Box 1 

Figure 3
Structural domains of distinct Nrf1 isoforms and other CNC–bZIP factors

Within the NTD, the NHB1 (amino acids 11–30) signal sequence dictates Nrf1 to target to the ER and enables it to anchor within the membrane [40], whereas NHB2 (amino acids 81–106, along with a space region between NHB1 and NHB2) controls potential proteolytic processing of Nrf1 to generate N-terminally truncated isoforms [59]. NHB1 and/or NHB2 are highly conserved with other CNC–bZIP factors such as Nrf3, CncC and Skn-1, and thus they, together with Nrf1 and its isoforms (TCF11, Nrf1D and Nrf1ΔS), are grouped into an NHB1–CNC subfamily of membrane-bound transcription factors [37]. The Neh1L domain is composed of both CNC and bZIP regions [38,79], which enables heterodimerization with a sMaf or other bZIP proteins before DNA binding to NF-E2/AP1-like ARE sequences in the promoter regions of target genes. The TADs of Nrf1 comprise AD1, NST, AD2 and SR regions. Within AD1, Neh2L contains the DIDLID/DLG element and ETGE motif; both are conserved with equivalents of Nrf2 and CncC and act as two Keap1-binding sites to negatively regulate Nrf2 and CncC rather than Nrf1. In addition to Neh2L, the Neh5L, but not Neh4L, subdomain is included within AD1 of Nrf1, when compared with its longer TCF11. The N-linked glycosylation of the NST domain causes Nrf1 and TCF11 to migrate electrophoretically at estimated masses of ∼120 kDa and ∼140 kDa respectively, whereas their non-glycosylated/deglycosylated proteins exhibit fast electrophoretic mobilities at ∼95 kDa and ∼110 kDa. They may be processed further through selective proteolysis to give rise to multiple cleaved forms of between 85 kDa and 25 kDa, of which the putative activated 85-kDa protein could be also translated through an alternative transcript-encompassed initiation signal (in Nrf1 Δ767 clones designated originally [62,63]) so that it lacks an ER-anchoring NTD (hence called Nrf1ΔN), and may be processed further into an unstable polypeptide of 65 or 55 kDa. The 55-kDa form, originally designated LCR-F1 (also called Nrf1β), is produced by in-frame translation and/or selective proteolysis, and may be rapidly degraded to yield short dominant-negative isoforms of 46 (called Nrf1β2), 36 (Nrf1γ) and 25 (Nrf1δ) kDa. For convenience, Nrf1 variant ΔD and Δ10 clones [62,63] are renamed Nrf1D or Nrf1ΔS respectively. The Nrf1ΔS mutant lacks both SR and Neh6L domains by alternative splicing, whereas the Nrf1D mutant lacks an originally C-terminal portion of ZIP and Neh3L existing in the prototypic Nrf1 protein, which is replaced by an additional 80-amino-acid region containing a predicted hydrophobic TM domain. For detailed definitions of these domains and motifs, see Box 1 

DISCOVERY OF Nrf1 AND ISOFORMS WITH DISTINCT POTENTIALS

In the study of the human β-globin gene regulation, its upstream LCR (locus control region) is identified to act as an erythroid-specific enhancer to monitor differential expression of the genes in the embryonic, fetal and adult haemopoietic tissues [4244]. The enhancer activity is mediated directly by a cis-acting AP1 element repeat (5′-GCTGAGTCATGATGAGTCA-3′, in which the core AP1-binding sites are underlined) [45]. The binucleotide GC is added 5′ to the left AP1-binding site to yield an antisense ARE sequence (5′-GCTGAGTCA-3′; Table 1), which was originally identified as a consensus NF-E2-binding site [4648]. The functional dimer of NF-E2 is composed of an erythroid-specific p45 subunit and a widely expressed p18 subunit. The p45 subunit enables NF-E2 to predominantly transactivate the β-globin gene through its DNA-binding CNC–bZIP domain [49]. The p18 subunit of NF-E2 was later identified as a sMaf (small Maf) protein (including MafG, MafF and MafK) [5052].

Three major founding isoforms: Nrf1, TCF11 and LCR-F1

The tandem consensus NF-E2/AP1-binding site from the β-globin LCR was used as an oligonucleotide probe to isolate Nrf1 from a cDNA expression library, which consists of either 742 amino acids in humans [53] or 741 amino acids in mice [54]. Similar strategies were also employed in two additional independent laboratories to clone cDNA sequences of LCR-F1 [55] and TCF11 [56], which comprise 447 and 772 amino acids respectively. Except for length variations, both nucleotide and amino acid sequences of LCR-F1 and TCF11 are identical with those of Nrf1, and thus they are thought of as different length isoforms (Figure 3). In fact, Nrf1 is yielded by alternative splicing of mRNA to remove exon 4 [that encodes V242PSGEDQTALSLEECLRLLEATCPFGENAE271, called Neh4L (Nrf2–ECH homology 4-like) region] from human TCF11 [56]. Although such Neh4L is lost in Nrf1, the latter isoform was shown to exhibit similar transactivation activity to that of TCF11 [57]; this long form is not found in mice [54]. Furthermore, it should be noted that selective post-translational processing of either Nrf1 or TCF11 yields multiple polypeptides of between 140 kDa and 25 kDa, which together dictate its overall activity to regulate differential expression of target genes [5860].

Nrf1β/LCR-F1 along with small dominant-negative isoforms Nrf1γ and Nrf1δ

By amino acid sequence comparison, LCR-F1 is a shorter form of Nrf1 (called Nrf1β [24,41]), which is thought to be translated by an in-frame perfect Kozak initiation signal (5′-puCCATGG-3′) existing around the methionine codons between positions 289 and 297 in mice [53,54,56]. Nrf1β/LCR-F1 migrates at a molecular mass of 55 kDa in the pH 7.0 LDS (lithium dodecyl sulfate)/NuPAGE gel [40,41], but the pH 8.9 Laemmli SDS/PAGE allows its mobility to be exhibited at an estimated size of 60 kDa [30] or 65 kDa (called p65Nrf1 [53,61]). Relative to Nrf1, Nrf1β/LCR-F1 lacks the N-terminal AD1 (acidic domain 1) [38,57] and exhibits a weak transactivation activity [55,60,62,63], but induction of Nrf1β/LCR-F1 activity may be dependent on distinct stressors in different cell lines [6264]. Intriguingly, an exception was reported that Nrf1β/LCR-F1 was thought to be a significant dominant-negative inhibitor of ARE-driven gene transactivation against the wild-type Nrf1 and Nrf2 [61]. The dispute on Nrf1β/LCR-F1 suggests that it is an unstable protein that might be proteolytically processed to yield two small polypeptides of 36 kDa and 25 kDa. This is supported by the finding that the transactivation activity of Nrf1β/LCR-F1 is significantly increased by blocking generation of either 36-kDa Nrf1γ or 25-kDa Nrf1δ [59,60]; both can be produced by the potential in-frame translation, as well as the selective endoproteolytic processing of longer Nrf1 proteins. Importantly, these small dominant-negative Nrf1γ and Nrf1δ, when overexpressed, are able to competitively interfere with the functional assembly of the active transcription factors, so as to down-regulate expression of NF-E2/AP1-like ARE-driven genes [60,62].

Box 1
Definitions of the major domains and motifs shown in bold
AD1 (acidic domain 1) functions as the major TAD (transactivation domain) in Nrf1, and comprises amino acids 125–298. This domain contains the PEST1 sequence (amino acids 141–170), the Neh2L subdomain (amino acids 156–242), the CPD (Cdc4 phosphodegron) (L267LSPLLT273) and the Neh5L subdomain (amino acids 280–298). 
AD2 (acidic domain 2) contributes to transactivation activity of Nrf1, and is particularly important for the short Nrf1β/LCR-F1 isoform. This domain includes an acidic-hydrophobic amphipathic region (amino acids 403–440) and SDS1 (serine/aspartate/serine motif 1) (amino acids 441–455) that contains the D447SGLS451 β-TrCP-binding degron. 
ARE (antioxidant-response element), also called the EpRE (electrophile-response element), comprises the consensus DNA sequence 5′-TGACnnnGC-3′ in the cognate gene promoter that is recognized by CNC–bZIP factors. 
CNC (cap‘n'collar) domain was originally identified in a Drosophila transcription factor. The CNC family includes C. elegansSkn-1, the four vertebrate activators NF-E2 p45 subunit, Nrf1 (including its long form TCF11 and short form Nrf1β/LCR-F1), Nrf2 and Nrf3, and two distantly related repressors Bach1 and Bach2. This family shares a highly conserved 45-amino-acid CNC domain (e.g. comprising amino acids 581–624 in Nrf1). 
CRAC (cholesterol-recognition/amino acid consensus motif) adjoins membrane-associated segments to enable interaction of the protein with membrane lipids. CRAC1 (amino acids 62–70) and CRAC2 (amino acids 74–82) are located close to the TM1 (amino acids 7–26) region within the NTD (N-terminal domain) of Nrf1. CRAC3 (V191XXYXXRXK199) lies immediately adjacent to the DIDLID/DLG element (amino acids 171–186, situated on the border between Neh2L and PEST1). 
Neh1L (Nrf2–ECH homology 1-like) region contains both CNC and bZIP domains and functions as the DBD (DNA-binding domain). 
Neh2L (Nrf2–ECH homology 2-like) subdomain is situated in the centre of AD1 in Nrf1. It is overlapped N-terminally by the PEST1 sequence and is flanked C-terminally by the CPD and Neh5L regions. Importantly, the Neh2L contains DLG and ETGE motifs, but these do not target Nrf1 for the Keap1-mediated proteasomal degradation. The DLG motif overlaps with the DIDLID element; both are integrated together and therefore referred to as the DIDLID/DLG element (amino acids 171–186). 
Neh3L (Nrf2–ECH homology 3-like) region, also called the C-terminal domain, includes a CRAC5 motif (amino acids 683–695) that lies adjacent to TMc (C-terminal transmembrane region, amino acids 705–725) and a putative arginine-enriched ER retention signal (amino acids 730–741). 
Neh4L (Nrf2–ECH homology 4-like) acts as a TAD in TCF11, but not in Nrf1. It is lost in Nrf1 by alternative splicing. 
Neh5L (Nrf2–ECH homology 5-like) subdomain functions as an essential TAD. It shares homology with the DIDLID/DLG element and an amphipathic region of the AD2 region (amino acids 409–428). 
Neh6L (Nrf2–ECH homology 6-like) domain is situated between the SR (serine-repeat) domain and the DBD, and contributes to the negative regulation of Nrf1. The N-terminal 30-amino-acid region of Neh6L that overlaps with the PEST2 sequence, contains a core SDS2 (serine/aspartate/serine motif 2) (amino acids 497–506) and also is adjacent to CRAC4/TMp (a proline-kinked hinge structure folded by amino acids 507–525). 
NHB1 (N-terminal homology box 1) comprises amino acids 11–30 in the NTD (amino acids 1–124) that negatively regulates Nrf1. The ER-targeting NHB1 sequence is highly conserved in equivalents in TCF11, Nrf3, CncC and Skn-1, and they are therefore proposed to be grouped together as the ‘NHB1–CNC’ subfamily of membrane-binding transcription factors. 
NHB2 (N-terminal homology box 2) comprises amino acids 81–106 in the NTD of Nrf1, which is conserved with equivalents in TCF11, Nrf3 and CncC, but does not exist in Skn-1. Although the amphipathic NHB2 sequence is identified to act as an ER-luminal anchor, if being repositioned on the cyto/nucleo-plasmic sides of membranes, it is predicted to function as a putative degron targeting to the ERAD pathway. This process is monitored possibly by a spacer region (such as CRAC1/2) between NHB1 and NHB2. 
NST (asparagine/serine/threonine) (amino acids 299–400) domain is situated between AD1 and AD2. It exists as a glycodomain in the ER, and has the capability to function as a bona fide TAD, which would be exerted only after it is repartitioned out of membranes into the cyto/nucleo-plasm. This process appears to be controlled by the TMi glycopeptide (amino acids 375–393) and additional CPD (amino acids 348–356). 
PEST (proline/glutamate/serine/threonine) sequence acts as a degron that targets the protein for either proteasome-dependent or -independent proteolysis pathway. Besides the PEST1 sequence (amino acids 141–170) in the N-terminal one-third of AD1, the PEST2 sequence (amino acids 456–519) covers the entire SR domain comprising amino acids 454–488 (as an inducible TAD), the SDS2 motif (amino acids 497–506), and the CRAC4/TMp core region (amino acids 508–519). 
AD1 (acidic domain 1) functions as the major TAD (transactivation domain) in Nrf1, and comprises amino acids 125–298. This domain contains the PEST1 sequence (amino acids 141–170), the Neh2L subdomain (amino acids 156–242), the CPD (Cdc4 phosphodegron) (L267LSPLLT273) and the Neh5L subdomain (amino acids 280–298). 
AD2 (acidic domain 2) contributes to transactivation activity of Nrf1, and is particularly important for the short Nrf1β/LCR-F1 isoform. This domain includes an acidic-hydrophobic amphipathic region (amino acids 403–440) and SDS1 (serine/aspartate/serine motif 1) (amino acids 441–455) that contains the D447SGLS451 β-TrCP-binding degron. 
ARE (antioxidant-response element), also called the EpRE (electrophile-response element), comprises the consensus DNA sequence 5′-TGACnnnGC-3′ in the cognate gene promoter that is recognized by CNC–bZIP factors. 
CNC (cap‘n'collar) domain was originally identified in a Drosophila transcription factor. The CNC family includes C. elegansSkn-1, the four vertebrate activators NF-E2 p45 subunit, Nrf1 (including its long form TCF11 and short form Nrf1β/LCR-F1), Nrf2 and Nrf3, and two distantly related repressors Bach1 and Bach2. This family shares a highly conserved 45-amino-acid CNC domain (e.g. comprising amino acids 581–624 in Nrf1). 
CRAC (cholesterol-recognition/amino acid consensus motif) adjoins membrane-associated segments to enable interaction of the protein with membrane lipids. CRAC1 (amino acids 62–70) and CRAC2 (amino acids 74–82) are located close to the TM1 (amino acids 7–26) region within the NTD (N-terminal domain) of Nrf1. CRAC3 (V191XXYXXRXK199) lies immediately adjacent to the DIDLID/DLG element (amino acids 171–186, situated on the border between Neh2L and PEST1). 
Neh1L (Nrf2–ECH homology 1-like) region contains both CNC and bZIP domains and functions as the DBD (DNA-binding domain). 
Neh2L (Nrf2–ECH homology 2-like) subdomain is situated in the centre of AD1 in Nrf1. It is overlapped N-terminally by the PEST1 sequence and is flanked C-terminally by the CPD and Neh5L regions. Importantly, the Neh2L contains DLG and ETGE motifs, but these do not target Nrf1 for the Keap1-mediated proteasomal degradation. The DLG motif overlaps with the DIDLID element; both are integrated together and therefore referred to as the DIDLID/DLG element (amino acids 171–186). 
Neh3L (Nrf2–ECH homology 3-like) region, also called the C-terminal domain, includes a CRAC5 motif (amino acids 683–695) that lies adjacent to TMc (C-terminal transmembrane region, amino acids 705–725) and a putative arginine-enriched ER retention signal (amino acids 730–741). 
Neh4L (Nrf2–ECH homology 4-like) acts as a TAD in TCF11, but not in Nrf1. It is lost in Nrf1 by alternative splicing. 
Neh5L (Nrf2–ECH homology 5-like) subdomain functions as an essential TAD. It shares homology with the DIDLID/DLG element and an amphipathic region of the AD2 region (amino acids 409–428). 
Neh6L (Nrf2–ECH homology 6-like) domain is situated between the SR (serine-repeat) domain and the DBD, and contributes to the negative regulation of Nrf1. The N-terminal 30-amino-acid region of Neh6L that overlaps with the PEST2 sequence, contains a core SDS2 (serine/aspartate/serine motif 2) (amino acids 497–506) and also is adjacent to CRAC4/TMp (a proline-kinked hinge structure folded by amino acids 507–525). 
NHB1 (N-terminal homology box 1) comprises amino acids 11–30 in the NTD (amino acids 1–124) that negatively regulates Nrf1. The ER-targeting NHB1 sequence is highly conserved in equivalents in TCF11, Nrf3, CncC and Skn-1, and they are therefore proposed to be grouped together as the ‘NHB1–CNC’ subfamily of membrane-binding transcription factors. 
NHB2 (N-terminal homology box 2) comprises amino acids 81–106 in the NTD of Nrf1, which is conserved with equivalents in TCF11, Nrf3 and CncC, but does not exist in Skn-1. Although the amphipathic NHB2 sequence is identified to act as an ER-luminal anchor, if being repositioned on the cyto/nucleo-plasmic sides of membranes, it is predicted to function as a putative degron targeting to the ERAD pathway. This process is monitored possibly by a spacer region (such as CRAC1/2) between NHB1 and NHB2. 
NST (asparagine/serine/threonine) (amino acids 299–400) domain is situated between AD1 and AD2. It exists as a glycodomain in the ER, and has the capability to function as a bona fide TAD, which would be exerted only after it is repartitioned out of membranes into the cyto/nucleo-plasm. This process appears to be controlled by the TMi glycopeptide (amino acids 375–393) and additional CPD (amino acids 348–356). 
PEST (proline/glutamate/serine/threonine) sequence acts as a degron that targets the protein for either proteasome-dependent or -independent proteolysis pathway. Besides the PEST1 sequence (amino acids 141–170) in the N-terminal one-third of AD1, the PEST2 sequence (amino acids 456–519) covers the entire SR domain comprising amino acids 454–488 (as an inducible TAD), the SDS2 motif (amino acids 497–506), and the CRAC4/TMp core region (amino acids 508–519). 

The minor (or tissue-specific) isoforms: Nrf1D, Nrf1ΔN and Nrf1ΔS

Alternative splicing of mRNA can remove a coding region for a peptide I175DILWRQDIDL185 from the N-terminal AD1 of Nrf1, which is conserved with an essential portion of the DIDLID element within Skn-1 [65]. It was also found that other variations in the 3′- and 5′-UTRs yielded four different types of mRNA transcripts, that are consequently translated into distinct Nrf1 isoforms with different trans-acting potentials [56]. As anticipated, several variant isoforms of Nrf1 were isolated from mouse mast CPII or dendritic DC18 cells by a PCR cloning strategy employing different primer pairs, including those covering the exon/intron boundaries delineated from the genomic sequence [62,63].

Of the Nrf1 variants, two were originally designated as Nrf1 clones Δ767 and D, containing a deletion of the functional translational start and stop codons respectively. Removal of the first translation start codon may yield a small form that was simply designated Nrf1ΔN in 2009 [24,58] (recently recalled Nrf1b [66]), in which the first N-terminal 181-amino-acid region of Nrf1 is replaced with a dodecylpeptide MGWESRLTAASA. By contrast, Nrf1 variant clone D not only has a deletion of the translation stop codon, but also contains a change in the second half of the leucine zipper motif. The C-terminal 72 amino acids of wild-type Nrf1 that are extremely positively charged are exchanged for an additional 80-amino-acid stretch that is enriched with negatively charged residues in this variant D containing a putative transmembrane segment (Figure 3). The third form was called Nrf1 clone Δ10 with a 137-amino-acid deletion covering both SR (serine-repeat) and serine-rich Neh6L (Nrf2–ECH homology 6-like) regions. For convenience, variant clones D and Δ10 are simply renamed Nrf1D or Nrf1ΔS respectively.

Together, at least 11 isoforms of Nrf1 [58], although differentially expressed in different mammalian species [22,24,62,63,67], are produced from several variant transcripts of the single nfe2l1 gene (e.g. in humans and mice; Figures 4B and 4C) [56,68,69]. An exception is two copies of the zebrafish nfe2l1 gene (i.e. nrf1a and nrf1b), along with two copies of the nfe2l2 gene (i.e. nrf2a and nrf2b) [70].

Distinct genome-targeting strategies to knock out the Nrf1 gene, with its products of multiple transcripts and proteins that are differentially expressed in mice and humans

Figure 4
Distinct genome-targeting strategies to knock out the Nrf1 gene, with its products of multiple transcripts and proteins that are differentially expressed in mice and humans

(A) Schematic representation of the wild-type Nrf1 locus (1) and distinct genome-targeting strategies (2–6) to knock out the gene. Of note, 2–4 represent three different strategies to make global KO of Nrf1 in mice, whereas 5 and 6 indicate two targeting approaches to create tissue-specific conditional KO of Nrf1 in the indicated organs. Various lengths of the Nrf1 gene loci and its artificial mutants are represented by double-stranded lines, with gene transcriptional expression being directed by arrows. Untranslated and translated exons are represented by light and dark green boxes respectively. Distinct portions of the gene that had been manipulated are also indicated. (B and C) Diagrammatic representation of chromosomal locations of the Nfe2l1 gene loci (expressed as Nrf1, TCF11 and/or LCR-F1) in humans and/or mice, with different numbers of exons. The left-hand side shows different lengths of multiple transcripts with altered numbers of the exons indicated, which were predicted to translate different protein isoforms shown on the right-hand side. Of note, exon 2a is generally considered to be untranslated, but indeed is predicted to contain a uORF, exons 3–5 located within the mORF can also be allowed for no, partial or complete translation so as to give rise to various lengths of distinct protein forms. The production of multiple isofoms is also attributable to alternative translation from mRNA variants arising from different transcription start sites (TSSs, so as to yield, for example, Nrf1ΔN and Nrf1β), alternative splicing of longer mRNA transcripts (e.g. to remove exon 4), and putative regulation of the long 3′-UTR containing two poly(A) tail signals.

Figure 4
Distinct genome-targeting strategies to knock out the Nrf1 gene, with its products of multiple transcripts and proteins that are differentially expressed in mice and humans

(A) Schematic representation of the wild-type Nrf1 locus (1) and distinct genome-targeting strategies (2–6) to knock out the gene. Of note, 2–4 represent three different strategies to make global KO of Nrf1 in mice, whereas 5 and 6 indicate two targeting approaches to create tissue-specific conditional KO of Nrf1 in the indicated organs. Various lengths of the Nrf1 gene loci and its artificial mutants are represented by double-stranded lines, with gene transcriptional expression being directed by arrows. Untranslated and translated exons are represented by light and dark green boxes respectively. Distinct portions of the gene that had been manipulated are also indicated. (B and C) Diagrammatic representation of chromosomal locations of the Nfe2l1 gene loci (expressed as Nrf1, TCF11 and/or LCR-F1) in humans and/or mice, with different numbers of exons. The left-hand side shows different lengths of multiple transcripts with altered numbers of the exons indicated, which were predicted to translate different protein isoforms shown on the right-hand side. Of note, exon 2a is generally considered to be untranslated, but indeed is predicted to contain a uORF, exons 3–5 located within the mORF can also be allowed for no, partial or complete translation so as to give rise to various lengths of distinct protein forms. The production of multiple isofoms is also attributable to alternative translation from mRNA variants arising from different transcription start sites (TSSs, so as to yield, for example, Nrf1ΔN and Nrf1β), alternative splicing of longer mRNA transcripts (e.g. to remove exon 4), and putative regulation of the long 3′-UTR containing two poly(A) tail signals.

Box 2
Notable differences in the salient features between Nrf1 and Nrf2
Nrf1 is more conservative than Nrf2 in the process of biological evolution: Nrf1 is highly conserved with the ancestral orthologues CNC and Skn-1 because they, but not Nrf2 (Figure 3), belong to the membrane-binding NHB1–CNC subfamily of transcription factors that are essential for both normal development and physiological homoeostasis. 
Nrf1, but not Nrf2, is indispensable for normal development and growth because its functional loss results in significant pathological phenotypes: Global KO of Nrf1, but not Nrf2, leads to embryonic lethality in the uterus. Conditional or inducible deletion of Nrf1 results in significant pathological phenotypes in tissue-specific KO mice (Table 2). By contrast, no obvious phenotypes are found in Nrf2-KO mice, although it is required for adaptation against cellular stress. 
Distinct roles for Nrf1 and Nrf2 in regulating distinct subsets of ARE-battery genes are conferred on Nrf1 with the unique functionality to maintain cellular homoeostasis and organ integrity: Both factors enable differential regulation of distinct AP1-like ARE/EpRE-driven genes: the basal constitutive expression of genes (critical for antioxidant, detoxification, cytoprotection, homoeostasis and development) appears to be controlled predominantly by Nrf1, whereas Nrf2 is a master regulator of inducible gene expression in response to cellular stress, although it is also involved in regulating basal expression of some genes. 
Nrf1 is essential for maintaining normal homoeostatic metabolism (i.e. proteostasis, lipidostasis and glucostasis), as well as the Nrf2-inducible response to metabolic or nutrient stress: Nrf1 is required for regulating key genes controlling normal homoeostatic metabolisms of protein, lipid and glucose to certain steady-state levels, and thus the functional loss leading to metabolic disorders such as diabetes, NASH and neurodegenerative diseases. In addition, some genes are also regulated by Nrf2 principally in response to metabolic or nutrient stress. 
Differences in the subcellular locations between Nrf1 and Nrf2 to mediate distinct biological responses: Under normal homoeostatic conditions, Nrf2 is sequestered by Keap1 in the cytoplasm where it is targeted for Keap1-mediated ubiquitination degradation, whereas cellular stress enables the release of Nrf2 to translocate the nucleus before transactivating target genes. By contrast, Nrf1 is not regulated by Keap1, because it is targeted and anchored in the ER and then sorted to import the nuclear envelope. If Nrf1 is integrated in the INM, it could gain a direct access to cognate genes; once it is stimulated by biological cues, the protein is subject to its dynamic topovectorial repositioning and selective processing in the ER and extra-ER compartments so as to yield an active or negative isoforms before translocating the nucleus (as illustrated in Figure 6). 
Distinct isoforms of Nrf1, but not of Nrf2, exhibit different or opposing functional activities: There exist at least 11 isoforms of Nrf1, though being differentially expressed in different metazoan species, arising from the single Nfe2l1 gene. Distinct Nrf1 isoforms enable formation of different functional heterodimers with each of sMaf or other bZIP proteins to differentially bind AP1-like AREs or its homologues (Table 1) and thus exert different or even opposing activities so as to fine-tune target genes. Such resulting differences are featured in Figures 3, 4 and 6. In contrast, no other isoforms are found in Nrf2. 
Differences in the structural and functional domains between Nrf1 and Nrf2: Both the ER-targeting NTD and NST glycodomain are present in Nrf1, but are absent from Nrf2; the structural difference dictates their biological functional distinctions. Furthermore, differential functionalities are also embodied within Neh1L–Neh6L (Box 1), which are represented by equivalents (i.e. Neh1–Neh6) in Nrf2:
  •  (i) The basic Neh1L region is positioned on the cyto/nucleo-plasmic side of membranes gaining access to genes.

  •  (ii) Neh2L does not serve as a putative degron despite containing Keap1-binding DLG and ETGE sites.

  •  (iii) Neh3L negatively regulates Nrf1, in contrast with positive regulation of Nrf2 by Neh3.

  •  (iv) Neh4L is dispensable in Nrf1, despite positive contribution to transactivation by its long form TCF11.

  •  (v) Neh5L, along with the DIDLID/DLG element and an amphipathic AD2 region, positively regulates Nrf1 via possible mechanisms accounting for Nrf2, and contributes to topovectorial repartitioning of Nrf1 out of ER.

  •  (vi) By comparison with the Neh6 domain of Nrf2, a similar Neh6L region that negatively regulates Nrf1 appears to be split by its 35-SR domain, which is O-GlcNAcylated so as to promote PEST2-directed proteolysis. Notably, the topovectorial repositioning of TADs (including AD1, NST, AD2 and SR) from the luminal side across membranes into cyto/nucleo-plasmic compartments is also monitored by TMp (situated on the border between PEST2 and Neh6L), together with the TMi (sited on the C-terminus of the NST domain adjacent to AD2).

  •  (vii) The amphipathic region of AD2 is considerably conserved with the Neh7 domain of Nrf2, but it is unknown whether AD2 of Nrf1 interacts with RXRα as Neh7 does directly.

 
Differences in the membrane topobiology between Nrf1 and Nrf2: The membrane topology of Nrf1 is determined by TM1 (amino acids 7–26), in co-operation with amphipathic TMi, TMp and TMc, and thus the proper topology of Nrf1 enables it to be integrated within and around the ER and nuclear envelope membranes, where it is selectively processed through various post-translational mechanisms into distinct isoforms with different topovectorial functionalities (as shown in Figure 6). Such topobiological features of Nrf1 dictates the unique functioning of the CNC–bZIP factor that is distinct from that of Nrf2. However, whether Nrf1 is involved in regulating genes critical for the topology of the ER and nucleus remains to be elucidated. 
Nrf1 is more conservative than Nrf2 in the process of biological evolution: Nrf1 is highly conserved with the ancestral orthologues CNC and Skn-1 because they, but not Nrf2 (Figure 3), belong to the membrane-binding NHB1–CNC subfamily of transcription factors that are essential for both normal development and physiological homoeostasis. 
Nrf1, but not Nrf2, is indispensable for normal development and growth because its functional loss results in significant pathological phenotypes: Global KO of Nrf1, but not Nrf2, leads to embryonic lethality in the uterus. Conditional or inducible deletion of Nrf1 results in significant pathological phenotypes in tissue-specific KO mice (Table 2). By contrast, no obvious phenotypes are found in Nrf2-KO mice, although it is required for adaptation against cellular stress. 
Distinct roles for Nrf1 and Nrf2 in regulating distinct subsets of ARE-battery genes are conferred on Nrf1 with the unique functionality to maintain cellular homoeostasis and organ integrity: Both factors enable differential regulation of distinct AP1-like ARE/EpRE-driven genes: the basal constitutive expression of genes (critical for antioxidant, detoxification, cytoprotection, homoeostasis and development) appears to be controlled predominantly by Nrf1, whereas Nrf2 is a master regulator of inducible gene expression in response to cellular stress, although it is also involved in regulating basal expression of some genes. 
Nrf1 is essential for maintaining normal homoeostatic metabolism (i.e. proteostasis, lipidostasis and glucostasis), as well as the Nrf2-inducible response to metabolic or nutrient stress: Nrf1 is required for regulating key genes controlling normal homoeostatic metabolisms of protein, lipid and glucose to certain steady-state levels, and thus the functional loss leading to metabolic disorders such as diabetes, NASH and neurodegenerative diseases. In addition, some genes are also regulated by Nrf2 principally in response to metabolic or nutrient stress. 
Differences in the subcellular locations between Nrf1 and Nrf2 to mediate distinct biological responses: Under normal homoeostatic conditions, Nrf2 is sequestered by Keap1 in the cytoplasm where it is targeted for Keap1-mediated ubiquitination degradation, whereas cellular stress enables the release of Nrf2 to translocate the nucleus before transactivating target genes. By contrast, Nrf1 is not regulated by Keap1, because it is targeted and anchored in the ER and then sorted to import the nuclear envelope. If Nrf1 is integrated in the INM, it could gain a direct access to cognate genes; once it is stimulated by biological cues, the protein is subject to its dynamic topovectorial repositioning and selective processing in the ER and extra-ER compartments so as to yield an active or negative isoforms before translocating the nucleus (as illustrated in Figure 6). 
Distinct isoforms of Nrf1, but not of Nrf2, exhibit different or opposing functional activities: There exist at least 11 isoforms of Nrf1, though being differentially expressed in different metazoan species, arising from the single Nfe2l1 gene. Distinct Nrf1 isoforms enable formation of different functional heterodimers with each of sMaf or other bZIP proteins to differentially bind AP1-like AREs or its homologues (Table 1) and thus exert different or even opposing activities so as to fine-tune target genes. Such resulting differences are featured in Figures 3, 4 and 6. In contrast, no other isoforms are found in Nrf2. 
Differences in the structural and functional domains between Nrf1 and Nrf2: Both the ER-targeting NTD and NST glycodomain are present in Nrf1, but are absent from Nrf2; the structural difference dictates their biological functional distinctions. Furthermore, differential functionalities are also embodied within Neh1L–Neh6L (Box 1), which are represented by equivalents (i.e. Neh1–Neh6) in Nrf2:
  •  (i) The basic Neh1L region is positioned on the cyto/nucleo-plasmic side of membranes gaining access to genes.

  •  (ii) Neh2L does not serve as a putative degron despite containing Keap1-binding DLG and ETGE sites.

  •  (iii) Neh3L negatively regulates Nrf1, in contrast with positive regulation of Nrf2 by Neh3.

  •  (iv) Neh4L is dispensable in Nrf1, despite positive contribution to transactivation by its long form TCF11.

  •  (v) Neh5L, along with the DIDLID/DLG element and an amphipathic AD2 region, positively regulates Nrf1 via possible mechanisms accounting for Nrf2, and contributes to topovectorial repartitioning of Nrf1 out of ER.

  •  (vi) By comparison with the Neh6 domain of Nrf2, a similar Neh6L region that negatively regulates Nrf1 appears to be split by its 35-SR domain, which is O-GlcNAcylated so as to promote PEST2-directed proteolysis. Notably, the topovectorial repositioning of TADs (including AD1, NST, AD2 and SR) from the luminal side across membranes into cyto/nucleo-plasmic compartments is also monitored by TMp (situated on the border between PEST2 and Neh6L), together with the TMi (sited on the C-terminus of the NST domain adjacent to AD2).

  •  (vii) The amphipathic region of AD2 is considerably conserved with the Neh7 domain of Nrf2, but it is unknown whether AD2 of Nrf1 interacts with RXRα as Neh7 does directly.

 
Differences in the membrane topobiology between Nrf1 and Nrf2: The membrane topology of Nrf1 is determined by TM1 (amino acids 7–26), in co-operation with amphipathic TMi, TMp and TMc, and thus the proper topology of Nrf1 enables it to be integrated within and around the ER and nuclear envelope membranes, where it is selectively processed through various post-translational mechanisms into distinct isoforms with different topovectorial functionalities (as shown in Figure 6). Such topobiological features of Nrf1 dictates the unique functioning of the CNC–bZIP factor that is distinct from that of Nrf2. However, whether Nrf1 is involved in regulating genes critical for the topology of the ER and nucleus remains to be elucidated. 

Four confusing abbreviations of other gene products that differ from Nrf1 referred to in the present review

In an attempt to provide a clear explanation of the nomenclature of the CNC–bZIP factor, we also note four additional confusing abbreviations similar to Nrf1, but definitely different from it, in the literature: (i) nuclear respiratory factor 1 (NRF-1, along with NRF-2), an activator of transcriptional expression of mitochondrial respiratory chain genes through its DBD (DNA-binding domain) that is closely related to the Drosophila EWG (erect wing) protein [71]; (ii) not really finished (Nrf), encoding a transcription factor for development of the zebrafish outer retina, that is homologous with human nuclear respiratory factor-1 and avian initiation-binding repressor [72]; (iii) nitrogen-response factor 1 (NRF1), a major regulator of the avirulence gene Avr9 through its zine-finger DBD, that is highly conserved in other nitrogen-regulatory proteins (i.e. AREA and NIT2) [73]; and (iv) negative regulator of Cdc forty-two (nrf1 designated in yeast), encoding a membrane protein Nrf1p, that inhibits the Cdc42p GTPase in Saccharomyces cerevisiae and Schizosaccharomyces pombe [74].

STRUCTURAL DOMAINS OF Nrf1 WITH CONSERVED MOTIFS DICTATE DISTINCT FUNCTIONS

Using bioinformatic analysis [38,75], Nrf1 is identified as a modular protein with nine discrete structural domains, which were designated NTD (N-terminal domain), AD1, NST (asparagine/serine/threonine), AD2, SR, Neh6L, CNC, bZIP and Neh3L (Nrf2–ECH homology 3-like) (each of which with distinct functional motifs is defined in Box 1, and Figures 3 and 5). Accumulated experimental evidence has revealed that these domains dictate the selective post-synthetic processing of Nrf1 to yield multiple polypeptide isoforms with distinctive and even opposing functions in regulating target gene expression [37,40,41,5860,76].

Two founding domains, CNC and bZIP (collectively called Neh1L), enable Nrf1 to bind target genes

The high homology among the CNC–bZIP family members is largely restricted to the CNC and bZIP regions, which are involved in their DNA binding to target genes and functional heterodimerization with a sMaf protein [49,77,78]. These two regions in Nrf1 were merged together as Neh1L (Nrf2–ECH homology 1-like) domain, as described in Nrf2 [79], and share approximately 40% amino acid identity with equivalent domains in NF-E2 p45, Nrf2 and Nrf3. An oligonucleotide library selection revealed that Neh1L enables Nrf1/TCF11 to bind an NF-E2/AP1-like consensus site (5′-TGCTgaGTCAT-3′ in which the binucleotide ‘ga’ is not essential; Table 1) originally found in the LCR of β-globin and the PBGD (porphobilinogen deaminase) promoter [78,80]. Similar DNA-binding activity of Nrf1 to the ARE (with the sequence 5′-TGAC/GnnnGC-3′) and ARE-like sequences was identified in the promoter regions of human and mouse Nqo1 [81,82]. Structural analysis of the functional heterodimers revealed that the CNC–bZIP proteins bind to the TGAC core of the ARE sequence, whereas sMaf partners specifically recognize the flanking GC regions [83,84]; their alterations render distinct heterodimers to mediate different sets of target genes [85,86].

Within the Neh1L domain, the basic region (R625DIRRRGKNKMAAQNCRKRKL645) also functions as a classic bipartite NLS (nuclear localization signal) [60], enabling nuclear translocation of Nrf1 to gain access to its target genes before transactivating their expression. In turn, it seems plausible that sets of NF-E2/AP1-like ARE-battery genes could also be differentially activated or repressed by distinct Nrf1 isoforms depending on different heterodimers with each of the AP1 family members [e.g. c-Jun, JunD, c-Fos, FosB, Fra-1, Fra-2 and ATF2 (activating transcription factor 2)] [62,63,81,87,88], ATF4 [89,90], and C/EBPβ (CCAAT/enhancer-binding protein β) [91], as well as sMaf [80,89,92]. In addition, the Neh1L domain of Nrf1 can also enable it to be associated with other DNA-binding complexes of NFAT (nuclear factor of activated T-cells) [62,63], NF-κB (p50 and p65) [88] or androgen receptor [64], such that their cognate gene transcription may also be regulated by the CNC–bZIP factor.

The amphipathic Neh3L region within the CTD (C-terminal domain) negatively regulates Nrf1

The Neh3L region immediately C-terminal to the CNC and bZIP regions is highly conserved among the family members (Figure 3). The homologous Neh3 domain of Nrf2 was identified to be required for its transactivation activity through an interaction with CHD6 (chromo-ATPase/helicase DNA-binding protein 6) [93]. However, CHD6 was not detected as one of the Nrf1-interacting proteins by LC–MS/MS [94], although the Neh3L region of Nrf1 shares 50% sequence identity and 70% similarity with the Neh3 domain of Nrf2. Conversely, the activity of shorter Nrf1β/LCR-F1 was repressed through direct interaction of its Neh3L with the human cytomegalovirus IE2 (immediate-early protein 2) [95] or MCRS2 (microspherule protein 2), a cell-cycle-dependent regulator that is involved in the telomere shortening by inhibiting telomerase activity [96], but both its heterodimerization with a sMaf protein and its ability to bind ARE sequences in Nrf1 target genes are unaffected by these two interacting partners.

Further examination reveals that the negative regulation of Nrf1 by the CTD is associated with the topological folding of the core amphipathic portion (Y707ALQYAGDGSVLLIPRTM724, called TMc or transmembrane C-terminal region) of Neh3L within and around the ER; the dynamic association with membranes could also be enhanced by either CRAC5 (cholesterol-recognition/amino acid consensus motif 5) or the basic C-tail (R730RQERKPKDRRK741) through potential interaction with the cholesterol-rich microdomains (i.e. rafts and caveolae) [41,58,60]. In addition, the basic C-tail peptide is not required for the nuclear localization of Nrf1, but rather for its primary cytoplasmic location. Yet, whether the arginine-enriched C-tail acts as a putative ER/NE-retention signal remains to be determined.

The Neh2L region does not negatively regulate Nrf1 that differs from the Neh2 domain of Nrf2

Within AD1 of Nrf1, the Neh2L (Nrf2–ECH homology 2-like) region is represented by the Neh2 domain of Nrf2, and is also conserved with the equivalent of CncC (Figure 3). The N-terminal Neh2 domain of Nrf2 contains a redox-sensitive Keap1 (Kelch-like ECH-associated protein 1)-binding degron (that comprises both the DLG and ETGE motifs), which targets the normal homoeostatic Nrf2 protein to the ubiquitin ligase cullin-3/Rbx1-dependent proteasomal degradation pathway [9799]. Subsequently, a two-site substrate recognition model was proposed that the forked-stem homodimer of Keap1 binds to the DLG and ETGE motifs within the Neh2 domain [100103]. A mechanism similar to the Keap1/Nrf2 signalling pathway has also been shown to be involved in the regulation of the Drosophila CncC protein [104,105], but not of the closely related Nrf1 [34,38].

Although the putative physical interaction between Nrf1 and Keap1 was examined in total cell lysates by co-immunoprecipitation [39] and LC–MS/MS [94], deletion of either the DLG and ETGE motifs or the entire Neh2L region influenced neither the stability of Nrf1 nor its transactivation activity [34,38,41,106]. In turn, the protein stability of Nrf1 and its functional activity were not affected by either KD (knockdown) of Keap1 [38] or expression of a dominant-negative mutant of cullin-3 [107]. Moreover, depletion of Keap1 by siRNA neither increased the steady-state level of Nrf1/TCF11 nor decelerated its protein degradation [34]. These results clearly demonstrate that the Neh2L of Nrf1 does not enable it to be negatively regulated via a mechanism similar to the Keap1SCF-dependent ubiquitin–proteasomal degradation pathway accounting for Nrf2. Conversely, Neh2L contributes to the stability of Nrf1, in particular its 120-kDa glycoprotein [59]. Further topology study reveals that, during biosynthesis of Nrf1, its Neh2L within AD1 along with the NST glycodomain is co-translationally translocated into the ER lumen [37]. Thus, while the Neh2L region is buried in the ER lumen, it cannot be allowed for the putative binding sites to gain access to the cytosolic Keap1. Exclusively, just after the Neh2L of Nrf1 could enter the extra-ER compartments, it might act as a bona fide degron, if only the DLG-flanking lysine/glutamate-rich sites (i.e. K169EX28K199EX4K205E; Figure 5) would be targeted for ubiquitination by an E3 ligase in the vicinity of the ER.

Locations of distinct functional motifs within Nrf1

Figure 5
Locations of distinct functional motifs within Nrf1

Within NTD, the TM1-associated NHB1 signal anchor sequence dictates Nrf1 to be integrated within and around the ER membrane. Of note, the Ncytoplasm/Clumen orientation of TM1 (amino acids 7–26) is determined by charge difference between its N-terminal region and C-terminal region flanking the core hydrophobic h-region, and thus may be regulated by potential phosphorylation, ubiquitination and lipidation of TM1-adjoining peptides. The space region (amino acids 31–81) between NHB1 and NHB2 was identified as a possible ubiquitin ligase Hrd1-binding site [106]. Both PEST1 and PEST2 degrons may also target Nrf1 to ubiquitin–proteasomal degradation pathways, and the selective processing of the protein by PEST1 and PEST2 may be regulated by possible modifications of DIDLID/DLG and SR-adjoining peptides (e.g. DSGLS), as they allow for dynamic repositioning from the ER luminal side of the membrane into the cyto/nucleo-plasmic side. On the C-terminal border of the NST glycodomain immediately to the left of the acidic-hydrophobic amphipathic AD2, the TMi segment (amino acids 375–393) may act as a lumen-anchoring switch to control repartitioning of its flanking AD1 and AD2/SR domains. Therefore only after Neh2L within AD1 is dislocated into the cytoplasmic compartments, could it enable Nrf1 to bind Keap1. The topovectorial process also directs regulation of two Cdc4 phosphodegrons (CPDs) by GSK-3β (and/or MAPKs) and its effect on Nrf1. In addition, both the selective processing of Nrf1 and its functional activity may also be monitored by its physical interaction of the indicated motifs with cognate binding proteins. For detailed definitions of these domains and motifs, see Box 1 

Figure 5
Locations of distinct functional motifs within Nrf1

Within NTD, the TM1-associated NHB1 signal anchor sequence dictates Nrf1 to be integrated within and around the ER membrane. Of note, the Ncytoplasm/Clumen orientation of TM1 (amino acids 7–26) is determined by charge difference between its N-terminal region and C-terminal region flanking the core hydrophobic h-region, and thus may be regulated by potential phosphorylation, ubiquitination and lipidation of TM1-adjoining peptides. The space region (amino acids 31–81) between NHB1 and NHB2 was identified as a possible ubiquitin ligase Hrd1-binding site [106]. Both PEST1 and PEST2 degrons may also target Nrf1 to ubiquitin–proteasomal degradation pathways, and the selective processing of the protein by PEST1 and PEST2 may be regulated by possible modifications of DIDLID/DLG and SR-adjoining peptides (e.g. DSGLS), as they allow for dynamic repositioning from the ER luminal side of the membrane into the cyto/nucleo-plasmic side. On the C-terminal border of the NST glycodomain immediately to the left of the acidic-hydrophobic amphipathic AD2, the TMi segment (amino acids 375–393) may act as a lumen-anchoring switch to control repartitioning of its flanking AD1 and AD2/SR domains. Therefore only after Neh2L within AD1 is dislocated into the cytoplasmic compartments, could it enable Nrf1 to bind Keap1. The topovectorial process also directs regulation of two Cdc4 phosphodegrons (CPDs) by GSK-3β (and/or MAPKs) and its effect on Nrf1. In addition, both the selective processing of Nrf1 and its functional activity may also be monitored by its physical interaction of the indicated motifs with cognate binding proteins. For detailed definitions of these domains and motifs, see Box 1 

Intriguingly, another such potential, but unclear, relationship seems to exist between Keap1, Nrf1/TCF11 and Nrf2; this is suggested by the finding that silencing of Keap1 augmented both basal and arsenic (As3+)-inducible levels of Nrf2 as expected, whereas the As3+-induced accumulation of Nrf1 was not increased, but was conversely attenuated, by KD of Keap1, as accompanied by no changes in the basal steady-state level of Nrf1 [108,109]. Therefore further study is warranted to determine what effects of the DIDLID/DLG element (D171IDLIDILWRQDIDLG186, in which the underlined DIDLID element overlaps the DLG motif) and ETGE motif are accurately elicited on Nrf1 and/or TCF11. In Skn-1, the conserved DIDLID element was reported to be a transactivation potential for this C. elegans transcription factor [65]. Nevertheless, a similar role for the DIDLID/DLG element has not been determined in either Nrf1 or Nrf2 [41,97,110]. Also, it is unknown whether PKC (protein kinase C) phosphorylates Ser195 within the central CRAC3 (V191FDYSHRQK199) motif immediately C-terminal to the DIDLID/DLG element in Nrf1, although Ser195 is located at a position equivalent to Ser40 in Nrf2 that is phosphorylated by PKC [111].

In addition, the first PEST (Pro/Glu/Ser/Thr-rich) sequence immediately N-terminal to the DIDLID/DLG element was predicted to be a potential degron to regulate the turnover of Nrf1 and TCF11 [67]. Yet, further examination indicates that the PEST1-adjoining region close to the N-terminal boundary of AD1 contributes to the stability of the 120-kDa glycoprotein of Nrf1, rather than its 95-kDa and smaller proteins [59], and also appears to positively regulate the factor at transactivating target genes [41]. Collectively, dual-opposing effects for the PEST1-adjoining region on AD1-mediated transactivation by Nrf1 may depend on the extent to which it is topologically repartitioned from the ER luminal side across membranes into the cyto/nucleo-plasmic side [37]. Thus only after it reaches the subcellular compartments, might it function as a potential degron or others.

The Neh5L subdomain is essential for the positive regulation of Nrf1 by AD1 lacking the Neh4L region

In Nrf1/TCF11, two regions, Neh4L and Neh5L (Nrf2–ECH homology 5-like), on the C-terminal boundary of AD1 are represented by the Neh4 and Neh5 domains of Nrf2, of which the Neh5L region also shares significant similarity with equivalents of all other CNC–bZIP proteins (Figures 3 and 5). In Nrf2, the central Neh4 and Neh5 regions function as two TADs (transactivation domains) that contribute to positive regulation of the CNC–bZIP factor by recruiting CBP {CREB [CRE (cAMP-response element)-binding protein]-binding protein} to the target gene transcriptional machinery [112,113]. The transactivation activity of Nrf2 is also controlled through a tight interaction with nuclear RAC3 (receptor-associated co-activator 3), also called SRC-3 (steroid receptor co-activator-3) [114]. By contrast, Nrf1 lacks the Neh4L region, which is removed by alternative splicing of the mRNA transcript encoding its longer form TCF11 [53,56]. Nevertheless, the transactivation activity of Nrf1 appeared to be unaffected by its loss of Neh4L function, because it still exhibits an activity similar to that measured from TCF11 [57]. Hence it is postulated that Neh4L-mediated transactivation by TCF11 could be negatively regulated via a mechanism involving the potential nuclear export signal (L251SLEECLRLL260 within Neh4L) [90]. This signal enables TCF11 rather than Nrf1 to export from the nucleus to the cytoplasm through a pathway involving Crm1 (chromosome region maintenance 1), an export receptor that forms a functional trimeric complex with cargos in the presence of Ran-GTP [90].

By contrast, it is important that the Neh5L region (with a core sequence D280LEQQWQDLMSIMEMQ295) is essential for the positive regulation of Nrf1 and TCF11 to transactivate target genes [37,41,57,113]. However, whether or how Neh5L-mediated transactivation by Nrf1/TCF11 is down-regulated by a CPD (Cdc4 phosphodegron)-containing region (S260PXXXXXLLSPLLTXXXSP278, the putative degron is underlined between the potential phosphorylation sites (italic)) immediately N-terminal to the Neh5L region is not well elucidated [107]. This region is inferred to enable phosphorylation by site-specific kinases such as GSK-3 (glycogen synthase kinase 3), MAPK (mitogen-activated protein kinase) and CDK (cyclin-dependent kinase).

AD2, NST and SR domains, as well as AD1, positively regulate Nrf1 and Nrf1β/LCR-F1, but both are negatively regulated by the Neh6L domain

Although AD1 acts as an essential TAD of Nrf1, its full transactivation activity is contributed by its AD2, NST and/or SR domains together with AD1 [24,41,60]. Relative to Nrf1, shorter Nrf1β/LCR-F1 lacks AD1, but still acts as a weak activator; the transactivation activity is mediated by its AD2, NST and SR domains [55,60,6264]. Of these TAD regions, AD2 is essential for transactivation by Nrf1β/LCR-F1, whereas both NST and SR domains appears to be dispensable for either Nrf1 or Nrf1β [24,41,55,60]. Further study of Nrf1 topobiology has unravelled that the major acidic-hydrophobic amphipathic portion (L409GGLLDEAMLDEISLMDLAI428) of AD2 shares structural homology with both the DIDLID/DLG element and the core Neh5L subdomain within AD1; they are suggested to be involved in the dynamic membrane-topological organization of Nrf1 to control its protein stability and transactivation activity [37,59]. Moreover, it should be noted that the amphipathic portion of AD2 in Nrf1 is also conserved with equivalent AD2L (AD2-like) regions of other CNC–bZIP proteins (Figure 3), of which the AD2L-adjoining region in Nrf2 has been identified as the Neh7 domain, that inhibit its activity through directly binding to RXRα (retinoic X receptor α) [115].

The NST region is a glycodomain in Nrf1, but its functioning as a bona fide TAD is elicited only after it has been repartitioned and retrotranslocated from the ER lumen into the cyto/nucleo-plasm where it is deglycosylated before transactivating target genes. Within the NST glycodomain, the core phenylalanine/leucine-rich semihydrophobic region (L375NSTFGSTNLAGLFFPSQL393, called TMi or transmembrane intermediate region in which the FG and GLFF motifs (italic) could also enable the protein to facilitate the passage through the central channel of nuclear pore complex) may serve as a membrane-tethered determinant to topologically lie on the plane of the luminal leaflet of membrane lipid bilayer, and anchor the adjoining amphipathic portion of AD2 closely to the luminal interface of membranes. In this case, once the TMi glycopeptide is deglycosylated, its adjacent regions (i.e. AD1, AD2 and SR) should be liberated from the luminal confinement, and repartitioned across membranes into cyto/nucleo-plasmic compartments, enabling transactivation of Nrf1 target genes. This is also supported by the finding that the basal Nrf1 activity and/or its stimulation by glucose deprivation were, to various degrees, prevented by deletion of the entire AD2, SR domain or their major portions [37,59]. In general, the ER lumen-resident L348FSPEVESL356 (the core degron underlined) motif within NST domain is not allowed for functioning as a bona fide degron, unless it would be repositioned out of the ER into the cyto/nucleo-plasmic subcellular compartments, whereupon it is enabled to become a GSK-3β-mediated phosphodegron, which targets Nrf1 (or Nrf1β/LCR-F1) to the Fbw7SCF (F-box and WD repeat domain-containing 7)-dependent ubiquitin–proteasomal degradation pathway, inhibiting its activity [107,116].

Spanning the boundary between the AD2 of Nrf1 and its SR region is the SDS1 sequence (E442EFDSDSGLSLDSS455, the core degron underlined) that is similar to the peptide E329FNDSDSGISLNTS342 (the core degron underlined) in the region of Nrf2 between the Neh5 and Neh6 domains; both are also conserved with the counterparts of Nrf3 and NF-E2 p45 subunits [59]. The latter DSGIS motif in Nrf2 acts as a redox-insensitive β-TrCP (β-transducin repeat-containing protein)-binding degron that negatively regulates the CNC–bZIP protein [110,117]. Similarly, the former DSGLS motif in Nrf1 was also identified as a GSK-3β-mediated phosphodegron, that targets the factor to the β-TrCPSCF-dependent ubiquitin–proteasomal degradation pathway [106]. Also, the DSGLS degron-flanking sequence (i.e. D453SSHSPSSLSS463, the putative phosphorylation sites are italic) on the N-terminal boundary of SR domain in Nrf1 shares homology with the GSK-3β phosphorylation site (D387SSSTCSRLSS397, the putative phosphorylation sites are italic) found in Skn-1 [118], but the latter factor lacks the DSGLS degron.

The SR region as TAD has the ability to positively regulate Nrf1 and TCF11, but not Nrf1β/LCR-F1; this was assessed by domain mapping [57,59,60]. However, Gal4-based experiments suggested that SR negatively regulates Nrf1, particularly Nrf1β/LCR-F1 within distinct fusion contexts [55,60]. The controversial functioning of SR might be exerted through different interacting partner(s) and/or modification (e.g. phosphorylation and O-glycosylation) in a way necessary for up- or down-regulation of Nrf1 target gene expression in different cell lines under different conditions. Furthermore, the positive or negative regulation of Nrf1 by SR might also be dependent on functioning of its adjacent sequences (i.e. SDS1 and SDS2). In addition to the DSGLS-SDS1 degron, the PEST2 sequence acts as a degron which is composed of the entire SR and the flanking SDS2 sequence (E489EGAVGYSSDSETLDLEEAEGAVGYQPEYSK519, in which CRAC4 is underlined) in Neh6L [37,59]. In addition, the SR portion of PEST2 has recently been identified to be required for negative regulation of Nrf1 by its O-GlcNAcylation status [119].

The SDS2 portion of PEST2 is conserved with the amino acids 351–385 region of the Neh6 domain in Nrf2, which also represses the factor through another β-TrCP-binding D373SAPGS378 degron mediated by GSK-3β [110,117]. The PEST2 degron contributes to the negative regulation of Nrf1 by Neh6L [37,55,5860]; its deletion diminishes the generation of small dominant-negative Nrf1γ/δ isoforms (lacking all TAD regions; Figure 3). Furthermore, Nrf1 is negatively regulated by Neh6L through signalling towards its phosphorylation at Ser497 by protein kinase CK2 (as an Nrf1-interacting protein) [94] or at Ser568 (i.e. Ser599 in TCF11) by PKA (protein kinase A) (which enhances interaction with C/EBPβ) [91], resulting in target gene transcriptional suppression. It should be noted that Nrf1 is monitored by a CRAC4-adjoining amphipathic peptide (A507EGAVGYQPEYSKFCRMSY525, the CRAC4 underlined) which is allowed to fold a putative proline-kinked hinge structure (called TMp or proline-kinked transmembrane hinge region) [37]. The dynamic topology is dictated by its net positive semi-hydrophobic region such that TMp may lie on the interface of membranes or span the membranes, as consistent with the positive-inside rule [120,121].

The NTD determines the membrane topology of Nrf1 that enables it to be distinguished from Nrf2

The evidence that the negative regulation of Nrf1 by NTD is associated with the ER was presented originally in 2004 [75] and then published formally in 2006 [38,39]. The salient feature of NTD confers on Nrf1 the ability to exert the unique biological function that is distinctive from Nrf2. The NTD of Nrf1 shares 60% similarity with the N-terminal region of Nrf3; it contains two conserved hydrophobic regions, called NHB1 (N-terminal homology 1) and NHB2 (Box 1). Both sequences G11LLQFTILLSLIGVRVDVD29 and R82LLSQVRALDRFQVPTTEVN AWLVH106 in Nrf1 are represented by G12LLQLTILLSLVGLRVDLD30 and R76LLHEVRALGVPFIPRTRVDAWLVH100 in Nrf3 (the membrane-tethering motifs are underlined) (Figure 3). The NHB1 signal peptide is essential for targeting Nrf1 and Nrf3 to the ER [40,122,123]. The NHB1-associated TM1 (first transmembrane region) (Figures 5 and 6) of Nrf1 enables it to anchor within the membranes and determines the topological folding of other domains; in turn, they are then required to repartition dynamically in and out of ER across membranes into extra-ER subcellular compartments [37,40,41,76]. Interestingly, the ER-targeting signal sequence similar to NHB1 is also present in the ancestral orthologues CncC and Skn-1 (Figure 3), both involved in ER signalling to the response gene expression [124126]. Among the NHB1–CNC subfamily that comprises Nrf1, TCF11, Nrf3, CncC and Skn-1, the evolutionary conservation of the NHB1 sequence suggests that they are folded with a similar membrane topology and also are processed by similar mechanisms [37].

By contrast, the NHB2 peptide, particularly its C-terminal portion (P96TTEVNAWLVH106, the membrane-tethering motifs are underlined) was identified as an ER luminal anchor allowing Nrf1 to reside within the organelle [76], suggesting that it may control dynamic topology of Nrf1 in and out the ER membranes. Upon its dislocation into extra-ER subcellular compartments, the inactive 120-kDa glycoprotein is deglycosylated to yield an active 95-kDa factor, and is processed further by a cytosolic and/or nuclear protease to generate a cleaved 85-kDa protein [36,59,127] (Figures 6 and 7). As expected, the cleaved 85-kDa Nrf1 is completely blocked by Nrf1Δ100−102, Nrf1Δ103−105 or Nrf1W103A/L104A, but not Nrf1N101A/A102L [i.e. the last mutant markedly increases aliphaticity (by 76.1%) and hydrophobicity (by 36.4%) of the NHB2 C-terminal V100NAWLV105, whereas both parameters are decreased by the former three mutants] [59], but all of these mutants do not abolish the expression of the Nrf1 120-kDa glycoprotein and 95-kDa deglycoprotein. Thereby, it is deduced that the selective processing of Nrf1 to yield the cleaved ∼85-kDa protein is likely to depend on the extents to which the V100NAWLV105-adjoining peptide is released from the ER lumen into the cyto/nucleo-plasmic side of membranes insomuch as to gain access to proteolysis by not-as-yet-identified proteases.

A membrane-topobiological model of Nrf1 with distinct functional isoforms

Figure 6
A membrane-topobiological model of Nrf1 with distinct functional isoforms

A membrane-topobiological model is proposed to explain the regulatory feedback circuit between Nrf1 and the 26S proteasome. Nrf1 is a moving membrane protein entailing dynamic membrane topologies that are determined by its TM1 (amino acids 7–26) in co-operation with other amphipathic peptides TMi (amino acids 375–393), TMp (amino acids 507–525) and TMc (amino acids 707–724), each of which is oriented with changing status. After Nrf1 is anchored through its TM1 within the ER membrane, the TM1-connecting TAD (including AD1, NST, AD2 and/or SR) is transiently translocated into the lumen where it is N-glycosylated through its NST domain to become an inactive 120-kDa glycoprotein, although its DBD is retained on the cyto/nucleo-plasmic side. When required, some of the TAD elements within the intact 120-kDa glycoprotein are repositioned via an elusive retrotranslocation pathway driven by p97/VCP from the ER lumen into the cyto/nucleo-plasm where it is deglycosylated to yield an active 95-kDa isoform. The latter deglycoprotein is proteolytically processed to generate a cleaved 85-kDa isoform by a cytosolic proteasomal inhibitor-sensitive protease, e.g. the 26S proteasome, which is de facto activated by limited inhibition through a feedback loop. In turn, transcriptional expression of both the 26S proteasome and p97/VCP complexes is regulated by the 95-kDa and/or 85-kDa Nrf1 isoforms before the CNC–bZIP protein is subjected to further proteasome-mediated proteolytic processing so as to generate small isoforms of between 55 kDa and 25 kDa. Such small polypeptides of Nrf1 may also be produced by in-frame translation of long Nrf1 mRNA transcripts and subsequently degraded further by the proteasome. Of note, Nrf1 is sorted from the ER, and travels through the nuclear pore membrane and is imported to the INM (which is fused with the ONM where the nuclear pore complex is inserted at, and which are also continuous with the ER). Hence we propose that Nrf1 may be subjected to its protein quality-control systems mediated by ERAD and INMAD. Therefore the proteolytic processing of Nrf1 by the putative INMAD-dependent proteasomes (which are predominantly nuclear) might also give rise to several lengths of processed isoforms with distinct functionalities, which, together with those being translocated from the cytoplasm, control distinct subsets of ARE-driven target genes, particularly responsible for maintenance.

Figure 6
A membrane-topobiological model of Nrf1 with distinct functional isoforms

A membrane-topobiological model is proposed to explain the regulatory feedback circuit between Nrf1 and the 26S proteasome. Nrf1 is a moving membrane protein entailing dynamic membrane topologies that are determined by its TM1 (amino acids 7–26) in co-operation with other amphipathic peptides TMi (amino acids 375–393), TMp (amino acids 507–525) and TMc (amino acids 707–724), each of which is oriented with changing status. After Nrf1 is anchored through its TM1 within the ER membrane, the TM1-connecting TAD (including AD1, NST, AD2 and/or SR) is transiently translocated into the lumen where it is N-glycosylated through its NST domain to become an inactive 120-kDa glycoprotein, although its DBD is retained on the cyto/nucleo-plasmic side. When required, some of the TAD elements within the intact 120-kDa glycoprotein are repositioned via an elusive retrotranslocation pathway driven by p97/VCP from the ER lumen into the cyto/nucleo-plasm where it is deglycosylated to yield an active 95-kDa isoform. The latter deglycoprotein is proteolytically processed to generate a cleaved 85-kDa isoform by a cytosolic proteasomal inhibitor-sensitive protease, e.g. the 26S proteasome, which is de facto activated by limited inhibition through a feedback loop. In turn, transcriptional expression of both the 26S proteasome and p97/VCP complexes is regulated by the 95-kDa and/or 85-kDa Nrf1 isoforms before the CNC–bZIP protein is subjected to further proteasome-mediated proteolytic processing so as to generate small isoforms of between 55 kDa and 25 kDa. Such small polypeptides of Nrf1 may also be produced by in-frame translation of long Nrf1 mRNA transcripts and subsequently degraded further by the proteasome. Of note, Nrf1 is sorted from the ER, and travels through the nuclear pore membrane and is imported to the INM (which is fused with the ONM where the nuclear pore complex is inserted at, and which are also continuous with the ER). Hence we propose that Nrf1 may be subjected to its protein quality-control systems mediated by ERAD and INMAD. Therefore the proteolytic processing of Nrf1 by the putative INMAD-dependent proteasomes (which are predominantly nuclear) might also give rise to several lengths of processed isoforms with distinct functionalities, which, together with those being translocated from the cytoplasm, control distinct subsets of ARE-driven target genes, particularly responsible for maintenance.

Intriguingly, the cleaved ∼85-kDa Nrf1 protein is significantly diminished, but not completely abolished, by loss of either the entire NHB2 or its portions (i.e. amino acids 81–90 or 96–106), as well as the spacer region (amino acids 31–80) between NHB1 and NHB2 [59]. Similarly, the cleaved Nrf3-C expression cannot also be abolished by deletion of the NHB2 peptide or its C-terminal portion from the protein [123]. These demonstrate that NHB2-adjacent regions (e.g. AD1, NST and AD2), together with topogons (i.e. TM1, TMi, TMp and TMc), are involved in the topological organization of Nrf1 to modulate its repositioning out of the ER lumen into the cyto/nucleo-plasm [37,41]. This notion is also supported by the finding that cleaved Nrf1 protein was not detected in all cases of NTD/GFPx2, NTD/Nrf2, N156/Nrf2 and N170/Nrf2 (i.e. N156 and N170 indicate the N-terminal 156 amino acids and 170 amino acids of Nrf1) [40,41,76]. Collectively, the selective proteolytic processing of Nrf1 to yield distinct isoforms is controlled by its dynamic membrane topology. However, it remains to be determined whether the basic-hydrophobic amphipathic NHB2 serves as a degradation signal, as a similar amphipathic helical sequence was identified as a functional degron targeting the yeast MATα2 transcription factor to uniquitin–proteasome destruction by the ERAD (ER-associated degradation) cytoplasmic arm [128,129].

Furthermore, the residues 31–81 region between NHB1 and NHB2 is also required to control the proper topological orientation of Nrf1 within and around the ER membrane through the SAS (signal peptide-associated sequence), which shares 55% similarity with the TM1 region, and two close CRAC1/2 motifs (with potential ubiquitination at Lys70) [76]. This region is also postulated to contain a binding site of Hrd1 (3-hydroxy-3-methylglutaryl-CoA reductase degradation protein 1), an E3 ubiquitin ligase involved in ERAD [106]. However, the first dilysine pair (i.e. K4K5), which is positioned on the cytoplasmic side of the ER, at the N-terminal end may also be ubiquitinated in the vectorial processing of Nrf1, because the stability of Nrf1, particularly its 120-kDa glycoprotein, is enhanced by deletion of the K4K5-adjoining two to ten residues [40,41]. Nonetheless, it remains to be determined which lysine residues flanking NHB1 (and the DIDLID/DLG element, Figure 5) are ubiquitinated by E3 ligases (e.g. Hrd1, SCFβ-TrCP or SCFFbw7), before Nrf1 is subjected to proteolytic processing.

PHYSIOLOGICAL FUNCTION OF Nrf1 AND ITS PATHOLOGICAL PHENOTYPES OCCURRING IN ANIMAL MODELS

In mammals, the Nrf subfamily of CNC–bZIP transcription factors (i.e. NF-E2 p45, Nrf1, Nrf2 and Nrf3) are master regulators of the expression of NF-E2/AP1-like ARE-driven genes, that are responsible for antioxidant, detoxification and cytoprotective defences against cellular stress [9,2224]. Among the subfamily, NF-E2 p45 is haemopoietic-specific [46,130], Nrf3 is principally expressed in the placenta with no phenotypes [131133], whereas only Nrf1, rather than Nrf2, is indispensable for development and growth [26,29,30], although Nrf2 is as widely expressed in various tissues as Nrf1 [69]. However, Nrf1 allows for for differential expression to yield multiple mRNA transcripts (of between 1.5 kb and 5.8 kb; Figure 4) and distinct polypeptide isoforms (of between 25 kDa; and 140 kDa; Figure 3) in a vast variety of embryonic, fetal and adult tissues, including liver, brain, kidney, lung, heart, skeletal muscle, bone, testis, ovary, placenta and other tissues [53,55,69,78,134].

Nrf1 is required for embryogenesis because its loss results in lethality

Apart from elevated expression in the heart, midbrain and head mesenchyme between 8 and 9 dpc (days post-coitus), the Nrf1 gene is largely expressed with a constant level of mRNA transcripts detected in all tissues at all developmental stages from 7.5 to 17.5 dpc [134]. The study demonstrates no strict cell- and tissue-specific boundaries for Nrf1 expressed during embryonic development. As such, the unique in vivo functioning of Nrf1 in the development and life processes has been, at least in part, delineated by a list of gene-targeting experiments, in which ESCs (embryonic stem cells) are engineered through distinct strategies for the homologous genomic recombination to generate transgenic mutant mice (Table 2 and Figure 4A).

Table 2
Salient phenotypic features of distinct transgenic Nrf1-KO and -knockin mice

E, embryonic day.

Allele Salient features Reference 
Global Nrf1−/− (Lcrf1tm1uabBlocking mesoderm formation, with no expression of brachyury (T) gene, leads to embryonic lethality at ∼E6.5–E7.5. Decreases in both erythropoiesis and haemoglobin synthesis after in vitro differentiation of KO ESCs is rescued in chimaeric mice, having neither anaemia nor changes in haemoglobins and haematological indices; the defect of Nrf1−/− is non-cell-autonomous [29
Global Nrf1−/− (Nrf1rPGK-neoA non-cell-autonomous defect in fetal liver erythropoiesis results in an anaemia and embryonic death at mid-late gestation (∼E13.5–E16.5). Anaemia results from abnormal maturation of precursor cells in liver microenvironment, because neither histological abnormalities in embryos and fetal organs nor down-regulation of β-globin are examined. Polypeptides of ∼95, 65, 50 and 48 kDa, but not 60 kDa, are recognized by an antibody against amino acids 274–293 of Nrf1β [30
 No contribution of Nrf1−/− ESCs to adult hepatocytes in chimaeric mice is based on trace amounts of ESC-derived GPI-A (glucosephosphate isomerase A) in liver, but not other organs examined [135
 Loss of Nrf1 enhances sensitivity to cytotoxicity of oxidants paraquat and CdCl2, such that free radicals are accumulated with decreased GSH levels, resulting from 3–4-fold down-regulation of GSS and GCLM, but not GCLC, in Nrf1−/− MEFs [31
 ∼1.5–2.0-fold overexpression of NQO1, GCLC and GSTA2, but no alteration of GCLM, is thought to result from loss of p65Nrf1 in Nrf1−/− MEFs, in which major p60 and minor p110 are retained. These three polypeptides are immunoblotted with a purified antibody against Nrf1 [61
 MG132-inducible proteasomal expression is abolished in Nrf1−/− MEFs, which retains a major ∼p85 protein and four minor polypeptides between 120 kDa and 65 kDa being blotted with antibody against amino acids 274–293 of Nrf1. Their expression is eliminated by sh-Nrf1, but a truncated polypeptide recognized by antibodies against the C-terminus of Nrf1 is unaffected [35
Chimaeric Nrf1−/− (Nrf1lacZNrf1−/− ESCs compete with wild-type cells to form the fetal liver in E14.5 chimaeric mice. As fetuses develop, anaemia occurs before apoptosis of Nrf1−/− hepatocytes. The cell-autonomous defect results from ∼2–5-fold down-expression of antioxidant genes (Gclc, Gclm, Gpx-1, Mt-1, Mt-2 and Ho-1) and severe oxidative stress. Nrf1−/− fetal hepatocytes and fibroblasts are susceptible to apoptosis induced by tBHP (t-butylhydroperoxide) and TNFα [135
Double Nrf1−/−:Nrf2−/− (Nrf1rPGK-neo:Nrf2−/−Earlier lethality of double-KO mice (∼E9.5–E11.5) than Nrf1-KO results from increased apoptosis via p53 to Noxa, severe oxidative stress and decreased expression of MT-1, GCLM, GCLC, ferritin H, HO-1 and NQO1, but not SOD1/2. Induction of GCLM, GCLC, HO-1 and Aop2 by diethylmaleate are inhibited partially by Nrf1-KO and completely by Nrf2-KO, or Nrf1−/−:Nrf2−/− [28
 Basal and tBHQ-induced expression of GCLC, c-Jun, c-Fos, p50, p65 and RelB is reduced, along with no changes in JunB, JunD and c-Rel and an increase in Fra-1/2, in Nrf1:Nrf2-null fibroblasts. Nrf1 and Nrf2 restores response to tBHQ via the intact AP1/NF-κB-binding sites [88
Liver-specific Nrf1f/f:AlbCre Viable mice develop steatohepatitis and spontaneous hepatic cancer. Oxidative stress and damage results from reduced expression of GSTM3/6 and GSTP2 and induced ω-oxidation of fatty acids, leading to ER proliferation. CYP4a10, CYP4a14, GSTA1 and MT1/2 are overexpressed, whereas PPARα and target genes Acox1, Acca1 and Ehhadh are not altered [32
 An increase in micronuclei formation and chromosomal instability is linked to reduced expression of kinetochore (Ndc80, Nuf2 and Spc25) and spindle (Sgol1) checkpoint genes [136
Liver-specific Nrf1f/f:AlbCre Liver damage and NASH are examined at 8 weeks. Expression of 52 Nrf1-dependent genes (e.g. Mt-1/2) is decreased more than 3-fold, whereas induction of 20 Nrf2 target genes by Nrf1−/− is abrogated by double KO of Nrf1 and Nrf2 [33
Nrf1dN/−:AlbCre Lipid accumulation in Nrf1−/− liver at 5 weeks results from 1500 genes being up-regulated and 1700 genes being down-regulated. Besides the 26S proteasome, genes involved in lipid metabolism (e.g. Lipin1, Pgc-1β, Pparα and target genes), amino acid (e.g. methionine) metabolism, the TCA cycle and mitochondrial respiration are decreased, whereas genes involved in the cell cycle and DNA replication are increased. Nrf1 target genes are unaffected by loss of Nrf2 or Keap1 [234
Liver-inducible Nrf1f/f:Cyp1a1-Cre Acute 3-MC-inducible loss of Nrf1 causes lipid accumulation and hepatic damage, resulting from overexpression of lipoprotein receptors and several lipid-metabolizing enzymes. Increased GSH level is associated with cysteine accumulation and xCT up-regulation [138
Brain-specific Nrf1f/f:Camk2Cre (Cre-ERT2) Age-dependent forebrain atrophy from 3 to 6 months results from apoptotic death of neurons, with impairment of proteasomal function leading to ubiquitin accumulation. 26S proteasome subunit genes are down-regulated in Nrf1-deficient cells, which enable hypersensitivity to proteasome inhibitors. However, no oxidative stress is increased in the pathogenesis of neurodegeneration [147
CNS-specific Nrf1f/f:Nestin-Cre Both progressive motor ataxia and severe weight loss lead to postnatal lethality within 3 weeks. The pathological phenotypes are similar to those of sMaf-deficient mice. Ubiquitinated proteins are accumulated in various CNS regions leading to neuronal loss in the hippocampus. Increased oxidative stress up-regulates HO-1 (and possibly MafG) in the spinal cords [154
Bone-specific Nrf1f/f:Col1a2-iCre Bone size, peak bone mass, trabecular number and mechanical strength are reduced in osteoblast-specific Nrf1-null mice. Down-regulation of Osx by ∼60% leads to a striking reduction in cell differentiation by 68% and impairment of bone formation. Deficiency of Nrf1 also up-regulates Nrf2 by ∼40%, but down-regulates Nrf3 and MafF by ∼60% and 50% respectively [162
Nrf1−/−:Nrf1-Tg (C/T, rs3764400) Induction of Nrf1-Tg represses body weight gain through inhibiting glucose metabolism so as to increase the risk of diabetes mellitus with insulin resistance induced. Nrf1 suppresses insulin-regulated glycolysis genes leading to loss of hepatic glucose 6-phosphate and fructose 6-phosphate, but increases entry into the TCA cycle [247
Allele Salient features Reference 
Global Nrf1−/− (Lcrf1tm1uabBlocking mesoderm formation, with no expression of brachyury (T) gene, leads to embryonic lethality at ∼E6.5–E7.5. Decreases in both erythropoiesis and haemoglobin synthesis after in vitro differentiation of KO ESCs is rescued in chimaeric mice, having neither anaemia nor changes in haemoglobins and haematological indices; the defect of Nrf1−/− is non-cell-autonomous [29
Global Nrf1−/− (Nrf1rPGK-neoA non-cell-autonomous defect in fetal liver erythropoiesis results in an anaemia and embryonic death at mid-late gestation (∼E13.5–E16.5). Anaemia results from abnormal maturation of precursor cells in liver microenvironment, because neither histological abnormalities in embryos and fetal organs nor down-regulation of β-globin are examined. Polypeptides of ∼95, 65, 50 and 48 kDa, but not 60 kDa, are recognized by an antibody against amino acids 274–293 of Nrf1β [30
 No contribution of Nrf1−/− ESCs to adult hepatocytes in chimaeric mice is based on trace amounts of ESC-derived GPI-A (glucosephosphate isomerase A) in liver, but not other organs examined [135
 Loss of Nrf1 enhances sensitivity to cytotoxicity of oxidants paraquat and CdCl2, such that free radicals are accumulated with decreased GSH levels, resulting from 3–4-fold down-regulation of GSS and GCLM, but not GCLC, in Nrf1−/− MEFs [31
 ∼1.5–2.0-fold overexpression of NQO1, GCLC and GSTA2, but no alteration of GCLM, is thought to result from loss of p65Nrf1 in Nrf1−/− MEFs, in which major p60 and minor p110 are retained. These three polypeptides are immunoblotted with a purified antibody against Nrf1 [61
 MG132-inducible proteasomal expression is abolished in Nrf1−/− MEFs, which retains a major ∼p85 protein and four minor polypeptides between 120 kDa and 65 kDa being blotted with antibody against amino acids 274–293 of Nrf1. Their expression is eliminated by sh-Nrf1, but a truncated polypeptide recognized by antibodies against the C-terminus of Nrf1 is unaffected [35
Chimaeric Nrf1−/− (Nrf1lacZNrf1−/− ESCs compete with wild-type cells to form the fetal liver in E14.5 chimaeric mice. As fetuses develop, anaemia occurs before apoptosis of Nrf1−/− hepatocytes. The cell-autonomous defect results from ∼2–5-fold down-expression of antioxidant genes (Gclc, Gclm, Gpx-1, Mt-1, Mt-2 and Ho-1) and severe oxidative stress. Nrf1−/− fetal hepatocytes and fibroblasts are susceptible to apoptosis induced by tBHP (t-butylhydroperoxide) and TNFα [135
Double Nrf1−/−:Nrf2−/− (Nrf1rPGK-neo:Nrf2−/−Earlier lethality of double-KO mice (∼E9.5–E11.5) than Nrf1-KO results from increased apoptosis via p53 to Noxa, severe oxidative stress and decreased expression of MT-1, GCLM, GCLC, ferritin H, HO-1 and NQO1, but not SOD1/2. Induction of GCLM, GCLC, HO-1 and Aop2 by diethylmaleate are inhibited partially by Nrf1-KO and completely by Nrf2-KO, or Nrf1−/−:Nrf2−/− [28
 Basal and tBHQ-induced expression of GCLC, c-Jun, c-Fos, p50, p65 and RelB is reduced, along with no changes in JunB, JunD and c-Rel and an increase in Fra-1/2, in Nrf1:Nrf2-null fibroblasts. Nrf1 and Nrf2 restores response to tBHQ via the intact AP1/NF-κB-binding sites [88
Liver-specific Nrf1f/f:AlbCre Viable mice develop steatohepatitis and spontaneous hepatic cancer. Oxidative stress and damage results from reduced expression of GSTM3/6 and GSTP2 and induced ω-oxidation of fatty acids, leading to ER proliferation. CYP4a10, CYP4a14, GSTA1 and MT1/2 are overexpressed, whereas PPARα and target genes Acox1, Acca1 and Ehhadh are not altered [32
 An increase in micronuclei formation and chromosomal instability is linked to reduced expression of kinetochore (Ndc80, Nuf2 and Spc25) and spindle (Sgol1) checkpoint genes [136
Liver-specific Nrf1f/f:AlbCre Liver damage and NASH are examined at 8 weeks. Expression of 52 Nrf1-dependent genes (e.g. Mt-1/2) is decreased more than 3-fold, whereas induction of 20 Nrf2 target genes by Nrf1−/− is abrogated by double KO of Nrf1 and Nrf2 [33
Nrf1dN/−:AlbCre Lipid accumulation in Nrf1−/− liver at 5 weeks results from 1500 genes being up-regulated and 1700 genes being down-regulated. Besides the 26S proteasome, genes involved in lipid metabolism (e.g. Lipin1, Pgc-1β, Pparα and target genes), amino acid (e.g. methionine) metabolism, the TCA cycle and mitochondrial respiration are decreased, whereas genes involved in the cell cycle and DNA replication are increased. Nrf1 target genes are unaffected by loss of Nrf2 or Keap1 [234
Liver-inducible Nrf1f/f:Cyp1a1-Cre Acute 3-MC-inducible loss of Nrf1 causes lipid accumulation and hepatic damage, resulting from overexpression of lipoprotein receptors and several lipid-metabolizing enzymes. Increased GSH level is associated with cysteine accumulation and xCT up-regulation [138
Brain-specific Nrf1f/f:Camk2Cre (Cre-ERT2) Age-dependent forebrain atrophy from 3 to 6 months results from apoptotic death of neurons, with impairment of proteasomal function leading to ubiquitin accumulation. 26S proteasome subunit genes are down-regulated in Nrf1-deficient cells, which enable hypersensitivity to proteasome inhibitors. However, no oxidative stress is increased in the pathogenesis of neurodegeneration [147
CNS-specific Nrf1f/f:Nestin-Cre Both progressive motor ataxia and severe weight loss lead to postnatal lethality within 3 weeks. The pathological phenotypes are similar to those of sMaf-deficient mice. Ubiquitinated proteins are accumulated in various CNS regions leading to neuronal loss in the hippocampus. Increased oxidative stress up-regulates HO-1 (and possibly MafG) in the spinal cords [154
Bone-specific Nrf1f/f:Col1a2-iCre Bone size, peak bone mass, trabecular number and mechanical strength are reduced in osteoblast-specific Nrf1-null mice. Down-regulation of Osx by ∼60% leads to a striking reduction in cell differentiation by 68% and impairment of bone formation. Deficiency of Nrf1 also up-regulates Nrf2 by ∼40%, but down-regulates Nrf3 and MafF by ∼60% and 50% respectively [162
Nrf1−/−:Nrf1-Tg (C/T, rs3764400) Induction of Nrf1-Tg represses body weight gain through inhibiting glucose metabolism so as to increase the risk of diabetes mellitus with insulin resistance induced. Nrf1 suppresses insulin-regulated glycolysis genes leading to loss of hepatic glucose 6-phosphate and fructose 6-phosphate, but increases entry into the TCA cycle [247

A global mutant of Nrf1 (i.e. Lcrf1tm1uab, in which a 3.5-kb genomic sequence with codons of amino acids 172–741 was deleted) resulted in embryonic lethality in the early gastrulation on Black Swiss outbred backgrounds [29]. The Nrf1−/− embryos developed to the late egg cylinder stage indistinguishable from wild-type counterparts, but thereafter development was arrested at ∼6.5 dpc and most died before 7.5 dpc, owing to the failure to form the primitive streak mesoderm, although ectoderm and visceral endoderm layers appeared normal. The direct defect of the homozygous mutation is, however, not cell-autonomous, because these mutant ESCs were rescued after injection into the wild-type blastocysts and also contributed to all cell lineages in chimaeras examined [29]. It is hence demonstrated that the functioning of Nrf1 is, although indirectly, required for the transcriptional expression of some critical genes that constitutively control the mesoderm formation.

Another quite different phenotype was presented by the global knockin mutant (i.e. Nrf1rPGK–neo, in which the phosphoglycerate kinase–neomycin cassette was reversely inserted into 5′-end of the bZIP-coding region) on a C57BL/6J blastocyst background [30]. The targeted disruption of Nrf1 resulted in embryonic lethality at mid-late gestation from 13.5 to 18.5 dpc. The Nrf1−/− embryos died in utero of decreased definitive enucleated red cells and the ensuing anaemia that resulted from impaired maturation of erythroid progenitors in the fetal liver micro-environment without increased apoptosis of the haemopoietic cells [30]. Although the lack of cell autonomy suggested an indirect role for Nrf1 in erythopoiesis, Nrf1 acts as an essential gene in the erythroid cells because none of the putative compensatory functions were efficiently executed by other CNC–bZIP proteins, such as NF-E2 p45 and Nrf2 that are expressed at high levels [46,47,53].

Although early embryonic lethality of the Nrf1-deficient mice [30] and the hepatocyte-specific function in fetal and adult livers [135] are not rescued by the putative compensative function of Nrf2 that is not essential for development, it is surmised that there may exist certain overlapping functions of the two CNC–bZIP factors and their potential redundancy in embryogenesis, development and oxidative stress response. By contrast with the Nrf1−/− embryonic lethality at middle to late gestation starting at 13.5 dpc [30], the double-KO embryos of Nrf1:Nrf2 caused earlier death at 10.5 dpc due to extensively increased apoptosis and severe oxidative-stress-induced growth retardation [28]. The p53-Noxa-mediated apoptosis was markedly induced by elevated levels of ROS (i.e. by measurement of H2O2 and singlet oxygen) resulting from severely impaired expression of antioxidant defence genes [e.g. Mt-1 (metallothionein 1), Gclm, Gclc, Fth1 (ferritin H), Ho-1 and Nqo1, but not Sod1/2 (superoxide dismutase 1/2)], when compared with the individual KO of Nrf1 or Nrf2. Hence the functioning of Nrf2 is allowed for a certain compensation for the loss of Nrf1 function in regulating genes essential for the intracellular redox threshold status during early embryogenesis. It seems plausible that the partially overlapping functions of Nrf1 and Nrf2 is determined by both their co-expression patterns [47,53,56] and sequence similarity beyond the CNC–bZIP domains (Figure 3).

To provide an understanding of the molecular basis for the Nrf1−/− phenotype, different experimental settings were employed to identify which truncated Nrf1 species were deleted or retained, and which downstream genes were dysregulated in the mutant MEFs (mouse embryonic fibroblasts). Unexpectedly, these experiments have given rise to controversial results. First, only p60 protein was not detected (but polypeptides of ∼95, 65, 50 and 48 kDa were retained) by Western blotting of the Nrf1−/− lyastes with a peptide antibody against amino acids 565–586 of Nrf1 (i.e. amino acids 274–293 in Nrf1β/LCR-F1/p65Nrf1) [30]. Decreased glutathione (GSH) levels resulted from ∼3–4-fold down-expression of genes encoding GSH biosynthesis enzymes GSS and GCLM, but not GCLC, in Nrf1−/− MEFs [31], such that the intracellular levels of superoxide, a major representative of ROS, are relatively increased to compromise membrane integrity, leading to red cell destruction and even haemolytic anaemia. Secondly, rabbit antibody against Nrf1 (purified from anti-Nrf1/GST fusion sera [39]) was subjected to Western blotting of the Nrf1−/− lyastes, showing that p120 and p65 proteins were undetectable, with the minor p110 and major p60 being retained [61]. Conversely, ARE-battery genes such as Nqo1, Gclc and Gsta2 (GST Alpha 2), but not Gclm, were up-regulated in Nrf1−/− MEFs, although lacking p120 and p65 [61]; this appears to contradict the above claims [31]. Thirdly, the rabbit polyclonal antibody against the N-terminus of Nrf1 (raised in the Chan laboratory [30,61]) was subjected again to Western blotting, revealing that Nrf1−/− MEFs expressed a major extra ∼85-kDa protein, with the minor p120 and p110, together with two proteins closely related to p65, but not the exact p65, being retained to considerably lowered levels [35]. All of these remaining proteins were abolished further by retrovirus expressing si-Nrf1 (shRNA-interfering Nrf1), despite no changes in a small polypeptide recognized by a C-terminus-specific antibody (Novus Biologicals). More importantly, induced transcription of the proteasome (PSM) genes was abrogated in MEFs of Nrf1−/−, but not Nrf2−/− [35]. Collectively, Nrf1−/− (from the knockin mutant Nrf1rPGK–neo) cells are enabled to express certain Nrf1 isoforms to various extents, and thus the resulting experimental data should be interpreted with extreme caution. Such isoforms of Nrf1 are inferrable to be putative by-products from a fraction of mRNA transcripts being alternatively spliced to remove certain portions of the knocked-in exogene-coding sequence or en bloc; their residual expression may result in a disparity in the obvious pathophysiological phenotypes occurring between different strategic KO mice (cf. Nrf1rPGK–neo [30] with Lcrf1tm1uab [29]).

Nrf1 contributes to liver development and protects the hepatocytes against apoptosis

The livers in Nrf1−/− embryos (from Nrf1 rPGK–neo) were noticeably smaller, lighter and hypoplastic, prominently by 13.5–14.5 dpc [30]. This feature of the developmental arrest cannot be explained by the impaired erythropoiesis in the fetal liver, because these Nrf1−/− erythropoietic cells grew normally in vitro and Nrf1−/− ESCs contributed efficiently to erythroid cells of chimaeric mice (generated by injecting Nrf1−/− ESCs into the wild-type blastocysts) [30]. Several of the chimaeric embryos were observed with anaemia that occurred before detection of cell death in the liver parenchyma [135]; i.e. the anaemia is secondary to the failure to sustain haemopoiesis in the impaired liver microenvironment.

The embryonic lethal phenotype of Nrf1−/− (from mice homozygous for Nrf1lacZ, in which a 2.2-kb genomic fragment encoding the bZIP domain was deleted and replaced with an IRES-βGeo cassette [135]) was similar to that of KO mice from Nrf1rPGK–neo [30]. The unfortunate lethality of embryos precludes further analysis of functioning of Nrf1 in fetal liver development. Alternatively, the chimaeric mice were generated by injecting positive Nrf1−/− ESC clones with a 129/Sv background into the blastocysts from C57BL/6J mice in which Nrf1 is normally expressed [135]. Characterization of these chimaeric mice at 8–16 weeks of age revealed that loss of Nrf1 resulted in an impaired contribution of the mutant ESCs to adult, but not fetal, hepatocytes, although the Nrf1−/− ESCs contributed to other tissues in adult animals, including the lung, kidney, muscle and heart where Nrf1 was, if under normal genetic conditions, expressed at high levels [135].

Further analysis of these chimaeric embryos at 14.5 dpc showed that the Nrf1−/− ESCs contributed to fetal liver development. Whereas Nrf1 is not required for early fetal liver development, the hepatocytes in chimaeric embryos underwent widespread apoptosis in the late gestation [135]. The hepatocytic apoptosis was determined to be associated with increased oxidative stress and decreased expression of antioxidant genes (e.g. Gclc, Gclm, Gpx1, Ho-1 and Mt-1/2). Therefore these findings demonstrate an essential cell-autonomous role of Nrf1 in the survival of hepatocytes during embryonic and fetal development, possibly by maintaining redox homoeostasis and protecting embryonic hepatocytes from apoptosis induced during development.

Liver-specific loss of Nrf1 results in NASH and hepatoma

The above-described chimaeric model cannot be employed to determine whether the critical role of Nrf1 in fetal liver development is different from its function in the mature hepatocytes, because the massive cell death and degeneration existed in livers of chimaeric embryos derived from the Nrf1−/− ESCs. To bypass this obstacle and continue on the road, conditional KO of Nrf1 by using a Cre-loxP transgenic system was used to determine the function of Nrf1 in the hepatocytes beyond embryonic development [32,33].

Liver-specific disruption of Nrf1 (to delete the last one exon encoding amino acids 296–741) in the adult mouse results in the pathological phenotype that resembles NASH and hepatic neoplasia, including hepatocellular adenomas and carcinomas, which spontaneously developed as early as 4 months after birth [32]. Before cancer development, Nrf1−/− livers exhibited interrelated steatosis, apoptosis, necrosis, inflammation and fibrosis. In addition to the pre-cancerous lesions, loss of Nrf1 could directly contribute to tumorigenesis by promoting chromosome mis-segregation [136]. 4OHT (4-hydroxytamoxifen)-inducible KO of Nrf1 (from Nrf1flox/flox:Cre-ERT2) increased the numbers of abnormal nuclei and micronuclei 3-fold higher than controls. Similar abnormalities were also observed following sh-Nrf1 KD. The Nrf1−/−-led genetic instability appeared to be associated with decreased expression of both the kinetochore genes Ndc80, Nuf2 and Spc25 and the spindle assembly checkpoint gene Sgol1 [136]. However, further studies are required to determine whether Nrf1 can function as a tumour suppressor in hepatocytes.

To gain an insight into the molecular pathological basis for NASH, the Nrf1−/− livers and related cells were subjected to further examination. It was found that the Nrf1-deficient hepatocytes had increased susceptibility to oxidative stress and damage, along with down-regulation of some ARE-battery genes [e.g. Gstm3 (GST Mu 3), Gstm6 and Gstp2 (GST Pi 2)] and up-regulation of Cyp4A genes [32]. Loss of Nrf1 function resulted in a significant increase in intracellular level of superoxide (representing ROS), which was at least in part generated by ω-oxidation of fatty acids by means of proliferated microsomal CYP4A (cytochrome P450 4A) (i.e. in the ER). The increased superoxide level is associated with elevated lipid and other pathogenesis leading to NASH. Similar pathology also emerged from the damaged livers of hepatocyte-specific Nrf1−/− mice [33]. The pathophysiological phenotype demonstrates that Nrf1 is vital to mediate expression of such genes involved in maintaining both cellular redox and lipid homoeostasis in the liver.

Besides lipids, a bulk of ubiquitinated (and oxidative damaged) proteins were also accumulated in the Nrf1−/− hepatocytes, leading to the ER-stress-associated steatosis [137]. This results from impaired transcription of the PSM genes and increased expression of the ER-stress-response genes [e.g. Atf4, Atf6, Bip (immunoglobulin heavy-chain-binding protein), Chop (C/EBP-homologous protein), and Gadd45β (growth-arrest and DNA-damage-inducible protein 45β) and Herp (homocysteine-induced ER protein)], as accompanied by phosphorylation of both PERK [PKR (dsRNA-dependent protein kinase)-like ER kinase] and its substrate eIF2α (eukaryotic initiation factor 2α) [137] (Figure 8). Microarray analysis of the Nrf1−/− livers revealed that 52 of the Nrf1-dependent genes [i.e. Mt-1/2, Clrf (cytokine receptor-like factor), Gcn20 (general control non-derepressible 20), Gadd45γ, Mfsd3 (major facilitator superfamily domain-containing 3), Pdk4 (pyruvate dehydrogenase kinase 4) and Sp3 (specificity protein 3)] were down-regulated more than 3-fold. Intriguingly, adaptive activation of 20 Nrf2 target genes by the single KO of Nrf1 was terminated by the double KO of Nrf1:Nrf2 [33], implying that the functioning of Nrf1 is required for the basal constitutive expression of cytoprotective genes against the endogenous oxidative stress that activates Nrf2. However, the prototypical Nrf2 target genes (i.e. Gclc, Gclm, Gss and Nqo1) were not significantly changed by acute loss of hepatic Nrf1 upon stimulation of the inducible KO mice (Nrf1flox/flox:Cyp1A1-Cre) by 3-MC (3-methylcholanthrene) that leads to profound NASH, without obvious oxidative stress [138]. These findings demonstrate that Nrf1 regulates a separate battery of genes distinct from that for Nrf2.

Nrf1-interacting partners appear to be responsible for maintaining liver homoeostasis

Although Nrf1 has been determined to execute a unique function that is indispensable for liver development and its homoeostatic integrity [30,32,33,135], the Nrf1 gene is not exclusively expressed in the liver at a higher level than those measured in other tissues [134]. In view of this, it is postulated that the unique functioning of Nrf1 in the liver could also be regulated by one of its specific interacting proteins (e.g. sMaf and ATF4). Among the Nrf1-dimerizated partners, Atf4 has shown elevated expression pattern in many of the same in situ hybridization sites as Nrf1 [134]. Like the Nrf1−/− phenotype [30,135], global inactivation of Atf4−/− resulted in severe fetal anaemia because of impaired fetal liver haemopoiesis, but relatively normal steady-state erythropoiesis in adults [139]. However, it is different from Nrf1−/− in that the cell-autonomous defect in proliferation of either haemopoietic progenitors from the small Atf4−/− embryos or the primary MEFs was not prevented by exogenous growth factors [139]. Although ATF4 was indeed suggested to be essential for normal fetal liver haematopoiesis, the fact that both Nrf1−/− and Atf4−/− caused similar 2-fold decreases in haematocrit levels and the number of haemopoietic progenitors per fetal liver at 15.5 dpc [30,139] raises the possibility that a functional heterodimer of Nrf1 and ATF4 is allowed for putative co-operation to regulate certain key genes that are essentially controlling normal fetal liver haemopoiesis.

Of the founding heterodimerization partners, MafG is also as widely expressed during development as Nrf1 [134], but global MafG−/− mice only showed mild thrombocytopenia and motor atxia [140]. By contrast, MafK is only expressed in specific tissues (e.g. liver, heart and brain) at the later stages (13.5∼17.5 dpc), whereas MafF expression was only detectable [134]. Since neither MafK−/− nor MafF−/− showed any apparent phenotypes [141], it is reasoned that they are functionally redundant with compensation for an individual loss of function overlapping with one another. This is supported by the fact that the sMaf triple-KO mice showed severe growth retardation and liver hypoplasia from 9.5 dpc, until the embryos died at ∼13.5 dpc [142], indicating that sMafs are essential for embryogenesis after 9 dpc, particularly in fetal liver development. Importantly, the liver of sMaf triple-KO embryos showed various oxidative-stress-induced apoptosis, resulting from significantly reduced expression of ARE-driven cytoprotective genes [e.g. Gclc, Gsta4, Gstp2, Mrp3, Mrp5, Nqo1 and Txnrd1 (thioredoxin reductase 1)], which resembles the phenotypic livers of the Nrf1:Nrf2 double-KO embryos. The evidence strongly supports the notion that Nrf1 (and/or Nrf2) requires an obligatory Maf partner to form a bona fide functional heterodimer that co-operatively executes a uniquely important role in regulating NF-E2/AP1-like ARE-driven genes involved in embryogenesis and fetal liver development under normal physiological conditions.

An additional heterodimer of Nrf1 with each partner of AP1 complexes (Jun, JunB, JunD, Fos, FosB, Fra-1, Fra-2, ATF2 and CREB) enables it to regulate expression of Nrf1 target genes [62,63,81,87,88]. Interestingly, global Jun−/− embryos died at mid-gestation (∼12.5 dpc), with abnormalities in liver [143,144], whereas hepatocyte-specific deletion of Jun−/− in adult livers affects cell survival and cell cycle progression [145], demonstrating that Jun is essential for fetal liver development. The defect of Jun−/− fetal livers was determined to result from increased oxidative stress, down-regulation of Nrf1, Nrf2 and other genes [Gclc, Gclm, Gss, Cdc2, Cycd1 (cyclin D1), Egfr (epidermal growth factor receptor), Akt and Gsk-] and up-regulation of p53 and p21; the cell-autonomous defect was also rescued by restoring JunD [146]. Collectively, it is of critical importance that Jun (and JunD) is enabled to function as an upstream activator to mediate transcriptionl expression of Nrf1, Nrf2, ARE-driven genes and others that are involved in cytoprotection against oxidative stress (Figure 9).

Brain-specific loss of Nrf1 leads to neurodegenerative pathology

During embryonic development (from 8 to 9 dpc), elevated expression of Nrf1 becomes pronounced throughout the head mesenchyme and contributes to the migrating cranial neural crest [134]. In adult brains, although Nrf1 is widely expressed, its high levels were detected in the cortex, cornu ammonis subregions and dentate gyrus of the hippocampus and choroid plexus [147]. In the lesioned hippocampal neurons, strong inducible expression of Nrf1 mRNA and protein, along with the cytoprotective enzyme HO-1 co-localized in the nucleus, was accompanied by relatively weak elevation of Nrf2 [148], suggesting a potential role of Nrf1 in both brain homoeostasis and neuronal survival after acute brain injury. Notably, single KO of genes (i.e. Nfe2p45, Nrf2, Nrf3, Bach1 and Bach2) other than Nrf1 does not give rise to apparent neuronal phenotypes [149153]. Therefore it is postulated that the only remaining candidate within the CNC–bZIP family responsible for neuronal homoeostasis is Nrf1. In fact, Nrf1 does contribute to CNS (central nervous system) development after birth; this was shown by conditional KO of Nrf1 in brains [147,154], although global KO of the gene did not cause any apparent neuronal deficits in mouse embryonic stages before death [31,135].

To gain insight into the in vivo function of Nrf1 in adult brains, two different transgenic mouse lines were generated by cross-breeding floxed Nrf1 mice (i.e. Nrf1flox/flox) with transgenic mice bearing Camk2Cre (Ca2+/calmodulin-dependent protein kinase II-Cre) [147] or Nestin-Cre [154]. The former brain-specific Nrf1-KO mice exhibited a neurodegenerative pathological phenotype: age-dependent forebrain atrophy (from 3 to 6 months), resulting from apoptotic death of neurons and impaired proteasomal function in damaged neurons [147]. The Nrf1 deficiency caused down-regulation of the proteasome subunit genes in the neurons such that they acquired hypersensitivity to proteasome inhibitors. However, it is intriguing that Nrf1-deficient neuronal dysfunction was not accompanied by increased oxidative stress during the neurodegenerative pathogenesis [147]. By contrast, the entire CNS-specific Nrf1 KO in mice resulted in a progressive motor ataxia with severe weight loss leading to postnatal lethality (within 3 weeks) [154]. The Nrf1−/− phenotypes are similar to those of sMaf-deficient mice (such as MafG−/−:MafK−/−) [155]. Further examination revealed that accumulation of ubiquitinated proteins in various CNS regions appeared to result in neuronal losses in the hippocampus, while increased oxidative stress resulted from decreased expression of Nqo1 and Gsta4, as accompanied by increased expression of Ho-1 (and possibly MafG) in the spinal cords. Overall, Nrf1 is required for the steady-state of CNS homoeostasis.

Nrf1 is required for differentiation of odontoblasts and osteoblasts

In undifferentiated odontoblasts, the physical interaction of Nrf1 with C/EBPβ through both bZIP domains, with respect to the binding efficiency dependent on phosphorylation by PKA at Ser569 (i.e. Ser599 situated on the boundary between the Neh6L and CNC domains in TCF11; Figure 5), directs the specificity of the CNC–bZIP function to repress the basal constitutive expression of the DSPP (dentin sialophosphoprotein) gene encoding two specific markers DPP (dentin phosphophoryn) and DSP (dentin sialoprotein) [91]. Conversely, in fully differentiated odontoblasts, the loss of interaction between Nrf1 and C/EBPβ caused an increase in transcriptional expression of DSPP to yield DPP and DSP (Figure 10B).

An insight into the molecular mechanisms by which ascorbic acid (i.e. a redox inducer also called vitamin C) induces osteoblast differentiation [156,157] revealed that a significant inducible increase in the expression of Osx (osterix; an osteoblast-specific transcription factor) was mediated by Nrf1 binding to the ARE sequence (located from −1762 to −1733 upstream of the transcription start site of Osx) in impaired BMS (bone marrow stromal) cells [158]; the mutant cells were derived from the spontaneous fracture mice caused by deletion of the gulonolactone oxidase gene leading to deficiency of vitamin C [159]. Albeit all the Nrf subfamily members have a potential to bind the ARE sequence in the Osx promoter (Figure 10C), the inducible expression level of Nrf1 was 50- and 500-fold higher than those of Nrf2 and Nrf3 respectively, whereas other CNC–bZIP proteins were not altered in vitamin C-treated BMS cells [158]. The notion that Nrf1 is essential for transcriptional regulation of Osx-mediated gene expression is also evident from KD of si-Nrf1, which decreased Osx expression, and impaired osteoblast differentiation and mineralized nodule formation from BMS cells [158]. By contrast, Nrf2 was considered to negatively regulate differentiation of osteoblasts and chondrocytes [160,161].

The in vivo function of Nrf1 in regulating bone formation has been determined using osteoblast-specific KO mice that were generated by cross-breeding Nrf1flox/flox mice [147] with transgenic mice bearing Col1a2-iCre [i.e. improved Cre recombinase under the control of regulatory sequence of the Col1α2 (collagen-1α2) gene] [162]. The osteoblast-specific Nrf1-null mice exhibited a reduction of bone size, peak bone mass, trabecular number and mechanical strength. This phenotype was accompanied by decreased expression of Osx by 60% leading to a reduction in cell differentiation (by 68%) and impairment of bone formation. Intriguingly, deficiency of Nrf1 was found to enable up-regulation of Nrf2 by 40% and down-regulation of Nrf3 and MafF by 60% and 50% respectively [162]. It is hence deduced that a potential interdependent complementary pathway may be controlled between Nrf1 and Nrf3; this was supported by evidence that both transcript levels were decreased upon exposure of wild-type and Nrf3−/− mice to BHT (butylated hydroxytoluene) [163], and both glycoproteins were also co-localized within and around the ER, with a similar membrane topology determined [37,41,122,123]. However, whether such a transcriptional inter-regulatory feedback circuit exists between Nrf1 and Nrf3 remains to be further determined.

CONTRIBUTION OF Nrf1 TO MAINTAINING REDOX HOMOEOSTASIS AGAINST OXIDATIVE STRESS

Collectively, both Nrf1 and Nrf2 are enabled to exert distinct important roles in differentially or co-ordinately regulating the basal and inducible expression of either the same or different subsets of NF-E2/AP1-like ARE/EpRE-driven cytoprotective genes [33,164166]. Nrf2 is a master regulator of both to transactivate antioxidant genes, and has an ability to execute dual-opposing roles in: (i) adaptive cytoprotection of normal cells against oncogenic agents, which is considered a new strategy for cancer chemoprevention; and (ii) permanent cytoprotection of damaged (mutagenic or tumour) cells from apoptosis, which enables the promotion of carcinogenesis and the acceleration of malignant progression [167,168]. By contrast, Nrf1 should be viewed as a robust regulator to mediate the constitutive expression of ARE-driven cytoprotective genes, some of which are critical for maintaining redox homoeostasis, particularly under normal physiological conditions. Nrf1 is also involved in regulating homoeostatic cytoprotective defences against distinct cellular stresses (Table 3).

Table 3
Nrf1 mediates induction of ARE/EpRE-driven responses to distinct stresses
Reagents (dose) Salient responses Reference(s) 
β-Naphthoflavone (β-NF; ∼10–50 μM), tBHQ (∼50–150 μM) Induction of Nrf1-mediated hARE-tk-CAT reporter genes (from NQO1, GCLC or GST-Ya) by β-NF is inhibited by c-Fos and Fra-1, whereas dual opposing effects of c-Jun depend on its extent to be expressed in transfected HepaG2 cells [81,87
Paraquat (∼0.4–1.6 mM), t-butylhydroperoxide (tBHP; 250 μM), TNFα, CdCl2 (∼50–200 μM) Compared with wild-type MEFs, loss of Nrf1 impairs antioxidant gene expression, leading to more sensitivity to cell killing by paraquat and CdCl2. Nrf1-deficent cells are also susceptible to apoptosis induced by tBHP and TNFα (∼2–50 ng/ml) [31,135
Menadione (∼20–100 μM), H2O2 (∼50–300 μM) Resistance of CFSCs (CCl4-induced cirrhotic fat-sorting cells in rat livers) to menadione toxicity is associated with nuclear accumulation of Nrf1 and p53, and induction of NQO1 via Nrf1-EpRE [179
HNE (15 μM) EpRE of the γ-glutamyltranspeptidase gene is induced to bind Nrf1, Nrf2, c-Fos, Fra-1, plus JunB, c-Jun, FosB and Fra-2 in untreated epithelial cells. Both inducible binding activity and expression of its mRNA are blocked by inhibitors of ERKs and p38, but not PKC [172
Genistein (∼25–50 μM), daidzein (∼25–50 μM), H2O2 (50 μM) Induction of glutathione peroxidase (GPx) and GCLC by genistein is mediated via Nrf1, but not Nrf2, to defend against oxidative damages to endothelial EA.hy926 cells [180
Bisphenol A (∼100–200 μM) Induction of Nrf1, Nrf2 and target antioxidant genes Ho-1, Nqo1 and Gclc is accompanied by increased GSH levels [175
Hypoxia (1% O2), CoCl2 (200 μM) Stabilization of Nrf1 p120, p110 and p65, but not p23, by MG132 is enhanced by hypoxia. Nrf1 target reporter is activated by CoCl2 [67
H2O2 (∼10–100 μM: pre-treated for 10 min before recovery for 24 h, then stress for 30 min) Pre-conditioning increases nuclear accumulation of Nrf1 (and Nrf2) and expression of GR, TR, CAT and NQO1 to confer cardioprotection against oxidative-stress-induced apoptosis; this effect is prevented by either si-Nrf1 or inhibitors of ERKs, p38 and PI3Ks [173
Nanoparticles (nPM; 60 nm) (5 h/day, 3 days/week for 10 weeks) Silencing of Nrf1 p120, p65, p30 and p23 increases expression of GCLC and GCLM in HBE1 cells, although the Nrf1-EpRE-binding activity is promoted in the lung of 12-month-old rather than 6-month-old mice [176
X-ray radiation (∼0.05–0.50 Gy) Increased ROS and Ca2+ levels cause nuclear accumulation of Nrf1 and Nrf2 to up-regulate MnSOD, HO-1 and NQO1 against paraquat cytotoxicity, along with phosphorylation of ERKs, but not JNK or p38, in skin HS27 cells. The activation of Nrf1/2 is inhibited by PD98058 [174
As3+ (10 μM), tBHQ (30 μM for ∼2–4 h) As3+-inducible Gclc and Gclm is not altered in Nrf1−/− or Nrf2−/− MEFs, whereas induction by tBHQ is blocked by loss of Nrf2 but not Nrf1 [182
As3+ (∼1.25–40 μM), tBHQ (∼25–50 μM), sulforaphane (SFN; 7.5 μM) A dose- and time-dependent increase in Nrf1 p120/p140 rather than its mRNA expression in As3+-treated HaCat cells is promoted further by Nrf2 KD, but are repressed by Keap1 KD. As3+-pre-conditioning is blocked by Keap1 KD, but not Nrf2 KD [109
TU (2 μg/ml), TG (2 μM), BFA (1 μg/ml) (i) Nuclear accumulation of Nrf1 (p120/p140) by As3+ is more than tBHQ or SFN. Expression of NQO1, GCLM and GCLC, as well as Keap1, but not Nrf2 proteins, is attenuated by Nrf1 KD, leading to decreased GSH levels and As3+-induced apoptosis. (ii) Relative to TG-unaffected mobility, TU and BFA give rise to faster or slower migration of Nrf1, enabling redistribution of As3+-inducible Nrf1 in the nucleus and cytoplasm. (iii) Basal and As3+-induced ARE reporter activity is repressed by ER stressors [108
tBHQ (∼25–50 μM), N-acetyl-L-leucyl-L-leucylnorleucinal (ALLN; 5 μg/ml), TG (1 μM), BFA (1 μM), TU (1 μg/ml) Induction of Nrf1-mediated ARE reporter genes by tBHQ is not affected by deletion of Keap1-binding ETGE and DLG motifs, but is suppressed by ALLN and ER stressors in COS-1 cells. [41
EGF (10 ng/ml), serum (10%), insulin (100 nM), torun-1 (250 nM), rapamycin (10 nM) mTOR [mammalian (or mechanistic) target of rapamycin)] inhibitors repress induction of PSM genes and Nrf1 rather than Nrf2 by insulin and growth factors, but not by bortezomib or MG132. Also, si-SREBP1 suppresses increased expression of Nrf1 in Tsc2−/− (tuberous sclerosis complex 2-null) MEFs. Both TU and TG fail to induce Nrf1 in wild-type cells [212
Dihydrotestosterone (DHT; 1–10 nM) DHT increases the nuclear translocation of Nrf1p65 in C4-2B cells, enabling it to interact with androgen receptor and thus enhance its DNA-binding activity and induce androgen receptor targets PSA (prostate-specific antigen) and TMPRSS2 (transmembrane protease serine 2) [64
α-MSH (1 μM), UVB (10 mJ/cm2Down-regulation of antioxidant genes encoding HO-1, γ-glutamylcysteine synthetase and GSTPi by UVB is prevented or overcompensated for by α-MSH in human skin [300
UVB (∼20–30 mJ/cm2), H2O2 (∼10–50 μM), GSH (∼1–2 mM) Nrf1 p120/p140 is lowered in human tumours and UVB-irradiated mouse skin. Up-regulation of Bik by Nrf1 KD sensitizes cells to apoptotic killing by UVB, and UV-induced oxidative DNA damage repair is impaired by Nrf1 KD to down-regulate XPC (xeroderma pigmentosum C) expression [181
Reagents (dose) Salient responses Reference(s) 
β-Naphthoflavone (β-NF; ∼10–50 μM), tBHQ (∼50–150 μM) Induction of Nrf1-mediated hARE-tk-CAT reporter genes (from NQO1, GCLC or GST-Ya) by β-NF is inhibited by c-Fos and Fra-1, whereas dual opposing effects of c-Jun depend on its extent to be expressed in transfected HepaG2 cells [81,87
Paraquat (∼0.4–1.6 mM), t-butylhydroperoxide (tBHP; 250 μM), TNFα, CdCl2 (∼50–200 μM) Compared with wild-type MEFs, loss of Nrf1 impairs antioxidant gene expression, leading to more sensitivity to cell killing by paraquat and CdCl2. Nrf1-deficent cells are also susceptible to apoptosis induced by tBHP and TNFα (∼2–50 ng/ml) [31,135
Menadione (∼20–100 μM), H2O2 (∼50–300 μM) Resistance of CFSCs (CCl4-induced cirrhotic fat-sorting cells in rat livers) to menadione toxicity is associated with nuclear accumulation of Nrf1 and p53, and induction of NQO1 via Nrf1-EpRE [179
HNE (15 μM) EpRE of the γ-glutamyltranspeptidase gene is induced to bind Nrf1, Nrf2, c-Fos, Fra-1, plus JunB, c-Jun, FosB and Fra-2 in untreated epithelial cells. Both inducible binding activity and expression of its mRNA are blocked by inhibitors of ERKs and p38, but not PKC [172
Genistein (∼25–50 μM), daidzein (∼25–50 μM), H2O2 (50 μM) Induction of glutathione peroxidase (GPx) and GCLC by genistein is mediated via Nrf1, but not Nrf2, to defend against oxidative damages to endothelial EA.hy926 cells [180
Bisphenol A (∼100–200 μM) Induction of Nrf1, Nrf2 and target antioxidant genes Ho-1, Nqo1 and Gclc is accompanied by increased GSH levels [175
Hypoxia (1% O2), CoCl2 (200 μM) Stabilization of Nrf1 p120, p110 and p65, but not p23, by MG132 is enhanced by hypoxia. Nrf1 target reporter is activated by CoCl2 [67
H2O2 (∼10–100 μM: pre-treated for 10 min before recovery for 24 h, then stress for 30 min) Pre-conditioning increases nuclear accumulation of Nrf1 (and Nrf2) and expression of GR, TR, CAT and NQO1 to confer cardioprotection against oxidative-stress-induced apoptosis; this effect is prevented by either si-Nrf1 or inhibitors of ERKs, p38 and PI3Ks [173
Nanoparticles (nPM; 60 nm) (5 h/day, 3 days/week for 10 weeks) Silencing of Nrf1 p120, p65, p30 and p23 increases expression of GCLC and GCLM in HBE1 cells, although the Nrf1-EpRE-binding activity is promoted in the lung of 12-month-old rather than 6-month-old mice [176
X-ray radiation (∼0.05–0.50 Gy) Increased ROS and Ca2+ levels cause nuclear accumulation of Nrf1 and Nrf2 to up-regulate MnSOD, HO-1 and NQO1 against paraquat cytotoxicity, along with phosphorylation of ERKs, but not JNK or p38, in skin HS27 cells. The activation of Nrf1/2 is inhibited by PD98058 [174
As3+ (10 μM), tBHQ (30 μM for ∼2–4 h) As3+-inducible Gclc and Gclm is not altered in Nrf1−/− or Nrf2−/− MEFs, whereas induction by tBHQ is blocked by loss of Nrf2 but not Nrf1 [182
As3+ (∼1.25–40 μM), tBHQ (∼25–50 μM), sulforaphane (SFN; 7.5 μM) A dose- and time-dependent increase in Nrf1 p120/p140 rather than its mRNA expression in As3+-treated HaCat cells is promoted further by Nrf2 KD, but are repressed by Keap1 KD. As3+-pre-conditioning is blocked by Keap1 KD, but not Nrf2 KD [109
TU (2 μg/ml), TG (2 μM), BFA (1 μg/ml) (i) Nuclear accumulation of Nrf1 (p120/p140) by As3+ is more than tBHQ or SFN. Expression of NQO1, GCLM and GCLC, as well as Keap1, but not Nrf2 proteins, is attenuated by Nrf1 KD, leading to decreased GSH levels and As3+-induced apoptosis. (ii) Relative to TG-unaffected mobility, TU and BFA give rise to faster or slower migration of Nrf1, enabling redistribution of As3+-inducible Nrf1 in the nucleus and cytoplasm. (iii) Basal and As3+-induced ARE reporter activity is repressed by ER stressors [108
tBHQ (∼25–50 μM), N-acetyl-L-leucyl-L-leucylnorleucinal (ALLN; 5 μg/ml), TG (1 μM), BFA (1 μM), TU (1 μg/ml) Induction of Nrf1-mediated ARE reporter genes by tBHQ is not affected by deletion of Keap1-binding ETGE and DLG motifs, but is suppressed by ALLN and ER stressors in COS-1 cells. [41
EGF (10 ng/ml), serum (10%), insulin (100 nM), torun-1 (250 nM), rapamycin (10 nM) mTOR [mammalian (or mechanistic) target of rapamycin)] inhibitors repress induction of PSM genes and Nrf1 rather than Nrf2 by insulin and growth factors, but not by bortezomib or MG132. Also, si-SREBP1 suppresses increased expression of Nrf1 in Tsc2−/− (tuberous sclerosis complex 2-null) MEFs. Both TU and TG fail to induce Nrf1 in wild-type cells [212
Dihydrotestosterone (DHT; 1–10 nM) DHT increases the nuclear translocation of Nrf1p65 in C4-2B cells, enabling it to interact with androgen receptor and thus enhance its DNA-binding activity and induce androgen receptor targets PSA (prostate-specific antigen) and TMPRSS2 (transmembrane protease serine 2) [64
α-MSH (1 μM), UVB (10 mJ/cm2Down-regulation of antioxidant genes encoding HO-1, γ-glutamylcysteine synthetase and GSTPi by UVB is prevented or overcompensated for by α-MSH in human skin [300
UVB (∼20–30 mJ/cm2), H2O2 (∼10–50 μM), GSH (∼1–2 mM) Nrf1 p120/p140 is lowered in human tumours and UVB-irradiated mouse skin. Up-regulation of Bik by Nrf1 KD sensitizes cells to apoptotic killing by UVB, and UV-induced oxidative DNA damage repair is impaired by Nrf1 KD to down-regulate XPC (xeroderma pigmentosum C) expression [181

Nrf1 is required for maintaining redox homoeostasis through constitutively regulating ARE-driven genes

Since Nrf1, like Nrf2, binds to an extended AP1-like NF-E2/ARE in the promoter region of Nqo1, as in those of PBGD and LCR of β-globin [78] (Figure 10A), it up-regulates ARE-mediated gene activity of the human Nqo1 and the rat Gstya subunit in HepG2 liver cells [81,87]. Although the basal expression of Nqo1 was not significantly decreased in Nrf1- or Nrf2-deficient cells [28,69,169], Nrf1:Nrf2 double-KO almost completely abolished Nqo1 expression [28]. The compound Nrf1−/−:Nrf2−/− mutant also abrogated the basal expression of Ho-1, Fth1, Mt-1 and Mt-2, even though they were expressed at considerably low levels in the individual Nrf1-deficient MEFs [28,135]. Furthermore, reduced expression of Ho-1, Gstm3, Gstm6 and Gstp2 was apparent in the hepatocytes of Nrf1−/− [32], and decreased expression of Nqo1 and Gsta4 was seen in the Nrf1−/− spinal cords [154]. These findings demonstrate that Nrf1 is required in mediating the basal expression of ARE-driven genes encoding antioxidant and detoxifying enzymes.

It is also noteworthy that Nrf1 is also necessary to regulate the basal expression of genes encoding GSH biosynthesis (i.e. Gclc, Gclm and Gss) and modulating enzymes (e.g. Gpx1). This conclusion is supported by the observation showing that expression of these genes was significantly decreased in MEFs from single Nrf1−/− or double Nrf1−/−:Nrf2−/− mice, but, conversely, increased GSH levels and enhanced expression of Gclc occurred in Nrf1-overexpressing cells [88,170]. Nrf1-deficient cells that have less activity of the antioxidant and detoxifying genes described above exhibited increased oxidative-stress-induced apoptosis and necrosis, with intracellular GSH at low levels accompanied by elevated oxidant levels (i.e. GSSG, H2O2 and singlet oxygen examined) [28,135]. The increased oxidative stress may be the cause of haemolytic anaemia leading to embryonic lethality of Nrf1−/− mice [28,135]. Severe oxidative stress caused massive hepatocytic death of fetal and adult livers, and was also involved in NASH and hepatoma [28,32]. Furthermore, progressive motor ataxia was determined to result from oxidative-stress-induced neuronal apoptosis in the hippocampus of CNS-specific Nrf1−/− mice [154]. From these findings, it is inferred that the basal expression of antioxidant and detoxifying enzymes mediated by Nrf1 is involved in setting up a threshold redox state under normal homoeostatic conditions, because this may be regarded as a prominent base of cellular defences against oxidative stress through the inactivation of endogenous oxidants (or pro-oxidants) and electrophiles derived from various metabolic processes, including fatty acids, phospholipds and DNA [171]. This conclusion is supported further by the evidence that a certain subset of Nrf2-dependent antioxidant genes were activated constitutively by endogenous oxidative stress caused by loss of Nrf1 function [33,154].

Nrf1-deficient cells were sensitive to the toxic effect of pro-oxidants such as paroquat, cadmium choride (CdCl2), diamide, tBHP (t-butylhydroperoxide) and TNFα (tumour necrosis factor α) [31,32,135], but the sensitivity is at a lesser degree than that of Nrf2-deficient cells [31]. This Nrf1:Nrf2 double-KO model revealed that both Nrf1 and Nrf2 are involved in regulating cellular redox status and they functionally compensate for the loss of each other's function, but not all ARE-containing target genes are mediated similarly by the two CNC–bZIP factors. For example, basal expression of Gclm, Mt-1 and Nqo1 was decreased to a similar low extent in Nrf1−/− or Nrf2−/− cells, and almost completely abolished in their double-KO cells [31]. Compared with wild-type cells, Nrf1-deficient cells prevented diethylmaleate-induced expression of Gclc, Gclm, Prx6 (peroxiredoxin 6) and Ho-1 to lower levels, but the efficient inducible expression appeared to be sufficient to be abolished by Nrf2 deficiency. These observations suggest that Nrf1 and Nrf2 have important differential functions in regulating the basal and inducible expression of ARE-battery genes, of which the basal expression appears to be predominantly mediated by Nrf1 whereas Nrf2 is a key regulator of inducible ARE-gene expression.

Activation of Nrf1 in a way similar to Nrf2 in antioxidant cytoprotection against distinct stresses

Induction of GGT (γ-glutamyltranspeptidase), which plays a critical role in GSH homoeostasis and metabolism, by the lipid peroxidation product HNE (4-hydroxynonenal) is mediated by an EpRE/ARE-bound complex, which contains Nrf1, Nrf2, c-Fos and Fra-1, together with JunB, c-Jun, FosB and Fra-2, but only the last four named factors were detectable in untreated epithelial cells [172], suggesting the involvement of Nrf1, like Nrf2, in the antioxidant response to HNE. The signalling to induce the Ggt gene was triggered by HNE dependent on activation of MAPKs, including ERK (extracellular-signal-regulated kinase) and p38 kinase, but neither PKC nor PI3K (phosphoinositide 3-kinase) [172]. Similar nuclear accumulation of Nrf1 and Nrf2 was also enhanced by pre-conditioning with H2O2 to induce expression of GR, TR, CAT (catalase) and Nqo1, which was involved in cardioprotection against oxidative-stress-induced apoptosis; this effect was prevented by si-Nrf1 or inhibitors of ERKs, p38 and PI3Ks [173]. In addition, the ERK inhibitor PD98058 blocked activation of Nrf1 and Nrf2 by X-ray radiation that increased both H2O2 and Ca2+ levels, leading to the nuclear accumulation of both CNC–bZIP factors and ensuing up-regulation of cognate cytoprotective genes MnSOD (manganese superoxide dismutase), Ho-1 and Nqo1 against paraquat cytotoxicity in skin HS27 cells; this signalling network may be triggered via phosphorylation of ERKs, but neither JNK (c-Jun N-terminal kinase) nor p38 kinase [174].

Similar Nrf1–and/or Nrf2–ARE pathways were also activated by bisphenol A so as to promote gene expression of antioxidant and detoxification enzymes (i.e. those encoded by Ho-1, Nqo1 and Gclc), resulting in increased GSH levels to protect HEK (human embryonic kidney)-293 cells from oxidative stress [175]. Another study revealed that Nrf1 was not directly responsible for the loss of Nrf2-dependent inducibility of antioxidant cytoprotective genes because of increased expression of Gclc and Gclm in the HBE1 (human bronchial epithelial) cells that had been transfected with si-Nrf1 to knock down the abundance of its targeting p120, p65, p30 and p23; all isoforms remained unaffected in the lungs of mice in response to nPM (nano-particular matter), although the overall binding activity of Nrf1-EpRE/ARE was promoted in the lung of 12-month-old rather than 6-month-old mice [176].

Activation of Nrf1 differs from that for Nrf2 in antioxidant defences against distinct stresses

Recently, functional network and gene ontology analyses revealed that Nrf1 and Nrf2 are distinctively involved in copper-responsive regulation of the HepG2 transcriptome [177]. In fact, vitamin C was identified to significantly induce expression of the transcription factor Osx through Nrf1, but not Nrf2, binding to its ARE sequence in the promoter region, leading to normal osteoblast differentiation [158]. During rat liver injury, similar Nrf1-ARE signalling pathway could also enable resistance of CFSCs (CCl4-induced cirrhotic fat-sorting cells) to the cytotoxicity of menadione (which can be reduced by NQO1 and NADPH cytochrome c reductase to form hydroquinone and semiquinone radicals respectively, and in turn the latter radical either reacts with cellular constituents or recycles back to menadione through a rapid reaction with molecular oxygen [178]). In fact, Nrf1 was determined to exist only in CFSCs, but not in NFSCs (normal fat-sorting cells), indicating that Nrf1, but not Nrf2, is involved in the adaptive resistance primarily through transactivating EpRE/ARE-driven genes (e.g. Nqo1) [179]. Together with the observation that the nuclear accumulation of Nrf1 was enhanced by p53 stabilization, Nrf1-medated resistance of the hepatic stellate cells against oxidative damage is suggested to be part of the adaptive activation process to increases in cell proliferation and production of extracellular matrix. The assumption is also supported by an investigation into cytoprotective effects of isoflavones against oxidative damages to endothelial EA.hy926 cells, revealing that genistein, but not daidzein, protected cells predominantly via Nrf1- rather than Nrf2-mediated activation of Gpx and Gclc [180].

The stabilization of Nrf1 p120, p95 and p65, but not p23, by the proteasomal inhibitor MG132 was enhanced by hypoxia such that activation of Nrf1 target reporter was also augmented by the hypoxia-inducer CoCl2 [67]. Interestingly, human Nrf1/TCF11 (p120/p140) expression in either human tumours or UVB-irradiated mouse skin were detected at lower levels than those from their counterpart controls [181]. KD of Nrf1 was allowed for up-regulation of Bik (Bcl2-interacting killer) that sensitizes cells to apoptotic killing by UVB, but the UV-derived oxidative DNA damage repair was impaired due to down-regulation of Xpc (xeroderma pigmentosum C) expression [181].

As3+ caused a dose- and time-dependent increase in human Nrf1/TCF11 (p120/p140), rather than its mRNA, expression in HaCat cells [109]. The As3+-induced nuclear accumulation of Nrf1/TCF11 is more than that induced by tBHQ (t-butylhydroquinone) or SFN (sulforaphane) [108]. More interestingly, As3+-inducible expression of Nrf1/TCF11 was enhanced further by Nrf2 KD, but was repressed by Keap1 KD, and also As3+-pre-conditioning stress response is blocked by Keap1 KD, but not Nrf2 KD [109]. Conversely, the expression of Keap1, but not Nrf2, was attenuated by Nrf1 KD, as accompanied by abolishment of Nqo1, Gclm and Gclc expression, leading to decreased GSH levels and increased As3+-induced apoptosis [108]. These findings suggest that a potential interaction exists between Keap1 and Nrf1 that is critical for maintaining their protein stability in response to As3+-induced stress, albeit it warrants further studies. However, on the other hand, neither Nrf1 nor Nrf2 was reported to be involved in As3+-inducible expression of Gclc and Gclm in MEFs [182], because the two gene induction by As3+ was unaffected by either Nrf1−/− or Nrf2−/−, although their expression induced by tBHQ was blocked by loss of Nrf2 but not Nrf1. In addition, induction of hepatocytic Gclc expression by anthocyanin cyaniding-3-O-β-glucoside, a flavonoid that can defend against H2O2 (examined as a representative of oxidants) during hyperglycaemia, was also revealed to be independent of Nrf1 and Nrf2, but not CREB [183].

Distinct roles for Nrf1 and Nrf2 in regulating ARE/EpRE-driven cytoprotection from pathophysiology

In an attempt to predict ARE/EpRE-regulated genes on the basis of the initially proposed core sequence [184] (Table 1), only one of several ARE (and ARE-like) batteries existing in each of the responsive genes such as Gclc, Gclm or Nqo2 [185187] was identified to be inducible through Nrf2-mediated pathways. The in-depth insights into what allows a certain ARE sequence to be functional whereas others are not [82,188] have revealed that differential effects of the nucleotide variables within the core ARE and flanking sequences on basal and inducible activity (i.e. the sequence extended to 5′-TMnnnRTGABnCnGCRnWWW-3′, in which B=C, G or T; M=A or C; R=A or G; and W=A or T) are of more importance than those predicted previously, but the functionality appears to be unaffected by its genomic location. As for instances of the human Gclc, just one (i.e ARE4) of 13 putative AREs/EpREs in the promoter region was induced by Nrf2 in response to HNE [185,188,189]. However, it cannot be excluded that Nrf1 may be invoved in regulating expression of Gclc driven by ARE4, because the functional battery sequence (5′-cgTGACTCAGCgc-3′; Table 1) encloses a consensus AP1-binding site, as reviewed previously [9]. This assumption is also supported by evidence showing that, besides ARE4, additional four cis-regulatory batteries {i.e. ARE1, ARE3, n7 and n8, each of which contains either an AP1/TRE [TPA (PMA)-response element] en bloc or AP1/TRE-like element} exhibited constitutive activity to contribute to the basal expression of Gclc, even though they were not responsive to HNE induction of Nrf2 [188]. Contribution of the AP1/TRE-like element to the constitutive expression of ARE/EpRE-driven genes is validated by altered responsiveness of relevant deletion mutants to HNE or PMA [190]. These findings have demonstrated that a putative complex of c-Jun with Nrf1 (and/or Nrf2) ensures only a bona fide (but non-enforced) commitment to basic transcriptional expression of ARE-driven genes, whereas the conversion of c-Jun by phosphorylation required for transactivation of an oncogenic AP1 dimer allows it to exert somewhat inhibitory effects depending on distinct contexts of ARE/EpRE sequences [191,192]. Collectively, it is plausible that the nuances of the internal nucleotides within the discrete ARE/EpRE core and flanking sequences can enable differential recruitment of distinct heterodimers of Nrf1 or Nrf2 with sMaf or other bZIP proteins to bind discrete target genes, leading to a discrepancy in transactivation responsible for antioxidant defences.

More notably, Nrf2 is well documented as a master regulator of adaptive responses to oxidative stressors and electrophiles [9,25], but it appears to be dispensable for the basal expression of many (but not all) ARE/EpRE-driven cytoprotective genes against carcinogenesis and other pathologenesis. This is supported by the fact that Nrf2−/− mice normally develop with neither spontaneous cancer nor other phenotypes [26], albeit they are more susceptible than wild-type mice to chemical carcinogens [27], and hence induction of Nrf2 has been considered as a chemopreventive target [9,167,193]. The Nrf2-inducible antioxidant, detoxification and cytoprotective effects against carcinogenesis appeared to be enhanced by forced expression of a constitutively dominant-active Nrf2 in transgenic mice [194], but it is of significant importance to note that the basal expression of ARE-driven genes, rather than their inducible expression, is crucial for the intrinsic anti-tumour prevention conferred on the host against DMBA [7,12-dimethylbenz(a)anthracene]/PMA-induced carcinogenesis in transgenic mice expressing a dominant-negative Nrf2 (that also inhibits other Nrf/CNC factors (e.g.Nrf1) [195]. Conversely, permanently hyperactive induction of Nrf2-mediated ARE-driven genes allows depravation of the factor as an unrecognized mediator of oncogenesis insofar as to promote cancer cell survival and tumourigenesis [196198]. The opposing potentials of Nrf2–ARE/EpRE signalling in tumour prevention and progression should thus be taken into account, as described elsewhere [199,200].

By contrast, KO of Nrf1 by distinct gene-targeting strategies in mice leads to embryonic lethality and other significant pathophysiological phenotypes resulting from severe oxidative stress [2933,138]. Further assays have revealed that loss of Nrf1’s function cannot be compensated for by the presence of Nrf2, albeit the latter CNC–bZIP factor possesses certain overlapping functions in regulating ARE-driven gene expression as confirmed by the double-KO (Nrf1−/−:Nrf2−/−) animal model [28]. Tissue-specific loss of Nrf1 in mouse liver, pancreas, brain and bone results in distinct pathologies of NASH and hepatoma [32,33], Type 2 diabetes [201], neurodegeneration [147,154] and reduced bone formation [162]; these remarkable phenotypic changes are accompanied by significant disorders of glucose, lipid and protein metabolism, in addition to severe endogenous oxidative stress. Together, these findings demonstrate that Nrf1 (and its isoforms) fulfils a unique and indispensable biological functions that is distinctive from that of Nrf2, in maintaining cellular redox and metabolic homoeostasis to certain steady-state levels during normal development and growth. However, there remains one great unknown about which isoforms (e.g. Nrf1, Nrf1β/LCR-F1, Nrf1γ and Nrf1δ; Figures 3 and 6) contribute to its unique pathobiological functions, in particular ARE/EpRE-driven antioxidant, detoxification and cytoprotection against oxidative-stress-induced carcinogenesis, because none of the single isoform-specific animal models are available to date.

DISTINCTIVE CONTRIBUTIONS OF Nrf1 AND Nrf2 TO MAINTAINING ER HOMOEOSTASIS

There exists a cross-talk between ER-derived oxidative stress response and UPR (unfolded protein response), as reviewed previously [202,203]. Loss of Nrf1 down-regulates antioxidant, detoxification and proteasome genes, leading to severe oxidative stress and disruption of protein folding within the ER and ERAD functioning so that unfolded and misfolded proteins are accumulated in the organelle [32,137]. The resulting pathophysiological phenomenon, called ER stress [204206], activates the UPR signalling pathways (Figure 8), giving rise to an adaptive remodelling of the ER that diminishes nascent polypeptide loading, removes aberrant folded proteins and restores its biological function [207,208]. Activation of UPR occurs principally through three different signal pathways involving PERK, IRE1 (inositol-requiring enzyme 1) and ATF6, each of which possesses a transmembrane domain with a similar topology (Figure 8). Thus similar membrane-bound Nrf1 (and Nrf3) should theoretically have a strong capacity of being induced by ER stress, as described for the C. elegans Skn-1 [123,126,209]. However, Nrf1 is de facto not activated via each of the UPR signals and, conversely, activation of Nrf1 by either tBHQ or As3+ to up-regulate ARE-driven antioxidant and detoxification genes was repressed by classic ER stressors, such as TU (tunicamycin), TG (thapsigargin) and BFA (brefeldin A) [41,108], although the electrophoretic mobility of Nrf1 and its subcellular redistribution were altered by TU and BFA, but not TG [108], suggesting that Nrf1 is modified post-translationally (i.e. glycosylation, deglycosylation, ubiquitination and proteolysis). By contrast, it is intriguing that the water-soluble Nrf2 was identified as a direct target of PERK triggered by TU so as to activate antioxidant and detoxification responses [210,211].

It is important to note a unique UPR-independent mechanism whereby the Nrf1 transactivation activity to up-regulate PSM genes is induced by low doses of its inhibitors [36,212], albeit proteasomal inhibition can be allowed for accumulation of oxidative ubiquitinated proteins triggering ER stress [213,214]. In fact, ER-stress-inducible UPR signalling pathways were endogenously activated by loss of Nrf1 function in the homozygous hepatocytes, and both similar ER stress response and steatosis were also enhanced by half loss of Nrf1 in the heterozygous livers, when compared with wild-type livers, in response to proteasomal inhibition [137]. It is therefore postulated that Nrf1 plays an essential cytoprotective role for maintaining ER homoeostasis, whereas its loss of function caused the ER transformation and proliferation of Nrf1−/− cells from conditional KO mice with spontaneous cancer [32]. This should be viewed as a consequence of chronic ER stress and prolonged activation of UPR signalling, coupled with severe oxidative stress, which together lead ultimately to carcinogenesis [215217].

The UPR signalling was reported to link to a lipid-stress-response pathway mediated by the ER-membrane-bound SREBP (sterol-response-element-binding protein) [218220]. Activation of SREBP signalling occurs only after being translocated into the Golgi apparatus where it is sequentially cleaved by both Site-1 and Site-2 proteases in a similar fashion to the case of ATF6. However, Nrf1 is neither translocated into the Golgi apparatus nor cleaved by Site-1 and Site-2 proteases [40,58]. In spite of the recent identification of SREBP1 as a direct upstream regulator of Nrf1 [212], which is involved in the mTORC1 [mammalian (or mechanistic) target of rapamycin complex 1] signalling response to insulin and EGF (epidermal growth factor) (Figure 9), it remains unknown whether there exists a co-ordinated cross-talk between the oxidative stress response and the lipid-coupled UPR. However, it is possible that Nrf1 senses redox changes in the vicinity of the ER, in which relevant signals are appropriately integrated and transduced to activate target genes; the ER signalling to regulate cytoprotective responses against stress enables Nrf1 to be selectively processed via a unique mechanism distinctive from that which accounts for ATF6 and SREBP1.

Nrf1 CONTRIBUTES TO PROTEOSTASIS THROUGH CONTROLLING PROTEASOMAL SUBUNIT EXPRESSION

As the proteome of a living organism faces a variety of challenges to their organ integrity, proteostasis (i.e. protein homoeostasis) is maintained by multilayered regulatory networks governing the steady-state balance of protein synthesis, folding and degradation, hence restoring cellular physiological homoeostasis to protect healthy development and growth against stress or disease [221223]. Within such networks, antioxidant transcription factors CncC, Skn-1 and Nrf2 appear to be involved in maintaining eukaryotic proteostasis through controlling expression of target genes involved in protein folding, UPR and proteasomal degradation [124,224228]. By contrast, both Nrf1 and its long TCF11 contribute primarily to bidirectional regulation of proteasomal subunit genes [36,166,203,212]. Nonetheless, it should be noted that these conserved CNC–bZIP family proteins are absent from yeast, such that expression of their PSM genes is mediated by the transcription factor RPN4, which is also required for feedback regulation of proteasome activity [229,230].

Nrf1/TCF11 controls transcriptional expression of genes encoding proteasomal subunits and p97/VCP

Chemical inhibition of proteasome also leads to up-regulation of mRNA levels of proteasomal subunits followed by increased formation of proteasomes (Figures 6 and 7); this was termed as the ‘bounce-back’ response [231233]. The proteasome recovery pathway is predominantly mediated by the membrane-bound CNC–NHB1 subfamily factors Nrf1/TCF11, Skn-1 and CncC in different metazoan species; they enable transactivation of PSM genes to increase their de novo synthesis and functional activity [34,35,124,226]. Notably, the bounce-back response to lower doses of proteasomal inhibitors (Table 4) has been determined to be mediated dominantly by Nrf1 rather than Nrf2 through binding to ARE sequences in the PSM promoter regions, because proteasome recovery is abolished by Nrf1 deficiency (Nrf1−/− and sh-Nrf1) but is unaffected by KO of Nrf2−/− [35]. Similarly, overexpression of TCF11, but not of Nrf2, causes ∼2-fold increases in endogenous expression of proteasome subunits and its de novo formation in transfected cells; in turn, silencing of TCF11 blocks inducible expression of 46 PSM-related genes and reduces stimulatory biosynthesis of proteasomes upon treatment of cells with its inhibitor expoxomicin [34]. Moreover, KD of Nrf1 enhances sensitivity of cells to apoptotic killing by the proteasomal inhibitor YU101 via caspase 3; this effect is blunted by the pan inhibitor Z-VAD-FMK (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) [35]. Collectively, it is postulated that Nrf1/TCF11-mediated proteasome recovery contributes to cancer chemoresistance to proteasomal inhibitors.

Table 4
Interdependence of Nrf1 with p97 and 26S proteasomes on the extent of inhibition
Reagents (dose) Salient responses and effects Reference(s) 
MG132 (1 μM), YU101 (∼0.15–0.25 μM), MLN4925 (5 μM), bortezomib (0.5 μM), benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone (Z-VAD-FMK; 50 μM) (i) Partial inhibition of proteasome activity up-regulates seven PSM genes in human LNCap and HT29 cells. The bound-back response is mediated by Nrf1, but not Nrf2, because it is abolished in Nrf1-KO MEFs. The recovery occurs via the Nrf1-mediated proteasome synthesis pathway. (ii) si-Nrf1 sensitizes MDA-MB-231 and U2OS cells to killing by YU101 via caspase 3; this is blunted by its inhibitor Z-VAD-FMK [35
MG132 (0.1 μM), epoxomicin (10 nM), BSc2118 (0.1 μM), lactacystin (1 μM), cotransin (2 μM) (i) Endogenous proteasome expression and its de novo formation are enhanced 2-fold in TCF11-transfected Ea.hy926 cells. (ii) si-TCF11 abolishes induction of 46 proteasome-related and ten oxidative defence genes by expoxomicin and its de novo formation. (iii) Oxidative proteotoxic stress promotes ER-to-nuclear translocation of TCF11 as a faster-migrating form to bind PSM genes; this process is blocked by si-Sec 61 and cotransin. (iv) Glycated and deglycated TCF11 proteins are stabilized by si-p97, but not si-Keap1. (v) Its stability is enhanced by si-Hrd1 and Hrd1-C329S, but not by si-gp78 or si-TEB4. Ubiquitination of TCF11 is inhibited by Hrd1-C329S [34
Bortezomib (2.5 mg/kg), sulforaphane (SFN; 5 mg/kg), 4-phenylbutyrate (120 mg/kg) (i) Basal and bortezomib-induced expression of PSM genes and Pomp is abolished by KO of Nrf1, but not of Nrf2, in liver. (ii) Induction of PSM genes, but not Gclc or Nqo1 by sulforaphane is blunted by Nrf1 KO, accumulating ubiquitin proteins. (iii) ER-stress-response genes (Atf4, Chop, Bip and Gadd45β) are up-regulated, with phosphorylation of PERK and elF2α in Nrf1-KO liver. (iv) Bortezomib induces steatosis in Nrf1+/− liver possibly via ER stress signalling genes (e.g. Atf6, Bip, Chop and Herp[137
MG132 (∼0.25–3 μM), bortezomib (10 nM), lactacystin (∼2–3 μM) (i) Dose- and time-dependent increases in Nrf1, Nrf2 and ubiquitin proteins are caused by MG132, which accumulates cells in G2/M-phase and promotes apoptosis via p21 and caspase 9. (ii) MG132-inducible, rather than basal, expression of PSM genes is eliminated by si-Nrf1, with partial inhibition of some PSM genes by si-Nrf2 in SCC-13, A431, HaCat, but not in normal, cell lines. (iii) Reduction of Bmi-1 and Ezh2 by MG132-stimulated PSM genes is partially recovered by si-Nrf1 but not si-Nrf2 [235
MG132 (10 μM), SB216763 (∼2–5 μM), LiCl (∼2–5 mM) (i) Nrf1 binds Fbw7α via a Cdc4 phosphodegron L348FSPEVE354 or L267LSPLLT273, targeting for SCF–ubiquitin–proteasome degradation. (ii) Nrf1 stability is enhanced by GSK-3 KD, and its inhibitors LiCl and SB216763 in HEK (human embryonic kidney)-293 cells. Phosphorylation of Nrf1 at Ser350 by GSK-3β is involved in Fbw7-mediated ubiquitin degradation [107,116
MG132 (10 μM) (i) Nrf1 is stabilized by MG132, si-β-TrCP1/2 but not si-Keap1. and DSGLS degron in Nrf1 is required for binding β-TrCP2, whereas amino acids 31–81 of Nrf1 is for degradation via Hrd1 and p97, but not gp78 or TEB4. (ii) si-p97 stabilizes 95-kDa and 120-kDa Nrf1; si-Hrd1 stabilizes 120-kDa, but not p95-kDa, Nrf1 in HeLa cells [106
MG132 (1 μM), epoxomicin (10 nM) (i) MG132-inducible expression of almost all of the 35 proteasome-related genes (e.g. PA200, Usp14 and Pomp) is blocked by si-Nrf1, with little impact on autophagy genes (Hsp70, p62 and Bag3), in HeLa cells. (ii) Protein kinase CK2 phosphorylates Nrf1 at Ser497 and represses gene transcription; this effect is blunted by Nrf1-S497A, with no change in its protein stability. (iii) Proteasome inhibitors cause p62-positive inclusion bodies; this is enhanced by si-Nrf1, with an increase in LC3-II rather than p62 [94
MG132 (10 μM), lactacystin (5 μM), staurosporine (15 nM), okadaic acid (30 nM) (i) MG132 increases ubiquitination of endogenous Nrf1 and decreases its activity. Its p120 and p65, but not p95, are stabilized, whereas p23 (blotted by antibodies against amino acids 191–475 of Nrf1) is destabilized. (ii) Nrf1-mediated reporter activity is promoted by okadaic acid, but is repressed by staurosporine [67
MG132 (∼1–5 μM), NMS-873 (10 μM), doxycycline (1 μg/ml) (i) MG132-stabilized p120 Nrf1 may be processed to yield an active p110 in HEK-293 cells; this is inhibited by mutating W103L104 to alanine (within V100NAWLVHRD108). (ii) Elimination of p110 by si-p97 and NMS-873 is attributable to blocking p120 retrotranslocation from the ER luminal side of membranes into the cyto/nucleo-plasmic side [127
MG132 (∼1–10 μM), MG262 (∼1–10 μM), bortezomib (0.01–10 μM), epoxomicin (∼0.05–10 μM), ML00603997 (0.5 μM) (i) Proteasomes are up-regulated by bortezomib, but not by ER stressors, with a decrease in the immunoproteasome. Pomp, Usp14, PA200, UBB, p97 and co-factors Npl4, Ufd1 and p47 are induced, with unaffected Rad23A/B. (ii) Induction of proteasomes and p97 by low doses of inhibitors is blocked by high doses (which completely inhibit its activity and trigger p-elF2α). (iii) Full inhibition and depletion of ATP repress generation of cleaved 75-kDa Nrf1, but elevates its full-length protein. (iv) During partial inhibition, other MG132-sensitive or ER-associated proteases were thought to be not involved in the processing of Nrf1, but the processing is co-ordinated with p97-driven dislocation, deglycosylation and ubiquitination of Nrf1 [36
MG132 (∼0.3–10 μM), calpain inhibitor I/N-acetyl-L-leucyl-L-leucylnorleucinal (CI/ALLN; ∼0.6–10 μg/ml), calpain inhibitor II (CII; 0.6–10 μg/ml), calpeptin (∼0.6–10 μg/ml) (i) Directional regulation of Nrf1 target proteasome reporter genes by CI/ALLN and MG132 depends on their doses being administrated. High doses block the processing of Nrf1 to yield cleaved 85 kDa, but small dominant-negative forms are enhanced. (ii) Calpain inhibitors also induce Nrf1 target proteasome reporter genes [59
Reagents (dose) Salient responses and effects Reference(s) 
MG132 (1 μM), YU101 (∼0.15–0.25 μM), MLN4925 (5 μM), bortezomib (0.5 μM), benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone (Z-VAD-FMK; 50 μM) (i) Partial inhibition of proteasome activity up-regulates seven PSM genes in human LNCap and HT29 cells. The bound-back response is mediated by Nrf1, but not Nrf2, because it is abolished in Nrf1-KO MEFs. The recovery occurs via the Nrf1-mediated proteasome synthesis pathway. (ii) si-Nrf1 sensitizes MDA-MB-231 and U2OS cells to killing by YU101 via caspase 3; this is blunted by its inhibitor Z-VAD-FMK [35
MG132 (0.1 μM), epoxomicin (10 nM), BSc2118 (0.1 μM), lactacystin (1 μM), cotransin (2 μM) (i) Endogenous proteasome expression and its de novo formation are enhanced 2-fold in TCF11-transfected Ea.hy926 cells. (ii) si-TCF11 abolishes induction of 46 proteasome-related and ten oxidative defence genes by expoxomicin and its de novo formation. (iii) Oxidative proteotoxic stress promotes ER-to-nuclear translocation of TCF11 as a faster-migrating form to bind PSM genes; this process is blocked by si-Sec 61 and cotransin. (iv) Glycated and deglycated TCF11 proteins are stabilized by si-p97, but not si-Keap1. (v) Its stability is enhanced by si-Hrd1 and Hrd1-C329S, but not by si-gp78 or si-TEB4. Ubiquitination of TCF11 is inhibited by Hrd1-C329S [34
Bortezomib (2.5 mg/kg), sulforaphane (SFN; 5 mg/kg), 4-phenylbutyrate (120 mg/kg) (i) Basal and bortezomib-induced expression of PSM genes and Pomp is abolished by KO of Nrf1, but not of Nrf2, in liver. (ii) Induction of PSM genes, but not Gclc or Nqo1 by sulforaphane is blunted by Nrf1 KO, accumulating ubiquitin proteins. (iii) ER-stress-response genes (Atf4, Chop, Bip and Gadd45β) are up-regulated, with phosphorylation of PERK and elF2α in Nrf1-KO liver. (iv) Bortezomib induces steatosis in Nrf1+/− liver possibly via ER stress signalling genes (e.g. Atf6, Bip, Chop and Herp[137
MG132 (∼0.25–3 μM), bortezomib (10 nM), lactacystin (∼2–3 μM) (i) Dose- and time-dependent increases in Nrf1, Nrf2 and ubiquitin proteins are caused by MG132, which accumulates cells in G2/M-phase and promotes apoptosis via p21 and caspase 9. (ii) MG132-inducible, rather than basal, expression of PSM genes is eliminated by si-Nrf1, with partial inhibition of some PSM genes by si-Nrf2 in SCC-13, A431, HaCat, but not in normal, cell lines. (iii) Reduction of Bmi-1 and Ezh2 by MG132-stimulated PSM genes is partially recovered by si-Nrf1 but not si-Nrf2 [235
MG132 (10 μM), SB216763 (∼2–5 μM), LiCl (∼2–5 mM) (i) Nrf1 binds Fbw7α via a Cdc4 phosphodegron L348FSPEVE354 or L267LSPLLT273, targeting for SCF–ubiquitin–proteasome degradation. (ii) Nrf1 stability is enhanced by GSK-3 KD, and its inhibitors LiCl and SB216763 in HEK (human embryonic kidney)-293 cells. Phosphorylation of Nrf1 at Ser350 by GSK-3β is involved in Fbw7-mediated ubiquitin degradation [107,116
MG132 (10 μM) (i) Nrf1 is stabilized by MG132, si-β-TrCP1/2 but not si-Keap1. and DSGLS degron in Nrf1 is required for binding β-TrCP2, whereas amino acids 31–81 of Nrf1 is for degradation via Hrd1 and p97, but not gp78 or TEB4. (ii) si-p97 stabilizes 95-kDa and 120-kDa Nrf1; si-Hrd1 stabilizes 120-kDa, but not p95-kDa, Nrf1 in HeLa cells [106
MG132 (1 μM), epoxomicin (10 nM) (i) MG132-inducible expression of almost all of the 35 proteasome-related genes (e.g. PA200, Usp14 and Pomp) is blocked by si-Nrf1, with little impact on autophagy genes (Hsp70, p62 and Bag3), in HeLa cells. (ii) Protein kinase CK2 phosphorylates Nrf1 at Ser497 and represses gene transcription; this effect is blunted by Nrf1-S497A, with no change in its protein stability. (iii) Proteasome inhibitors cause p62-positive inclusion bodies; this is enhanced by si-Nrf1, with an increase in LC3-II rather than p62 [94
MG132 (10 μM), lactacystin (5 μM), staurosporine (15 nM), okadaic acid (30 nM) (i) MG132 increases ubiquitination of endogenous Nrf1 and decreases its activity. Its p120 and p65, but not p95, are stabilized, whereas p23 (blotted by antibodies against amino acids 191–475 of Nrf1) is destabilized. (ii) Nrf1-mediated reporter activity is promoted by okadaic acid, but is repressed by staurosporine [67
MG132 (∼1–5 μM), NMS-873 (10 μM), doxycycline (1 μg/ml) (i) MG132-stabilized p120 Nrf1 may be processed to yield an active p110 in HEK-293 cells; this is inhibited by mutating W103L104 to alanine (within V100NAWLVHRD108). (ii) Elimination of p110 by si-p97 and NMS-873 is attributable to blocking p120 retrotranslocation from the ER luminal side of membranes into the cyto/nucleo-plasmic side [127
MG132 (∼1–10 μM), MG262 (∼1–10 μM), bortezomib (0.01–10 μM), epoxomicin (∼0.05–10 μM), ML00603997 (0.5 μM) (i) Proteasomes are up-regulated by bortezomib, but not by ER stressors, with a decrease in the immunoproteasome. Pomp, Usp14, PA200, UBB, p97 and co-factors Npl4, Ufd1 and p47 are induced, with unaffected Rad23A/B. (ii) Induction of proteasomes and p97 by low doses of inhibitors is blocked by high doses (which completely inhibit its activity and trigger p-elF2α). (iii) Full inhibition and depletion of ATP repress generation of cleaved 75-kDa Nrf1, but elevates its full-length protein. (iv) During partial inhibition, other MG132-sensitive or ER-associated proteases were thought to be not involved in the processing of Nrf1, but the processing is co-ordinated with p97-driven dislocation, deglycosylation and ubiquitination of Nrf1 [36
MG132 (∼0.3–10 μM), calpain inhibitor I/N-acetyl-L-leucyl-L-leucylnorleucinal (CI/ALLN; ∼0.6–10 μg/ml), calpain inhibitor II (CII; 0.6–10 μg/ml), calpeptin (∼0.6–10 μg/ml) (i) Directional regulation of Nrf1 target proteasome reporter genes by CI/ALLN and MG132 depends on their doses being administrated. High doses block the processing of Nrf1 to yield cleaved 85 kDa, but small dominant-negative forms are enhanced. (ii) Calpain inhibitors also induce Nrf1 target proteasome reporter genes [59

Re-examination of liver-specific Nrf1−/− mice that develop NASH and hepatoma revealed that the pathology is attributed to accumulation of ubiquitinated proteins, which results from abolishment of basal and inducible expression of PSM genes and Pomp (proteasome maturation protein) by KO of Nrf1 but not Nrf2 [137]. Microarray analysis of transcriptional profiling of livers from the hepatocyte-specific Nrf1−/− mice demonstrates constitutive decreases in expression of the PSM genes under normal conditions [234]. Consequently, the failure of homozygous Nrf1−/− liver to dispose of ubiquitinated and/or unfolded proteins leads to up-regulation of UPR genes encoding ATF4, CHOP (C/EBP-homologous protein), BiP (immunoglobulin heavy-chain-binding protein) and GADD45β (growth-arrest and DNA-damage-inducible protein 45β), accompanied by phosphorylation of PERK and eIF2α. Further study showed that heterozygous Nrf1+/− mice are also susceptible to the proteasomal inhibitor bortezomib-induced steatosis possibly through activation of the ER signalling pathway [e.g. ATF6, BiP, CHOP and HERP (homocysteine-induced ER protein)] [137]. The same group has also presented evidence showing that dysregulation of PSM expression contributes to neurodegeneration in the brain-specific Nrf1−/− mice [147].

Silencing of Nrf1 eliminates MG132-inducible, rather than basal, expression of PSM genes; in contrast, some of the PSM genes are down-regulated by silencing of Nrf2 in cancer cell lines (e.g. SCC-13, A431 and HaCat), but not in normal cell lines [235]. Similarly, MG132-inducible expression of 35 PSM-related genes [including PA200 (proteasome activator 200 kDa), Usp14 (ubiquitin-specific peptidase 14) and Pomp] is also blocked by Nrf1 KD in HeLa cells [94]. Silencing of Nrf1, but not Nrf2, causes a partial recovery from MG132-stimulated proteasomal degradation of Bmi-1 and Ezh2; loss of the two Polycomb group proteins is associated with reduced cell proliferation, accumulation of cells that are arrested in G2/M-phase, and increased apoptosis via p21 and caspase 9 [235]. Interestingly, formation of MG132-inducible p62-positive inclusion bodies is enhanced by KD of Nrf1, with an increase in the autophagy marker LC3-II (long chain 3 II), whereas expression of autophagy genes [e.g. Hsp70 (heat-shock protein 70), p62 and Bag3 (Bcl2-associated athanogene 3)] appears to be unaffected [94].

A convincing study of the ‘bounce-back’ response to bortezomib demonstrates that it allows induction of almost all PSM-related genes [including Pomp, Usp14, PA200 and UBB (ubiquitin B)] and those encoding p97/VCP and co-factors (i.e. Npl4, Ufd1 and p47) as an adaptive recovery response to limited inhibition of the proteasome [36], with an exceptional decrease in the immunoproteasomal complex. They found that the proteasome activity to degrade ubiquitinated proteins is partially or completely suppressed by its inhibitor, particularly at high concentrations, such that UPR signalling is triggered via eIF2α phosphorylation [36,137,235]. However, ER stress signalling is not required for induction of PSM genes by proteasomal inhibitors [36,212].

Nrf1/TCF11 is positively and negatively regulated by proteasomal inhibitors

A bidirectional regulatory feedback circuit existing between Nrf1 and the proteasome (Figure 7) is proposed on the basis of our and other findings that Nrf1-transactivated proteasomes differentially contribute to positive and negative regulation of the CNC–bZIP protein depending on the extents of proteasomal inhibition [34,36,59,127]. This hypothesis has been confirmed by experiments showing that induction of 26S proteasome subunits and p97/VCP by lower doses of bortezomib is, rather, blocked by higher doses of the inhibitor (that completely represses the proteasome activity so that ubiquitinated proteins are accumulated) [36]. Similar effects are obtained from different concentrations of MG132 and ALLN (N-acetyl-L-leucyl-L-leucylnorleucinal) that are administrated in Nrf1-expressed cells [59].

A bidirectional regulatory feedback circuit between Nrf1 and its target 26S proteasome

Figure 7
A bidirectional regulatory feedback circuit between Nrf1 and its target 26S proteasome

Schematic representation of the bidirectional regulatory feedback circuit existing between Nrf1 and the proteasome. Transcriptional expression of the 26S proteasomal subunits and p97/VCP is regulated predominantly by Nrf1 (but not Nrf2). In turn, different doses of proteasomal inhibitors (e.g. 0.1–1.0 μmol/l MG132) exert opposing effects on Nrf1’s activity to mediate expression of genes encoding proteasomal subunits and p97. Specifically, low doses of MG132 (A) stimulate the up-regulation of Nrf1 target proteasomal subunit and p97 genes by increasing the levels of active 95-kDa and/or 85-kDa Nrf1 isoforms, whereas high doses of MG132 (B) cause the down-regulation of Nrf1-mediated transcription by increasing the levels of either 36-kDa Nrf1γ or 25-kDa Nrf1δ dominant-negative isoforms and decreasing the levels of cleaved 85-kDa isoforms. However, the detailed rational mechanism(s) of this biphasic response remains to be determined. In addition, further mechanistic study is warranted to identify which proteases enable the selective endoproteolytic processing of 120-kDa or 95-kDa Nrf1 proteins to yield the cleaved 85-kDa and other smaller isoforms.

Figure 7
A bidirectional regulatory feedback circuit between Nrf1 and its target 26S proteasome

Schematic representation of the bidirectional regulatory feedback circuit existing between Nrf1 and the proteasome. Transcriptional expression of the 26S proteasomal subunits and p97/VCP is regulated predominantly by Nrf1 (but not Nrf2). In turn, different doses of proteasomal inhibitors (e.g. 0.1–1.0 μmol/l MG132) exert opposing effects on Nrf1’s activity to mediate expression of genes encoding proteasomal subunits and p97. Specifically, low doses of MG132 (A) stimulate the up-regulation of Nrf1 target proteasomal subunit and p97 genes by increasing the levels of active 95-kDa and/or 85-kDa Nrf1 isoforms, whereas high doses of MG132 (B) cause the down-regulation of Nrf1-mediated transcription by increasing the levels of either 36-kDa Nrf1γ or 25-kDa Nrf1δ dominant-negative isoforms and decreasing the levels of cleaved 85-kDa isoforms. However, the detailed rational mechanism(s) of this biphasic response remains to be determined. In addition, further mechanistic study is warranted to identify which proteases enable the selective endoproteolytic processing of 120-kDa or 95-kDa Nrf1 proteins to yield the cleaved 85-kDa and other smaller isoforms.

Nrf1/TCF11-mediated ‘bounce-back’ response to lower concentrations of proteasomal inhibitors is triggered by partial inhibition of proteasomes, particularly ERAD-dependent proteasomes that predominantly tether to the cytoplasmic side of ER–NE networks [236239] (Figure 6). The putative ERAD-dependent proteasomes still exert a limited proteolytic activity to enable the selective processing of Nrf1 during ER-to-cytosolic retrotranslocation of its inactive 120-kDa glycoprotein in order to generate an active ∼95-kDa deglycoprotein and a cleaved ∼85-kDa protein [34,36,59]. It is important to note that production of the cleaved 85-kDa Nrf1 occurs after Hrd1-mediated ubiquitination of the retrotranslocated 95-kDa Nrf1 followed by proteasome-mediated proteolytic processing [34,36,106]. In addition to the 26S proteasome, its inhibitor-insensitive and other ER-localized proteases (e.g. calpain, Clp, SPP and USP19) may also be involved in, although some cytosolic protease candidates [36] had been excluded from, the proteolytic processing of Nrf1/TCF11. The resulting ∼95-kDa and 85-kDa Nrf1 proteins are localized subcellularly in close proximity to the ER such that they could be, to a moderate extent, protected by membranes against degradation by proteasomes and/or other cytosolic proteases, before being dislocated from the ER to the nucleus prior to transactivation of Nrf1 target genes (e.g. PSM and p97). By contrast, the cyto/nucleo-plasmic isoforms of Nrf1 (i.e. 55-kDa Nrf1β, 36-kDa Nrf1γ and 25-kDa Nrf1δ) are exposed to proteolytic degradation by the remaining proteasome activity in the presence of its inhibitors, so as to eliminate their dominant-negative activity to repress transcriptional expression of PSM and p97 through competing against active Nrf1 forms of ∼95 kDa and 85 kDa (Figure 7A). Alternatively, it cannot be ruled out that limited inhibition of proteasome enhances the stability of active ∼95-kDa and 85-kDa Nrf1 such that its transactivation activity is concomitantly prolonged to promote de novo formation of proteasomes, particularly proteasome complexes that are functionally assembled in the cyto/nucleo-plasm, where those shorter unprotected Nrf1 proteins are placed in a temporospatial preference for access to proteolytic degradation. The process is likely to occur via β-TrCPSCF- and Fbw7αSCF-mediated ubiquitin–proteasome pathways [106,107]. In addition, the Nrf1 target p97/VCP as a versatile AAA (ATPase associated with various cellular activities) protein provides a force to drive retranslocation of the 120-kDa Nrf1 from the lumen of ER into the cytoplasmic side and subsequently present the 95-kDa deglycoprotein to the ERAD-dependent ubiquitin–proteasome pathway for further proteolytic processing to produce the cleaved 85-kDa protein. Conversely, silencing of p97 or depletion of ATP abolishes the ER-to-cytosolic retrotranslocation of Nrf1 such that glycated (and unglycated) Nrf1 is stabilized and accumulated in the ER lumen insomuch as to block generation of deglycated 95-kDa and cleaved 85-kDa Nrf1 proteins [34,36,106,127].

In contrast with the effect of lower doses of proteasomal inhibitors on Nrf1, its transactivation activity is suppressed by sufficient inhibition of proteasomes by higher doses of these inhibitors [36,59,67] (Figure 7B). If the ERAD-dependent proteasome is completely inhibited, its proteolytic processing of Nrf1 to yield the cleaved 85-kDa protein is attenuated subsequently [36]. As a consequence, the abundances in the ER-associated full-length 120-kDa glycoptein and 95-kDa deglycoprotein of Nrf1 are increased owing to the stabilization of these two proteins by full inhibition of proteasome activity, but the transactivation activity of the active 95-kDa Nrf1 to regulate target genes (e.g. PSM and p97) appears to be competitively inhibited by dominant-negative Nrf1γ and Nrf1δ forms that are accumulated when cells have been treated with higher concentrations of proteasomal inhibitors [59,60].

Collectively, proteasome inhibitors have dual-opposing effects on Nrf1 and, in turn, Nrf1 contributes to dual regulation of the 26S proteasomal subunits. We therefore propose that a bidirectional regulatory feedback circuit exists between Nrf1 and the proteasome and envisage that the regulatory feedback enables Nrf1 to maintain cellular homoeostasis and organ integrity. Hence it is reasonable to hypothesize that, during normal development and growth, organ integrity is fine-tuned by a steady-state balance between the selective proteolytic processing of Nrf1 by the 26S proteasome and the transcriptional activity of Nrf1 to regulate the expression of proteasome subunits. However, it is not known how Nrf1-mediated proteasome recovery pathway co-ordinates with an adaptive cross-talk between ERAD and UPR that is provoked by certain proteasomal inhibition, to promote the topovectorial processing of Nrf1 to yield retrotranslocated 95-kDa and cleaved 85-kDa proteins before being dislocated from the ER into the cyto/nucleo-plasm, whereupon they may not only transactivate Nrf1 target genes and also be processed further to generate the smaller dominant-negative isoforms to turn off the cognate gene expression. In addition, how the dual-regulatory feedback circuit between Nrf1 and the proteasome cross-talks with antioxidant response signalling networks is unknown [166], albeit its orthologue Skn-1 is activated in a selective oxidative stress response to proteasomal dysfunction [240].

Nrf1 IS REQUIRED FOR CONTROLLING THE METABOLISM OF THE HEPATIC FATTY AND AMINO ACIDS

The liver is a central organ for the metabolism of fatty acids through lipid synthesis and oxidation. Upon liver-specific KO of Nrf1, such mutant mice manifested pathological signs of fatty liver similar to human NASH [32,33]. The inactivation of Nrf1 causes marked lipid (triacylglycerol) accumulation in vacuolated hepatocytes at 4 weeks of age in liver-specific Nrf1−/− mice [32], but serum triacylglycerol and total cholesterol levels are not obviously different between genetypes examined [33]. Subsequently, active proliferation of fatty Nrf1−/− hepatocytes also leads ultimately to development of hepatocellular adenoma and carcinoma by 12 months of age in the mutant mice. The fatty liver phenotype of conditional Nrf1-KO mice was initially considered to be attributable to severe oxidative damage by constitutive stress derived from up-regulation of microsomal fatty acid ω-oxidation by CYP4A enzymes, with the exception of no changes in the expression of PPARα (peroxisome-proliferation-activated receptor-α) and its target genes, such as those encoding acyl-CoA oxidase 1 (Acox1), enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase (Ehhadh) and acetyl-CoA acyltransferase 1 (Acca1), which are required for the β-oxidation pathway [32]. However, microarray analysis of similar Nrf1−/− livers reveals a 3.68-fold decrease in the expression of PDK4 (pyruvate dehydrogenase kinase 4) with potential ARE sequences in the promoter region [33]. Of note, PDK4 is a key regulatory enzyme involved in the switching of energy sources from glucose to fatty acids in response to environmental conditions, but up-regulation of PDK4 does not require the obligatory participation of PPARα in dietary steatohepatitis [241,242].

Further transcriptional expression profiling of conditional Nrf1-deficient livers reveals down-regulation of genes involved in lipid and amino acid metabolism, the TCA (tricarboxylic acid) cycle and the mitochondrial respiratory chain, as well as proteasomal degradation, whereas other genes involved in the cell cycle and DNA replication are up-regulated [234]. For example, the loss of Nrf1 results in decreased expression of a group of PPARα target genes, such as Acox1 and those encoding hydroxyacyl-CoA dehydrogenase (Hadh), aldo-keto reductase family 1 member D1 (Akr1d1), 3-hydroxy-3-methylglutaryl-CoA synthase 2 (Hmgcs2) and fatty acid desaturase 1 (Fads1). In addition to Pparα, two co-activators, Lipin1 and Pgc-, required for transcriptional expression of hepatic lipid metabolic enzymes are down-regulated in the livers of Nrf1-KO mice, with unaltered expression of Pgc- and Srebp1, rather than Pparγ, up-regulated in the mutant livers when compared with control mice [234]. Of note, the Lipin1 gene (regulated by Nrf1/MafG) was originally discovered using a positional cloning approach to identify the genetic mutation responsible for the fatty liver dystrophic (fld) mouse phenotype [243], and later Lipin1 was determined to function as a phosphatidic acid phosphatase or an inducible co-activator of PPARα/PGC-1α (PPARγ co-activator 1α) regulatory pathway [244]. Furthermore, Pgc- is also required for activation of oestrogen-related receptor α and nuclear respiratory factor 1; the last two factors are key regulators of the respiratory chain through the control of mitochondrial biogenesis [245]. Importantly, Nrf1-regulated gene expression profiling is not significantly affected in the Nrf2-KO or Keap1-KD mouse livers [234].

Two weeks after administration of 3-MC as a ligand of the AhR (arylhydrocarbon receptor), which is enabled for binding to the promoter of Cyp1A1 fused with the Cre-coding sequence, to the inducible liver-specific KO mice of Nrf1flox/flox:Cyp1A1-Cre, acute loss of hepatic Nrf1 leads to accumulation of triacylglycerols and cholesterol in the liver of mutant mice with the typical pathology of NASH [138]. Further examination of Nrf1-deficient livers revealed up-regulation of apolipoprotein receptor 2 (Apoer2/LRP8) and very-low-density lipoprotein receptor (Vldlr), but not of other lipid transporters [i.e. Ldlr (low-density lipoprotein receptor), Lipc (hepatic lipase), Lpl (lipoprotein lipase), Cd36, Slc27a4 (solute carrier 27 A4) and Mttp (microsomal triacylglycerol transfer protein)]. This finding indicates that loss of Nrf1 results in hepatic triacylglycerol accumulation due to an excessive uptake of chylomicrons and VLDL particles as a consequence of increased expression of Apoer2 and Vldlr, suggesting that both target genes are repressed by Nrf1 under normal homoeostatic conditions.

Metabolomic assays of the 3-MC-inducible Nrf1-KO livers revealed a significant increase in the polar lipid content comprising mainly fatty acids derived from triacylglycerols, with no change in the non-polar lipid content being detected. Within the composition of polar lipids (including saturated, mono- and poly-unsaturated fatty acids), marked decreases in the abundance of arachidonic, palmitic and stearic acids, whereas relative increases in oleic, gadoleic, γ-linoleic and palmitoleic acids were determined [138]. Of note, the increases in palmitoleic and oleic acids were associated with significant up-regulation of Fads3, whereas moderate down-regulation of Fads1 and Fads2 appeared to be consistent with the decreased level of arachidonic acid. Despite no obvious changes in the expression of cyclo-oxygenase 1 (Cox-1) and Cox-2, an increment in the expression of 5-lipoxygenase-activating protein (Alox5ap) was found as a key gene involved in the arachidonic acid metabolism that is negatively regulated by Nrf1, implying that acute loss of hepatic Nrf1 triggers a switch from protaglandin to leukotriene synthesis.

In addition to the negative regulation of lipoprotein receptors (i.e. Apoer2 and Vldlr) and lipid-metabolizing enzymes (i.e. Fads3 and Alox5ap) by Nrf1, it is also a direct transrepressor of xCT (a light-chain of the cystine/glutamate antiporter system Xc that is modulated by the variant adhesion molecule CD44v). Following acute 3-MC-inducible loss of hepatic Nrf1, the xCT gene is transactivated to increase the cellular uptake of cystine that is reduced to cysteine into a markedly increasing pool, along with moderate increases in glutamate and glycine, which together lead to an increase in GSH and GSSG levels, albeit accompanied by unaltered expression of Gclc and Gclm involved in glutathione synthesis. Taken together, Nrf1 is proposed to be involved negatively in a two-step-responsible induction of xCT and possibly other genes [138], with the positive regulation of xCT by Nrf2 [10]. Under basal conditions, Nrf1 is recruited to ARE sequences in xCT and contributes to gene repression, but it is displaced from these sites during oxidative/electrophile stress and thus allows other factors such as Nrf2 to induce xCT expression.

Moreover, several genes responsible for methionine meta-bolism, including methionine adenosyltransferase 1α (Mat1α), S-adenosylhomocysteine hydrolase (Ahcy), glycine N-methyltransferase (Gnmt) and nicotinamide N-methyltransferase (Nnmt) are down-regulated by loss of Nrf1 in the livers of specific KO mice [234], although no changes in these enzymatic substrates was observed after acute loss of hepatic Nrf1 [138]. This indicates that Nrf1 positively regulates methionine metabolism genes.

Nrf1 CONTROLS HOMOEOSTATIC GLUCOSE METABOLISM AND ALSO SUPPRESSES DIABETES DEVELOPMENT

The steady-state level of glucose in animals is maintained through insulin regulatory mechanisms, thus enabling such a constant physiological homoeostatic status to be kept in the body so as to evolutionarily develop robust adaptation to changes in glucose and other nutritional metabolisms. Conversely, deficiency of insulin-producing β-cells in the pancreas results in Type 1 diabetes, which is considered an autoimmune disease although its aetiology has not been elucidated. By contrast, Type 2 diabetes is characterized by peripheral insulin resistance, along with relative insulin deficiency in the early stages, followed by β-cell toxicity leading to the cellular inability to produce and secrete insulin in response to glucose stimulation in the later stages of the disease.

Gain of function of Nrf1 results in glucose metabolic disorders, insulin resistance and diabetes mellitus

An SNP (single nucleotide polymorphism) rs3764400 is identified by GWAS (genome-wide association study) to be located in the 5′-flanking region of human Nrf1 gene, representing a risk factor for obesity with BMI (body mass index) at a P value of 5×10−6 [246]. Later, the reporter gene activity indicated that rs3764400 is a regulatory SNP of Nrf1 gene expression with a ∼2.2-fold higher level being driven by the risk C allele, than that mediated by the control T allele, within the 2.1-kb promoter region upstream of the first transcription start site [247]. These findings suggest that increased expression of the Nrf1 gene may contribute to the pathophysiology of obesity and relevant metabolic dysregulation. In fact, the rs3764400-containing sequence 5′-CAGT/CT-3′ [246] is homologous with the reverse sequence 5′-CNGTT-3′ of the consensus c-Myb-binding site (5′-AACNG-3′) [248]. Therefore it is deduced that the major T allele of rs3764400 composes a consensus c-Myb-binding sequence 5′-CAGTT-3′ (the normal major T) (Figure 9), whereas the minor C allele cannot allow c-Myb-related factors to bind the altered motif 5′-CAGCT-3′ (the minor C risk). Together with induction of Nrf1 gene by the risk C allele of rs3764400 [247], it is proposed that c-Myb may act as an upstream regulator of Nrf1 transcriptional expression controlling glucose metabolism. Moreover, it should be noted that differential regulation of glucose metabolism by Nrf1 and Nrf2 is supported by the evidence showing that gain of function of Nrf1 increases blood glucose levels [247], and, conversely, blood glucose levels are decreased by gain of function of Nrf2 [249,250].

Since no interpretable evidence had been provided by loss-of-function mutants of mouse Nrf1 gene showing severe liver dysfunction [32,33], it is hence warranted for the gain-of-functional transgenic mice to be created by engineering overexpression of Nrf13×Flag that is driven by the MafG regulatory promoter domain (designated Nrf1-Tg:MGRD) [247]. As a result, increased expression of Nrf1 causes a loss in mouse body weights fed on normal and high-fat diets. Interestingly, a pathology resembling human diabetes mellitus was developed in high-fat-diet-induced obesity model mice harbouring Nrf1-Tg:MGRD, with markedly higher levels of blood glucose after insulin injection [247]. Further examinations revealed that the pancreatic islets of Nrf1-Tg:MGRD mice were much larger than those of wild-type mice, and the insulin-positive areas were also increased in the pancreas of Nrf1-Tg:MGRD mice. This finding suggests constitutive activation of insulin secretion in the response of Nrf1-Tg:MGRD mice to systemic insulin resistance. The notion is also supported by the presence of insulin resistance in Nrf1-Tg:MGRD mice fed on the normal diets, and a positive correlation of increased Nrf1 expression with induction of insulin resistance.

The induction of insulin resistance appears to be interpreted by possible disruption of signalling towards metabolic perturbations by forced Nrf1 expression. The hypothesis is addressed by the findings [247] that: (i) insulin-induced phosphorylation of Akt-Ser473 is significantly prevented in the liver and skeletal muscle, but not white adipose tissue, of Nrf1-Tg:MGRD mice, implying that Nrf1 suppresses insulin signalling to Akt phosphorylation; (ii) Nrf1-Tg:MGRD mice showed an increase in the weight of liver, rather than skeletal muscle or white adipose tissue, and the increased weight of liver was enhanced further under high-fat-diet conditions, implying altered levels of hepatic metabolism; (iii) both glucose utilization and production are impaired in the livers of Nrf1-Tg:MGRD mice by repressing expression of insulin-regulated glycolysis-related genes [e.g. those encoding GCK (glucokinase), AldoB (aldolase B), PGK1 (phosphoglycerate kinase 1), PK (pyruvate kinase)], gluconeogenesis-related genes (i.e. Fbp1 and Pck1) and the Glut2 (glucose transporter 2) gene, particularly under the high-fat-diet conditions; (iv) Nrf1 suppresses the entry of glucose into the glycolytic pathway, leading to decreased levels of G6P (glucose 6-phosphate) and F6P (fructose 6-phosphate), accompanied by an increased level of pyruvate along with a reduced level of lactate, such that the entry of the TCA cycle is, however, increased with levels of acetyl-CoA, citrate and ATP being elevated, in the liver of Nrf1-Tg:MGRD mice fed on high-fat diets; (v) Nrf1 increases utilization of hepatic acetyl-CoA to generate β-hydroxybutyrate from other fuels (i.e. fatty acids), but from neither glucose nor amino acid catabolism; and (vi) impaired glucose metabolism is obviously ameliorated by decreased expression of Nrf1 in heterozygous Nrf1+/− mice, when compared with wild-type Nrf1+/+ mice, both fed on high-fat diets.

Collectively, impairment of insulin action and regulated glucose metabolism by constitutive induction of Nrf1 results in development of diabetes mellitus in Nrf1-Tg:MGRD mice. Nonetheless, except that overexpression of Nrf1 suppresses insulin-induced Akt phosphorylation, no changes in the expression levels of major components of the insulin signalling pathway were detected by microarray analysis of the liver and skeletal muscle of Nrf1-Tg:MGRD mice [247]. This suggests that Nrf1 induces insulin resistance through a pathway distinct from transcriptional repression of the insulin signalling components. In addition, Nrf1 function may also be affected by altered levels of metabolic regulators, such as acetyl-CoA and its product β-hydroxybutyrate.

Pancreatic loss of Nrf1 function impairs insulin secretion and glucose metabolism leading to diabetes

The failure of pancreatic β-cells to secrete insulin sufficiently to meet increasing demand of glucose metabolism is defined as a major contributor to Type 2 diabetes with secondary loss of β-cells and thus impairment of β-cell function to reduce GSIS (glucose-stimulated insulin secretion) is a critical pathophysiological event of diabetes. Interestingly, β-cell-specific Nrf1-KO mice exhibit severe fasting hyperinsulinaemia, with reduced GSIS and glucose intolerance [201]. The phenotype results from the fact that deficiency of all Nrf1 isoforms (between 25 kDa and 140 kDa) in mouse islets and MIN6 β-cells leads to a marked increase in the basal insulin release with reduced GSIS, a β-cell phenotype reminiscent of the early stage of Type 2 diabetes. Since hyperinsulinaemia is considered a compensatory response to insulin resistance and its prolonged status can also induce insulin resistance and obesity [251,252], the pathological phenotype of Nrf1-KO mice is likely to account for the ensuing glucose intolerance.

Biochemical studies of Nrf1-deficient islets and β-cells reveal substantial increases in both NADPH/NADP and ATP/ADP ratios, due to enhanced glycolysis and mitochondrial metabolism (e.g. β-oxidation), rather than oxidative phosphorylation [201]. Further examination has demonstrated that silencing of Nrf1 up-regulates the expression of Glut2 and glycolytic proteins, such as lactate dehydrogenase 1 (Ldh1), glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and hexokinase 1 (Hk1), with down-regulation of Gck. Of note, the impaired glycolysis and GSIS resulting from loss of Nrf1 is rescued by KD of HK1. By contrast, no changes in the expression of mitochondrial protein complexes I and II, cytochrome c and ATP synthase F1, along with mitochondrial biogenesis-related PGC-1α, PGC-1β and nuclear respiratory factor 1 were observed in Nrf1-deficient cells when compared with equivalent wild-type [201]. Collectively, β-cell-specific depletion of Nrf1 results in a shift from oxidative phosphorylation to aerobic glycolysis for energy, as a consequence similar to the Warburg effect in cancer cells. Moreover, the basal expression of p53, phosphorylated Akt and AMPKα (AMP-activated protein kinase α) are increased in Nrf1-deficient cells [253], implying that these signal molecules may together be involved in alteration of glucose metabolism, although the detailed mechanisms remain to be determined. In particular, it has become obscured by an aberrant coupling of oxidative stress with the switch of metabolic changes in Nrf1-depleted cells.

Nrf1 IS REGULATED AT VARIOUS LEVELS THROUGH DISTINCT MECHANISMS

The aforementioned discoveries have convincingly demonstrated that the unique biological functions of Nrf1 are indispensable for maintaining cellular homoeostasis and organ integrity during normal development and growth, as well as other pathophysiological processes. For this reason, the functional activity of Nrf1 is finely tuned by a steady-state balance between its production and the concomitant processing into distinct isoforms before being turned over, which together confer on the host robust cytoprotection against a variety of cellular stresses (Figure 11A). In fact, accumulating evidence reveals that Nrf1 is tightly regulated at various levels through distinct mechanisms, some of which have been recently reviewed by Willmore and colleagues [254].

The selective topovectorial processing of Nrf1 to yield distinct isoforms with different functionalities to fine-tune the steady-state expression of ARE/EpRE-driven genes

In distinct topovectorial processes after translation, Nrf1 is selectively processed within different temporospatial subcellular locations, in which it is specifically modified in a variety of post-translational fashions (i.e. N-glycosylation, O-GlcNAcylation, deglycosylation, phosphorylation, ubiquitination, degradation and proteolytic cleavage) so as to yield multiple distinct isoforms (i.e. 120, 95, 85, 55, 46, 36 and 25 kDa as estimated by NuPAGE; Figures 3 and 6); they exert different and even opposing transcriptional activities to mediate differential expression of target genes. Among such distint isoforms, there are some polypeptides that can also be generated through differential biosynthesis from alternative translation starting at different initiation signals (to yield TCF11, Nrf1α, Nrf1β and Nrf1γ) located within various lengths of multiple mRNA transcripts, which arise from distinct transcription start sites existing within the single gene loci (Figures 4B and 4C) and alternative splicing of its mRNA transcripts insomuch as to give rise to several deletion mutants (e.g. Nrf1ΔS and Nrf1D; Figure 3). Nonetheless, much remains unknown about the detailed mechanisms underlying alternative translation and transcription of Nrf1 to yield such multiple products from its single gene so that they are differentially expressed to meet the host requirements for antioxidant, detoxification and cytoprotection from changing pathophysiological stresses.

This being the case, it is of crucial importance to understand the molecular and subcellular basis for Nrf1 regulation, about which there is no doubt as it has been shown for the topobiological model (Figure 6) that the selective post-synthetic processing of the CNC–bZIP protein into multiple polypeptide isoforms with different or opposing functional potentials is dictated by distinct topovectorial processes, whereby it (and its derivates) can allow for the proper topological folding within and around the ER so as to move in and out of the luminal side of membranes into extra-ER subcellular compartments (i.e. cyto/nucleo-plasm) before being dislocated in order to enter the nucleus, whereupon it is allowed to access its cognate genes. During translation of Nrf1, its NHB1 signal anchor sequence enables it to be integrated within and around the ER; this vectorial process allows its DNA-binding CNC–bZIP domains to be positioned on the cytoplasmic side of the ER membrane, whereas its TADs (including AD1, AD2, NST and SR regions) are transiently translocated into the lumen, whereupon the NST domain is N-glycosylated to yield an inactive 120-kDa Nrf1 glycoprotein. Subsequently, when required for induction by biological cues, portions of the TAD elements are dynamically retrotranslocated into extra-luminal subcellular compartments, whereupon it is deglycosylated to yield an active 95-kDa Nrf1 isoform. The membrane topovectorial process monitors the selective proteolytic processing of Nrf1 to yield distinct isoforms (between 85 kDa and 25 kDa) by the putative Hrd1-p97-mediated ERAD-dependent ubiquitin–proteasome pathway (and other cytosolic proteasome-independent mechanisms) [34,36,59,106]. Yet, further convincing evidence is required to support the selective processing of Nrf1 directly by the 26S proteasome-limited proteolysis even upon the intracellular ‘bounce-back’ response to low doses of proteasomal inhibitors [34,35], because the topic is a subject of debate with points of doubt reported previously.

Of note, it is questionable whether the post-translational processing of ectopic Nrf1 protein to yield some cleaved polypeptides is impaired by its net charged N-terminal (and C-terminal) epitope (e.g. the strong negative 3×FLAG), which directs an improper topology of the protein to be folded in an error-prone orientation within membranes [127]. Intriguingly, there is no available evidence to show that a sufficient blockage of the proteasome activity leads to a significant attentuation or complete abolishment of the putative proteolytic processing of the intact full-length Nrf1/TCF11 (with both glycoprotein and deglycoprotein forms of between 140 kDa and 120 kDa estimated on the routine SDS/PAGE gels, whereas they are subjected to separation by LDS/NuPAGE gels to migrate between 120 kDa and 95 kDa); this contradicts a recent report by Sha and Goldberg [36], who assumed that the potential proteasome-limited protealytic processing of the intact full-length Nrf1/TCF11 protein with a mass of ∼100 kDa (resolved by using SDS/PAGE) to give rise to a cleaved ∼75-kDa protein seemed to be abolished by high doses of proteasomal inhibitors.

It is important to note that there is no available evidence showing that the cytoplamic non-membranous 26S proteasome complex is rather integrated within membranes, a prerequisite for putative regulated intramembrane proteolysis [255]. Thus further studies are warrented to determine whether the intact full-length glycoprotein and deglycoprotein of Nrf1/TCF11 is processed proteolytically by an unidentified ER-lumen-resident and/or membrane-bound protease, rather than the Golgi-localized Site-1 and Site-2 proteases accounting for regulated intramembrane proteolysis of ATF6 and SREBP1 [40,41,58], in order to be released from the membranes and translocated into the nucleus.

The ER-localized Nrf1 is quality-controlled by the Hrd1/p97/ERAD-dependent proteasome pathway

It is clear that the bona fide ERAD-dependent ubiqintin–proteasome-mediated proteolysis is well placed to occur on the cytosolic side of ER membranes [129,256258]. Hence it is postulated that the cytoplasmic portion of Nrf1 may serve as an ERAD-C substrate that enables it to be targeted for ubiquitin–proteasomal degradation mediated by the TEB4/Doa10 E3 ligase complex. In contrast, the proteolytic processing of Nrf1/TCF11 appeared to be unaffected by KD of TEB4 [34], suggesting that other β-TrCPSCF and Fbw7SCF-dependent ubiquitin ligases or cytosolic proteases are involved in the cytoplasmic processing of the CNC–bZIP protein. Especially, although the ER-lumen-resident and membrane-bound regions of Nrf1/TCF11 could act as ERAD-L and ERAD-M substrates respectively, both should be, to a greater or lesser extent, protected by membranes against the cytoplasmic proteasome and others, particularly in cell-cycle-arrested or quiescent cells as described in [259]. Conversely, no matter whether the luminal and membranous portions of Nrf1 would be misfolded or not, if such domains are topologically repartitioned and retrotranslocated out of the ER across membranes into extra-ER cytoplasmic compartments, the CNC–bZIP protein would become more susceptible to proteolytic degradation by the Hrd1/p97/ERAD pathway [34,36,106]. Consistently, production of so processed 85-kDa Nrf1 possibly by the proteasomes is blocked by deletion of the ER-luminal anchoring NHB2 sequence (amino acids 81–106) [59]. It is therefore proposed that NHB2 could act as a potential dual retrotranslocation/degradation signal (designated as a retrodegron) recognized by Hrd1 or other E3 ligases, as a similar amphipathic helical sequence was identified as a functional degron targeting the yeast MATα2 transcription factor to the uniquitin–proteasome destruction by either ERAD or INMAD (inner nuclear membrane-associated degradation) [128,129]. Moreover, upon dynamic repositioning of potential degrons (i.e. CPD, DSGLS-containing SDS1 and PEST2) to be presented on the cytoplasmic side of membranes, Nrf1 is subjected to further selective proteolytic processing by other ubiquitin–proteasome pathways (e.g. β-TrCPSCF and Fbw7SCF [106,107]) and/or other cytosolic proteases to generate certain cleaved active or dominant-negative isoforms. This is also supported by the finding that production of Nrf1γ is abolished by removal of SDS1 or PEST2 degrons [59].

Collectively, dynamic membrane topological organization of Nrf1 determines its selective post-translational modifications, such as glycosylation, deglycosylation, ubiquitination and partial proteolysis by the cytoplasmic proteasome to yield several membrane-free isoforms before being translocated into the nucleus, which together control its ability to regulate target genes. If this is true, it is unfortunate that, while approaching objective reality, ever-increasing evidence reveals that the proteasomes (as holoenzymes differentially composed of regulatory and core particles) are predominantly nuclear and their steady-state accumulation occurs primarily at the nuclear periphery of dividing cells [259]. The novel discovery demonstrates a major requirement for proteasomal proteolysis in the nuclear subcompartment abutting the NE [256,260], which is contiguous with the ER, but is enriched in a unique subset of proteins (and probably lipids) that distinguish the INM (inner nuclear membrane) from the ER. Only during quiescence and cell cycle arrest do the proteasomes accumulate in proximity to the ER adjoining the ONM (outer nuclear membrane) [259].

The INM-bound Nrf1 may be quality-controlled by the putative INMAD-specific ubiquitin–proteasome pathway

Since there is no protein synthesis in the nucleus [261263], it is deduced that the NE membrane-bound Nrf1 is isolated from an uncleavable fraction of the intact ER-resident glycoproteins and deglycoproteins so as to import it to the INM, such a process that occurs during or just after translation of the nascent protein being enabled to ensure proper topological folding within and around ER membranes (as illustrated in Figure 6). Nonetheless, detailed molecular mechanisms by which the uncleavable Nrf1 protein is sorted and transported to the INM through the pore membrane (adjacent to the nuclear pore complex inserted where the double NE membranes ONM and INM are fused together) remain unsolved, but this process could be interpreted as several potential models proposed previously [41].

Upon deletion of the N-terminal residues 2–10 to give rise to a major mutant glycoprotein of Nrf1Δ2−10, it is exclusively localized in the ER rather than in the NE [40], indicating that the L2SLKKYLTE10 peptide might serve as a signal sorting the integral membrane-bound Nrf1 protein to the INM. In addition, the C-terminal tail arginine/lysine-enriched peptide R730RQERKPKDRRK741 does not function as a canonical NLS, but its removal causes the Nrf1Δ730−741 mutant to be released from the membranes and thus causes an apparent increase in the transactivation mediated by Nrf1 (and its short Nrf1β) [60]. It is therefore inferrable that the presence of the basic C-terminal tail peptide alone or in combination with the adjoining amphipathic TMc region enables it to act as an INM-tethering signal or a retention signal for non-specific binding of Nrf1 (and possibly Nrf1β) to the heterochromatin within the nuclear periphery. Besides its transmembrane regions, Nrf1 contains several potential sorting determinants, including luminal asparagine-glycan and cytoplasmic dileucine and tyrosine-based motifs, as they had been shown to sort other proteins [264].

In general, the unprocessed Nrf1 protein with a proper membrane topology may travel during interphase through the ER membrane to the NE, and diffuse passively within the plane of the membrane from the ER/ONM for import to the INM [265]. Membrane protection experiments against cytosolic proteases have led us to deduce that Nrf1 can associate with a membrane microdomain that is resistant to Triton X-100 [266]. The detergent-resistant membrane is likely to represent a coalesced lipid-ordered microdomain (e.g. lipid raft and caveolae) where phospholipids are much more tightly packed with cholesterol and sphingolipids than their surrounding non-raft regions of the bilayer [267]. For this, another possibility is that Nrf1 may be integrated into raft-like caveolae for post-mitotic sorting to the NE membranes [268]. Following trafficking of Nrf1 to the INM, the CNC–bZIP protein may flip-flop in the sphingolipid-rich nuclear membrane, as the constituent lipids and other membrane proteins in the detergent-resistant microdomain move in and out of the microdomain with different partitioning kinetics [267]. Collectively, it is hence proposed that Nrf1 may be located within such detergent-resistant microdomain structures and then ferried in a lipid raft along the continuous membrane from the ER/ONM to reach the INM through an unidentified mechanism by which it is conferred to pass the pore membrane barrier in check.

The rates of such membrane trafficking of Nrf1 in route to the INM can be determined by distinct membrane topological folding of its intact full-length proteins, in co-operation with its sorting signals, retention motifs and other relevant participants to nuclear import factors. To provide an explicit understanding of Nrf1 traffic to the INM, the following four models could be proposed on the basis of current knowledge [269275], in combination with our previous publications [37,41,58,76]. Firstly, if the nascent polypeptide of Nrf1 is allowed a simple fold to integrate in a mono-spanning orientation of Ncytoplasm/Clumen, the very short N-terminal sorting sequence that faces the cytoplasmic side of ER membranes enables the simply folded protein to pass freely through the pore membrane barrier such that it is rapidly diffused in the membrane to reach the INM, without respect to its large C-terminal region that is translocated predominantly in the lumen. As anticipitated, the transport of such a large lumen-resident region of Nrf1 may also be ATP-dependent, but it is unknown whether this region is chaperoned by its partner heat-shock p70 [94] or associated with the luminally retained heterohexameric portion of INM-specific LAP1-torsin complex [276,277]; this structure is similar to that of p97/VCP required for ERAD substrates extracted from membranes. Secondly, as illustrated in Figure 6, the C-terminal portion of the 120-kDa Nrf1 glycoprotein with a classic NLS is positioned on the cytoplamic side of membranes (which is topologically transferrable to the nucleoplasmic side of the INM). Such a <60-kDa cyto/nucleo-plasmic protion of Nrf1 is reasoned to enable membrane trafficking so it can pass through the lateral channel (∼10 nm in diameter) of the nuclear pore complex; the process can also be enhanced by putative interactions of its NLS with facilitated nuclear import factors, albeit this remains unresolved. Thirdly, a relatively lower rate of trafficking of the 95-kDa Nrf1 deglycoprotein is possible to evaluate due to its cyto/nucleo-plasmic portion being much greater than 60 kDa to hardly pass the lateral channel of nuclear pore complex. However, the trafficking of the deglycoprotein to the INM may also occur after its functional NLS directs its cytoplasmic portion to form an effective transport complex with canonical nuclear import factors; the resulting formation of such a traditional import complex can enable nuclear translocation of the 95-kDa Nrf1 deglycoprotein to pass through the central channel (∼39 nm in diameter) of the nuclear pore complex for import to the destined INM. As it is so hard to import to the INM, it is fortunate that the cyto/nucleo-plasmic portion of the 95-kDa Nrf1 deglycoprotein possesses three FG (Phe-Gly) pairing signals that facilitate its passage through the nuclear pore complex. Finally, Nrf1 may also be sorted for its traffic to the INM through post-mitotic membrane-related events [278], such as vesicle trafficking and membrane fusion.

Once in the nucleus, it is unclear how the INM-bound Nrf1 factor gains access to the promoters of target genes and the transcriptional machinery. It is possible that some domains of Nrf1 might be dynamically repartitioned and retrotranslocated from the perinuclear space (which is topologically continuous with the ER lumen) across the INM in order to present on the nucleoplasmic side. In this scenario, even though the N-terminal domain of the CNC–bZIP factor is integrally tethered to the INM, both its DNA-binding and TAD regions are repositioned to face the nucleoplasmic side such that they can enable direct free access to target genes and transcriptional machineries respectively. Despite no available evidence having been obtained showing that Nrf1 can be cleaved by signal peptidases, as well as by Site-1 and Site-2 proteases, insofar as to release it from its membrane-anchoring sequence and other N-terminal regions under normal homoeostatic conditions [40,58,76], it cannot be ruled out that the INM-bound protein might also be allowed for its possible proteolysis by an unidentified INM-specific ubiquitin–proteasomal degradation (INMAD) pathway to yield certain active or dominent-negative isoforms, particularly when it is required for induction by biological cues in order to meet the functioning of ARE-driven genes against distinct challenges to various pathophysiological cellular stresses.

In spite of the INM being a specialized domain of the ER, the ERAD-required Hrd1 is absent from the nuclear location [256,260], whereas the TEB4/Doa10 E3 ligase is found to import the INM and also be involved in INMAD for quality control of the yeast MATα2 transcription factor [128,129], kinetochore protein Ndc10 [279] and spindle pole body-relevent Nbp1 [280], each of which contains a membrane-active amphipathic α-helix that is folded into a structure resembling that of ALPS (ArfGAP1 lipid packing sensor), and tethers to the curved nuclear membranes. However, the putative INMAD-dependent proteolytic processing of Nrf1/TCF11 appeared to be also unaffected by KD of TEB4/Doa10 [34], suggesting that the CNC–bZIP protein is quality-controlled by other INM-associated ubiquitin–proteasome pathways, such as a counterpart of the yeast INMAD-specific ubiquintin E3 ligase (Asi1–Asi3) complex [262,263,281], that mediates the degradation of transcription factors Stp1/2 required for activation of the Ssy1/Ptr3/Ssy5 amino-acid-sensing pathway and promotes the degradation of functional regulators (e.g. Erg11 and Nsg1) of sterol biosynthesis. Hence we envisage that such spatial segragation between ERAD and INMAD of the protein quality-control systems [129,256,257,260] may be conferred on Nrf1 to excute its unique intrinsic functions.

Of note, in quiescent or cell-cycle-arrested cells, some proteasomes are sequestered into the JUNQs (juxtanuclear quality-control compartments) [282,283], which behave similarly to proteasome storage granules, situated at the cytoplasmic side of the NE, but become dissipated just after cell growth is resumed, such that they are rapidly imported into the nucleus to contribute their function in cell proliferation [256,259]. Similarly, a considerable fraction of Nrf1 was found to be localized within juxtanuclear aggresome-like inclusion bodies (which resemble JUNQs) within the ER abutting the NE [37,59,94]. The observation suggests that possible co-localization of Nrf1 with the proteasomes stored enables its protein quality to be controlled by putative INMAD in real time and space, as described elsewhere [256,260]. Moreover, the real temporospatial processing of Nrf1 by putative INMAD-dependent proteasomes, as well as its dynamic access to distinct subsets of target genes, may also be selectively confined by the interwoven lamina (as a stable structure beneath the INM) to a subnuclear territory [271,284]. It therefore seems most probable that Nrf1 undergoes some triage of topological changes within proximity to the nuclear periphery, so enabling its bZIP domain to form distinct functional heterodimers with each of sMaf or other bZIP proteins before binding to diverse AP1-like ARE/EpRE enhancers in cognate gene promoters. Indeed, a well-documented example of the principle of sequestering soluble transcription factor to the NE is probably that of c-Fos by lamin A/C [285], such that the perinuclear localization of c-Fos enables it to gain access to its partner Nrf1 to form a functional dimer, besides the canonical AP1 dimer with c-Jun.

As an upstream transcriptional regulator of Nrf1, SREBP1 is synthesized as a precursor protein embedded within the ER membranes, but it was also found to be associated with the unprocessed pre-lamin A in the vicinity of the nuclear rim [286], which hence limits the access of the cholesterol transcription factor to target genes in the nuclear interior rather than the nuclear periphery, but it is unknown whether the Nrf1 gene is positioned in the perinuclear subcompartment. In fact, transcription factors in association with chromatin-modifying complexes seem to play a central role in regulating chromatin dynamics and topological positioning into distinct subnuclear compartments. This was shown by zinc-finger-mediated tethering of the GATA1-associated protein Ldb1 to the β-globin promoter, leading to the formation of a chromatin loop between the promoter and the LCR and to expression of the β-globin gene in the absence of GATA1 [287]. This suggests that Nrf1 (and/or its derivatives) is likely to be responsible for regulating its founding target β-globin gene expression.

Although the nucleus is a more reducing organelle (i.e. GSH/GSSG≈100:1), essential for gene replication and transcription, the ER lumen provides a more oxidizing environment (i.e. GSH/GSSG≈1:1–3:1) required for protein and lipid biosynthesis [265,288]. Such a large difference in the redox compartmentalization demonstrates that the INM serves as a redox barrier enabling for the temporospatial segregation of the reduced nucleoplasm from the oxidizing perinuclear space, which is contiguous with the ER lumen and also shares topological similarity to the extracellular environments. We therefore propose that the INM-bound Nrf1 should play a direct indispensable role in sensing redox changes on two sides of nuclear membranes and signalling transduction towards the regulation of ARE/EpRE-driven antioxidant, detoxification and cytoprotective genes against various stresses.

Transcriptional regulation of Nrf1 and its target genes by multiple signalling networks

The transcriptional expression of Nrf1 along with distinct subsets of downstream target genes was monitored by several signalling pathways (Figure 9). A recent report has showed that Nrf1, but not Nrf2, enables recruitment of Pitx2/3 on the PitxRE (Pitx-response element, with the sequence 5′-TAATCC…TAATCC-3′) in the promoter and hence is involved in Pitx2/3-mediated expression of antioxidant genes [i.e. Aldh1a1 (aldehyde dehydrogenase 1A1), Mt-1, Gsta1 and Sod1] as stem cells switch from a glycolytic proliferative progenitor state into an oxidative post-mitotic differentiated state at the onset of differentiation during embryonic development [289]. This demonstrates the coupling of the Pax3/7/Pitx2/3/Nrf1-mediated redox gene regulatory networks with the metabolic switch as stem cells become more specialized within the lineage. Similar conclusion is also drawn from the findings that gain of function of Nrf1 represses insulin-regulated glycolysis gene expression possibly through Myb-dependent signalling [247], but conversely its loss of function results in a shift from oxidative phosphorylation to aerobic glycolysis for rising energetic needs, with no changes in the expression of nuclear respiratory factor 1 and its target genes controlling mitochondrial biogenesis and function in order to meet metabolic needs of differentiated cells [201]. Such genes (i.e. Cytochromec, Tfam, Ndufv1 and Ndufv6) regulated by nuclear respiratory factor 1 [245] may also be monitored by Pitx2/3 [289], but not directly by Nrf1 [201]. Therefore, Nrf1 is considered to act as a vital nexus between redox signalling [e.g. Pax3/7/Pitx2/3/Nrf1/Aldh1a1 (aldehyde dehydrogenase 1A1)] and oxidative metabolic networks during cell differentiation. Moreover, stem cell differentiation and fate specification may also be regulated by Sox9/Nrf1/Aldh1a1 redox signalling, which enables transactivation of Nrf1 through the consensus Sox-binding sites (5′-CAATGa…AcCAATGa-3′) in the promoter [290].

In the proteostatic response to insulin and growth factors [212], the inducible expression of a group of Nrf1-mediated proteasomal subunit genes is activated transcriptionally via the TSC2/mTORC1/SREBP1 signalling pathway (Figure 9), but not through classical ER stress signalling pathways (Figure 8). The transactivation of Nrf1 by SREBP1 has been shown to be attributable to the presence of four consensus SREs (sterol-regulatory elements, with the forward and reverse sequences 5′-TCACXCCAc-3′ and 5′-gTGGXGTGA-3′) in close proximity to two potential transcription start sites located within the gene promoter region [212]. However, it is unknown whether (or which) SRE motifs within the Nrf1 promoter might be required to drive expression of the uORF (upstream ORF) or switch from the uORF to the mORF (main ORF) (Figure 11B).

Proposed relevance of Nrf1-regulatory gene networks to ER-derived stress-response pathways

Figure 8
Proposed relevance of Nrf1-regulatory gene networks to ER-derived stress-response pathways

It is unclear whether and how Nrf1, Nrf2 and Nrf3 are involved in the regulation of target genes responsible for ER-derived stress. Of note, the cytosolic Nrf2 has been shown to be phosphorylated by PERK in response to ER (i.e. unfolded protein) stress [202,210,211], whereas classic ER stressors have been reported to have opposing effects on the ER-resident Nrf1 and Nrf3 factors [41,108,123]. Conversely, ER-derived (i.e. redox, proteotoxic and metabolic) stresses resulting from loss of Nrf1 function are proposed to contribute to relevant pathophysiological phenotypes (e.g. NASH-based inflammation and carcinogenesis), which might be induced by three classic signalling pathways dependent on PERK, IRE1 (which acts dually as an endoribonuclease and serine/threonine protein kinase) and ATF6 towards the activation of UPR genes. Thus chronic ER stress could be concomitant with endogenous oxidative proteotoxic and metabolic stress resulting from dysregulated expression of Nrf1-mediated cytoprotective genes, required for antioxidant buffering, detoxification, protein degradation, apoptosis, anti-inflammation and tumour suppression in Nrf1-deficient cells. The resulting induction of relevant pathophysiological processes by chronic prolonged ER stress is attributable to altered activation of the following cis-regulatory response elements: ARE (5′-TGACNNNGC-3′); AARE (amino acid-response element, 5′-TGATGCAAA-3′); ERSE [ER-stress-response element, 5′-CCAAT-(N9)-CCACG-3′ and 5′-ATTGGNCCACG-3′]; UPRE (UPR element, 5′-TGACGTGG/A-3′) and other consensus binding sites (Table 1). Transcriptional expression of UPR genes are differentially regulated by XBP1 (X-box-binding protein 1), ATF4, ATF6, C/EBP and CHOP; each could serve as a partner of heterodimerization with Nrf1. Notably, Nrf1 (or Nrf3) is selectively processed by an ER-associated mechanism [36,37,59], which is distinctive from the regulated intramembrane proteolysis of ATF6 and SREBP1 in the Golgi apparatus such that their active N-terminal portions are released and translocated into the nucleus before transactivating genes responsible for cholesterol and other lipid metabolic processes [213,218,301].

Figure 8
Proposed relevance of Nrf1-regulatory gene networks to ER-derived stress-response pathways

It is unclear whether and how Nrf1, Nrf2 and Nrf3 are involved in the regulation of target genes responsible for ER-derived stress. Of note, the cytosolic Nrf2 has been shown to be phosphorylated by PERK in response to ER (i.e. unfolded protein) stress [202,210,211], whereas classic ER stressors have been reported to have opposing effects on the ER-resident Nrf1 and Nrf3 factors [41,108,123]. Conversely, ER-derived (i.e. redox, proteotoxic and metabolic) stresses resulting from loss of Nrf1 function are proposed to contribute to relevant pathophysiological phenotypes (e.g. NASH-based inflammation and carcinogenesis), which might be induced by three classic signalling pathways dependent on PERK, IRE1 (which acts dually as an endoribonuclease and serine/threonine protein kinase) and ATF6 towards the activation of UPR genes. Thus chronic ER stress could be concomitant with endogenous oxidative proteotoxic and metabolic stress resulting from dysregulated expression of Nrf1-mediated cytoprotective genes, required for antioxidant buffering, detoxification, protein degradation, apoptosis, anti-inflammation and tumour suppression in Nrf1-deficient cells. The resulting induction of relevant pathophysiological processes by chronic prolonged ER stress is attributable to altered activation of the following cis-regulatory response elements: ARE (5′-TGACNNNGC-3′); AARE (amino acid-response element, 5′-TGATGCAAA-3′); ERSE [ER-stress-response element, 5′-CCAAT-(N9)-CCACG-3′ and 5′-ATTGGNCCACG-3′]; UPRE (UPR element, 5′-TGACGTGG/A-3′) and other consensus binding sites (Table 1). Transcriptional expression of UPR genes are differentially regulated by XBP1 (X-box-binding protein 1), ATF4, ATF6, C/EBP and CHOP; each could serve as a partner of heterodimerization with Nrf1. Notably, Nrf1 (or Nrf3) is selectively processed by an ER-associated mechanism [36,37,59], which is distinctive from the regulated intramembrane proteolysis of ATF6 and SREBP1 in the Golgi apparatus such that their active N-terminal portions are released and translocated into the nucleus before transactivating genes responsible for cholesterol and other lipid metabolic processes [213,218,301].

In the anti-inflammatory (and innate immune) responses [291], Nrf1/TCF11 is stimulated by the TGFβ (transforming growth factor β)-dependent signalling pathway as to repress transcriptional expression of iNOS (inducible nitric oxide synthase), an important player in acute inflammation and septic shock that is synergistically induced by cytokines (e.g. TNFα, interleukin 1β and interferon γ). Thus Nrf1/TCF11 has been identified as a transrepressor through binding the AP1/NF-E2-like sequence (5′-TGCTGAAAGTGAG-3′) in the iNOS promoter. The induction of Nrf1/TCF11 to down-regulate iNOS by TGFβ occurs via Smad6 short form, which is suppressed by Smad7, but it is unknown whether the putative Smad6/7-binding site exists either in the promoter of Nrf1 or in close proximity to the Nrf1/MafG-binding site of the iNOS promoter.

Interestingly, up-regulation of Nrf1 by TGFβ is enhanced by PMA (a tumour-promoting agent); the effect of TGFβ on Nrf1 is suppressed by calphostin, whereas the combinational effect of TGFβ and Smad6 short form is blocked by the PKC inhibitor staurosporine [291], indicating that activation of Nrf1 by TGFβ and/or Smad6 short form is stimulated by PMA dependent on PKC (Figure 9). In addition, both protein kinase CK2 and ERK2 have been shown to phosphorylate Nrf1β/LCR-F1 and thereby render it active in binding to the NF-E2/AP1 site in the TNFα promoter [63]; however, it is unknown which sites were phosphorylated by ERK2 or PKC.

Transcriptional expression of Nrf1 target genes regulated by several upstream signalling pathways

Figure 9
Transcriptional expression of Nrf1 target genes regulated by several upstream signalling pathways

Multiple mRNA transcripts arising from alternative transcriptional expression of the single Nfe2l1 gene and subsequent processing of its products can be translated through different initiation signals within the ORF insofar as to generate distinct Nrf1 isoforms, which may be selectively processed in the ER or extra-ER subcellular compartments before being translocated into the nucleus. Clearly, there exist several AP1-like ARE sequences in the promoter region of Nfe2l1 such that gene transcription has been shown to be regulated by Jun, JunD and Nrf3 [146,162,163]. Downstream p53 (and p21) expression is differentially regulated by Jun/JunD directly or indirectly via Nrf1. Importantly, expression of Nrf1 target proteasomal subunit genes are transcriptionally activated by the upstream TSC2/mTORC1/SREBP1 signalling response to insulin and other growth factors [212]. More interestingly, Nrf1, but not Nrf2, acts as an essential nexus between its target antioxidant genes and its upstream Pax3/7/Pitx2/3 signalling, and the constitutive activation of Nrf1 target antioxidant gene networks (also regulated by Sox9 [290]) enables ESCs to switch from a glycolytic proliferative progenitor state to an oxidative metabolic differentiated state during normal development [289]. However, it should be noted that mitochondrial respiratory genes involved in oxidative metabolism are regulated by nuclear respiratory factor-1 (NRF-1 with a molecular mass of 68 kDa); this abbreviation is confusingly similar to Nrf1. Moreover, it was found that the rs3764400-containing sequence 5′-CAGT/CT-3′ exists in the promoter of human Nrf1 gene [246]; the major T allele of the sequence could serve as a consensus c-Myb-binding site (5′-AACNG-3′) [248], whereas the minor C-risk allele does not allow c-Myb to bind the altered motif (5′-CAGCT-3′), leading to dysregulation of insulin-activated glycolysis and ultimate development of diabetes. In the anti-inflammatory response, Nrf1 is induced by Smad6 short form signalling and this effect is repressed by Smad7, but the detailed mechanism remains to be determined.

Figure 9
Transcriptional expression of Nrf1 target genes regulated by several upstream signalling pathways

Multiple mRNA transcripts arising from alternative transcriptional expression of the single Nfe2l1 gene and subsequent processing of its products can be translated through different initiation signals within the ORF insofar as to generate distinct Nrf1 isoforms, which may be selectively processed in the ER or extra-ER subcellular compartments before being translocated into the nucleus. Clearly, there exist several AP1-like ARE sequences in the promoter region of Nfe2l1 such that gene transcription has been shown to be regulated by Jun, JunD and Nrf3 [146,162,163]. Downstream p53 (and p21) expression is differentially regulated by Jun/JunD directly or indirectly via Nrf1. Importantly, expression of Nrf1 target proteasomal subunit genes are transcriptionally activated by the upstream TSC2/mTORC1/SREBP1 signalling response to insulin and other growth factors [212]. More interestingly, Nrf1, but not Nrf2, acts as an essential nexus between its target antioxidant genes and its upstream Pax3/7/Pitx2/3 signalling, and the constitutive activation of Nrf1 target antioxidant gene networks (also regulated by Sox9 [290]) enables ESCs to switch from a glycolytic proliferative progenitor state to an oxidative metabolic differentiated state during normal development [289]. However, it should be noted that mitochondrial respiratory genes involved in oxidative metabolism are regulated by nuclear respiratory factor-1 (NRF-1 with a molecular mass of 68 kDa); this abbreviation is confusingly similar to Nrf1. Moreover, it was found that the rs3764400-containing sequence 5′-CAGT/CT-3′ exists in the promoter of human Nrf1 gene [246]; the major T allele of the sequence could serve as a consensus c-Myb-binding site (5′-AACNG-3′) [248], whereas the minor C-risk allele does not allow c-Myb to bind the altered motif (5′-CAGCT-3′), leading to dysregulation of insulin-activated glycolysis and ultimate development of diabetes. In the anti-inflammatory response, Nrf1 is induced by Smad6 short form signalling and this effect is repressed by Smad7, but the detailed mechanism remains to be determined.

Nrf1-binding transcriptional complexes dictate differential expression of distinct subsets of genes

Figure 10
Nrf1-binding transcriptional complexes dictate differential expression of distinct subsets of genes

The differences between AP1-like NF-E2/TCF11/ARE sequences and their homologues (Table 1) dictate recruitment of various heterodimeric complexes of Nrf1 (or other isoforms and family members such as NF-E2, Nrf2 and Nrf3) with its partners (sMaf and other bZIP proteins) in order to mediate differential transcriptional expression of distinct subsets of cognate target genes. (A) The reverse sequence of ARE (called the NF-E2/TCF11-binding site) enables Nrf1/TCF11, like NF-E2 and Nrf2, to mediate transcriptional expression of both haemopoietic genes [i.e. those encoding β/γ-globin and PBGD (porphobilinogen deaminase)] and non-haemopoietic genes (i.e. those encoding HO-1, ferritin H, MT1/2 (metallothionein 1/2), NQO1 and GST) respectively. Furthermore, Nrf1 also regulates genes controlling the cell cycle (e.g. Cdc2L1/2 and Cdc25B) and others (e.g. CYC1). (B) In the undifferentiated odontoblasts, DSPP promoter activity is repressed by a physical interaction of Nrf1 with C/EBPβ through their bZIP domains, which is monitored by PKA via phosphorylation of Nrf1 at Ser599 within the Neh6L domain [91]. Upon dissociation of Nrf1 from C/EBPβ, the single DSPP gene allows the transcriptional expression of both DPP and DSP insomuch as to promote odontoblast differentiation. In additional mineralized tissues, interaction of C/EBPβ with Runx2 enhances osteocalcin gene expression. (C) Ascorbic acid (also called vitamin C) up-regulates Nrf1-mediated Osx, an osteoblast-specific transcription factor, in BMS cells [158]. It also stimulates embryonic and mesenchymal stem cells to differentiate into osteoblasts through activation of transcription factor Cbfa1 via signalling of MAPK. This kinase is activated following interaction of integrin with collagen in the intracellular response to vitamin C (as a cofactor of prolyl hydroxylase required for hydroxylation and secretion of procollagen to form stable triple-helical collagen, in addition to a key redox inducer).

Figure 10
Nrf1-binding transcriptional complexes dictate differential expression of distinct subsets of genes

The differences between AP1-like NF-E2/TCF11/ARE sequences and their homologues (Table 1) dictate recruitment of various heterodimeric complexes of Nrf1 (or other isoforms and family members such as NF-E2, Nrf2 and Nrf3) with its partners (sMaf and other bZIP proteins) in order to mediate differential transcriptional expression of distinct subsets of cognate target genes. (A) The reverse sequence of ARE (called the NF-E2/TCF11-binding site) enables Nrf1/TCF11, like NF-E2 and Nrf2, to mediate transcriptional expression of both haemopoietic genes [i.e. those encoding β/γ-globin and PBGD (porphobilinogen deaminase)] and non-haemopoietic genes (i.e. those encoding HO-1, ferritin H, MT1/2 (metallothionein 1/2), NQO1 and GST) respectively. Furthermore, Nrf1 also regulates genes controlling the cell cycle (e.g. Cdc2L1/2 and Cdc25B) and others (e.g. CYC1). (B) In the undifferentiated odontoblasts, DSPP promoter activity is repressed by a physical interaction of Nrf1 with C/EBPβ through their bZIP domains, which is monitored by PKA via phosphorylation of Nrf1 at Ser599 within the Neh6L domain [91]. Upon dissociation of Nrf1 from C/EBPβ, the single DSPP gene allows the transcriptional expression of both DPP and DSP insomuch as to promote odontoblast differentiation. In additional mineralized tissues, interaction of C/EBPβ with Runx2 enhances osteocalcin gene expression. (C) Ascorbic acid (also called vitamin C) up-regulates Nrf1-mediated Osx, an osteoblast-specific transcription factor, in BMS cells [158]. It also stimulates embryonic and mesenchymal stem cells to differentiate into osteoblasts through activation of transcription factor Cbfa1 via signalling of MAPK. This kinase is activated following interaction of integrin with collagen in the intracellular response to vitamin C (as a cofactor of prolyl hydroxylase required for hydroxylation and secretion of procollagen to form stable triple-helical collagen, in addition to a key redox inducer).

The specificity of DNA-binding activity of Nrf1 at the NF-E2/AP1-like ARE/EpRE sequences (Table 1) existing in distinct gene promoters in order to differentially up- or down-regulate transcriptional expression of target genes are also dictated by subtle differences in functional heterodimerization with different partners (e.g. sMaf and other bZIP proteins such as FosB, c-Jun, JunD and ATF4). Among them, the sMaf proteins act as cognate partners repressing Nrf1-mediated transactivation [80,92,292], whereas ATF4 is critical for partnering with Nrf1 essential for normal embryonic development and growth [89,90,134,139]. In the transcriptional regulation networks, these partners enable formation of various functional dimers with distinct CNC–bZIP proteins and hence competition and/or synergy between family members and partners might monitor the ability of Nrf1 to differentially regulate distinct subsets of NF-E2/AP1-like ARE-driven genes [254].

It is important to note that Jun, a key partner to heterodimerize with Nrf1 [62,81,88], is also essential for fetal liver development as global KO of the gene (Jun−/−) leads to embryonic lethality at mid-gestation (∼12.5 pdc) [143,144]; this phenotype resembles that of Nrf1−/− mice [29,30]. Furthermore, hepatocyte-specific deletion of Jun in adult mice affects cell survival and cell cycle progression due to increased p53 expression and ensuring impairment of p21-dependent liver regeneration [293]. Interestingly, increased oxidative stress occurs in Jun−/− livers as accompanied by down-regulated expression of Nrf1 and Nrf2, but the impairment is rescued by JunD in place of Jun (i.e. Jund/d) fetal liver [146], indicating transcriptional regulation of Nrf1/2 by Jun or JunD (Figure 9). In contrast, JunD (similar to Nrf2) also regulates genes involved in the antioxidant defences, but Jund/d mice still exhibit early senescence through increased p19ARF expression, which is up-regulated in Jun−/− cells [294]. Moreover, the Jun-dependent cellular senescence can be restored by EGF and HB-FGF (human basic fibroblast growth factor) stimulation to activate the EGFR (EGF receptor)/Akt/GSK-3β signalling pathway [146]. Collectively, Jun and JunD have the ability to regulate antioxidant, detoxification and cytoprotective genes through monitoring the expression of Nrf1 and Nrf2 or relevant dimeric DNA-binding activity. Together with Jun, Nrf1 is critical for regulation of key genes controlling liver development, whereas Nrf2 appears to depend on the genetic backgrounds and environmental conditions.

CONCLUDING REMARKS

The unique biological functions of Nrf1 and its pathophysiological phenotypes (Figure 11A) are distinctive from those of Nrf2, as well as other CNC–bZIP factors. These sharp distinctions are determined by differences in the primary structures of their gene products and their functional subcellular locations, in which they are regulated through different mechanisms in distinct signalling processes. Collectively, the selective topovectorial processing of Nrf1 (and its long TCF11) via various mechanisms yields multiple isoforms of between 140 kDa and 25 kDa exerting different biological activities, which together regulate basal and inducible expression of distinct subsets of NF-E2/AP1-like ARE/EpRE-driven cytoprotective genes against a variety of cellular stresses, but each isoform-specific physiobiological function has not yet been determined in vivo.

Nrf1 is required for mediating multifunctional cytoprotective responses to various stimulators

Figure 11
Nrf1 is required for mediating multifunctional cytoprotective responses to various stimulators

(A) The cartoon shows that a functional heterodimer of Nrf1 with either sMaf or other bZIP proteins fine-tunes differential expression of distinct subsets of NF-E2/AP1/ARE-driven cytoprotective genes that are essential for maintaining cellular homoeostasis and organ integrity in multifunctional responses to a variety of endogenous and exogenous stimulators during normal development and growth. Notably, several significant pathological phenotypes have been observed to occur in different transgenic mice expressing loss-of-function or gain-of-function mutants of Nrf1 (and its isoforms). (B) Besides the mORF, an uORF is predicted to exist in the full-length Nrf1 mRNA sequence (but not a similar uORF to that in Nrf2 or Nrf3). The uORFs in mouse Nrf1 and human TCF11 could be translated into Nrf1u and TCF11u, which are postulated to sense redox stress and bind metals through the conserved cysteine-histidine motifs (in light blue), and/or be bound to membranes through the amphipathic region that is shown as a wheel (on the right-hand side). Of note, uORF may also have an effect on the expression profiling of distinct Nrf1 isoforms. Among them, TCF11 is a long form of Nrf1, which arises from alternative splicing to delete a Neh4L-coded exon from AD1 in TCF11. Both are broadly expressed in humans, but TCF11 does not exist in mice. Nrf1β/LCR-F1 is a short form, arising from the in-frame translation, and lacks its N-terminal 295 amino acids covering both NTD and AD1, whereas Nrf1δ is a dominant-negative form, arising from the in-frame translation, because it lacks the TADs AD1, NST and AD2. In addition, the naturally occurring Nrf1ΔN mutant is generated through an alternative transcription start site of the gene such that the first N-terminal 181-amino-acid region of wild-type Nrf1 is deleted and replaced by an additional 12 amino acids (i.e. MGWESRLTAASA).

Figure 11
Nrf1 is required for mediating multifunctional cytoprotective responses to various stimulators

(A) The cartoon shows that a functional heterodimer of Nrf1 with either sMaf or other bZIP proteins fine-tunes differential expression of distinct subsets of NF-E2/AP1/ARE-driven cytoprotective genes that are essential for maintaining cellular homoeostasis and organ integrity in multifunctional responses to a variety of endogenous and exogenous stimulators during normal development and growth. Notably, several significant pathological phenotypes have been observed to occur in different transgenic mice expressing loss-of-function or gain-of-function mutants of Nrf1 (and its isoforms). (B) Besides the mORF, an uORF is predicted to exist in the full-length Nrf1 mRNA sequence (but not a similar uORF to that in Nrf2 or Nrf3). The uORFs in mouse Nrf1 and human TCF11 could be translated into Nrf1u and TCF11u, which are postulated to sense redox stress and bind metals through the conserved cysteine-histidine motifs (in light blue), and/or be bound to membranes through the amphipathic region that is shown as a wheel (on the right-hand side). Of note, uORF may also have an effect on the expression profiling of distinct Nrf1 isoforms. Among them, TCF11 is a long form of Nrf1, which arises from alternative splicing to delete a Neh4L-coded exon from AD1 in TCF11. Both are broadly expressed in humans, but TCF11 does not exist in mice. Nrf1β/LCR-F1 is a short form, arising from the in-frame translation, and lacks its N-terminal 295 amino acids covering both NTD and AD1, whereas Nrf1δ is a dominant-negative form, arising from the in-frame translation, because it lacks the TADs AD1, NST and AD2. In addition, the naturally occurring Nrf1ΔN mutant is generated through an alternative transcription start site of the gene such that the first N-terminal 181-amino-acid region of wild-type Nrf1 is deleted and replaced by an additional 12 amino acids (i.e. MGWESRLTAASA).

Of note, Nrf1 (and TCF11) has been identified as a dynamic membrane protein entailing distinct topologies that are folded within and around the ER and NE membranes. The membrane topology of Nrf1 (Figures 6 and 8) is completely distinct from those of ATF6 and SREBP1 that adopt in an Ncytoplasm/Clumen structure spanning the Golgi apparatus membranes [213,218]. The NHB1-adjoining signal sequence of Nrf1 enables it to be integrated within the ER membranes, but is not processed via a mechanism similar to the regulated intramembrame proteolysis that accounts for the processing of ATF6 and SREBP1; this is attributable to the fact that the CNC–bZIP protein processes neither Site-1 nor Site-2 protease cleavage sites [40,58]. Once Nrf1 is anchored within the ER through the uncleavable NHB1 sequence, the NHB2-connecting portions of the nascent polypeptide are translocated through the Sec 61 complex into the ER lumen, in which its NST domain is N-glycosylated so as to become an inactive 120-kDa Nrf1 glycoprotein (a similar TCF11 glycoprotein is estimated with a molecular mass of between 130 kDa and 140 kDa). If required for induction by biological cues, the ER-protected transactivation domains of Nrf1/TCF11 are dynamically retrotranslocated through an unidentified mechanism and repositioned from the luminal side of ER membranes into the cyto/nucleo-plasmic side, whereupon the CNC–bZIP protein is deglycosylated to yield an active 95-kDa factor (enabling transactivation of target genes), which is subsequently processed by a proteasomal inhibitor-sensitive protease (e.g. 26S proteasome) to generate a cleaved 85-kDa protein. These longer isoforms of Nrf1 may be selectively processed further insomuch as to give rise to several smaller isoforms between 85 kDa and 25 kDa, some of which are also produced by in-frame translation from various lengths of longer mRNA transcripts. Furthermore, such isoforms of Nrf1 may also be affected by other post-translational modifications, i.e. N-deglycosylation, O-GlcNAcylation, de-GlcNAcylation, phosphorylation, ubiquitination and proteolytic degradation, occurring in the cyto/nucleo-plasmic compartments.

Notably, the following five points should be taken into serious consideration. Firstly, the proper orientation of the intact membrane topology of Nrf1 is determined by the positive-inside rule (i.e. the net positively charged terminus of a transmembrane region is positioned to face the cyto/nucleo-plasmic side of ER membranes); this is defined by charge differences between two regions flanking the hydrophobic core buried within the membrane lipid bilayer [37,40,58,76], which abide by the membrane topology dogma [120,121,295,296]. Secondly, the AAA protein p97/VCP may provide the power to drive dynamic retrotranslocation of Nrf1 from the ER-luminal side of membranes into the cyto/nucleo-plasmic side [34,36,127]; this facilitates the selective processing of the CNC–bZIP protein to generate a deglycosylated 95-kDa and/or a cleaved 85-kDa isoform. The latter endoproteolytic cleavage is theoretically catalysed by an unidentified cytosolic protease (i.e. the 26S proteasome as a candidate), but not by the AAA protein p97/VCP itself. Thirdly, an intact non-glycosylated and/or deglycosylated 95-kDa Nrf1 protein (with electrophoretic mobility similar to that of p110-Nrf1 separated by pH 8.9 Laemmli SDS/PAGE [35,39,127]) is expressed de facto, but is not a bona fide cleaved isoform, in the cells (that had been treated particularly with ER stressor TU). Fourthly, variations in the electrophoretic locations of distinct Nrf1/TCF11 isoforms (including its intact prototypic full-length protein, glycoprotein, deglycoprotein and several cleaved isoforms) that have migrated on different PAGE gels [36,59,127] should be considered with extreme caution. Finally, differences between Nrf1 and other confusingly similar abbreviations (particularly NRF-1, which represents nuclear respiratory factor-1) that have emerged in the literature [201,245,289] should be seriously taken into account.

In future research, it is very exciting to determine the mechanisms underlying differential transcriptional expression of the single Nrf1 gene and subsequent post-transcriptional processing of its products into multiple mRNA transcripts under different cellular pathophysiological conditions. The translational expression patterns of distinct Nrf1 isoforms may also be controlled by uORF (which was predicted to exist in the intact full-length and longer Nrf1 mRNA transcripts, as shown in Figure 11B), in addition to Nrf1-interacting heterodimeric partners and upstream transcription factors (e.g. SREBP1, Pitx2/3, Sox9, Smad6 and Myb). As is the case for c-Myc, C/EBP or ATF4 [297299], the uORF could act as a molecular pathophysiological switch driving carcinogenesis or other degenerative diseases. It is envisaged that, under cellular stress, the uORFs in mouse Nrf1 and human TCF11 are translated into a 58-amino-acid polypeptide, called Nrf1u and TCF11u, which share 69% amino acid identity and 80% similarity. Moreover, Nrf1u is a putative repressor of Nrf1 possibly through sensing redox stress, binding metals and targeting to membranes, but this prediction requires confirmation.

We thank the many researchers who have contributed to the understanding of the CNC–bZIP Nrf subfamily of transcription factors, but also apologize to those whose work in this field we were unable to cite due to space restrictions and unintentional ignorance. We are very thankful to Professor Guy Salvesen, as Reviews Editor of the Biochemical Journal, and the two anonymous reviewers for providing their thoughtful suggestions to improve the quality of this work. We also gratefully acknowledge the kind help of Professor John D. Hayes and Dr Albena T. Dinkova-Kostova (at Ninewells Hospital and Medical School, University of Dundee, Dundee, U.K.) for providing invaluable information.

FUNDING

This work was supported by the National Natural Science Foundation of China (NSFC) [key programme grant numbers 91129703 and 91429305 and project grant number 31270879] awarded to Y.Z., but the funding source had no direct involvement in the research work.

Abbreviations

     
  • AAA

    ATPase associated with various cellular activities

  •  
  • AD

    acidic domain

  •  
  • AD2L

    AD2-like

  •  
  • Aldh1a1

    aldehyde dehydrogenase 1A1

  •  
  • AP1

    activator protein 1

  •  
  • ARE

    antioxidant-response element

  •  
  • ATF

    activating transcription factor

  •  
  • Bach

    BTB and CNC homology

  •  
  • BFA

    brefeldin A

  •  
  • BiP

    immunoglobulin heavy-chain-binding protein

  •  
  • BMS

    bone marrow stromal

  •  
  • bZIP

    basic-region leucine zipper

  •  
  • C/EBPβ

    CCAAT/enhancer-binding protein

  •  
  • CFSC

    CCl4-induced cirrhotic fat-sorting cell

  •  
  • CHD6

    chromo-ATPase/helicase DNA-binding protein 6

  •  
  • CHOP

    C/EBP-homologous protein

  •  
  • CNC

    cap'n’collar

  •  
  • CNS

    central nervous system

  •  
  • CPD

    Cdc4 phosphodegron

  •  
  • CRAC

    cholesterol-recognition/amino acid consensus motif

  •  
  • CRE

    cAMP-response element

  •  
  • CREB

    CRE-binding protein

  •  
  • CTD

    C-terminal domain

  •  
  • CYP

    cytochrome P450

  •  
  • DBD

    DNA-binding domain

  •  
  • dpc

    days post-coitus

  •  
  • DPP

    dentin phosphophoryn

  •  
  • DSP

    dentin sialoprotein

  •  
  • EGF

    epidermal growth factor

  •  
  • eIF2α

    eukaryotic initiation factor 2α

  •  
  • EpRE

    electrophile-response element

  •  
  • ER

    endoplasmic reticulum

  •  
  • ERAD

    ER-associated degradation

  •  
  • ERK

    extracellular-signal-regulated kinase

  •  
  • ESC

    embryonic stem cell

  •  
  • Fbw7

    F-box and WD repeat domain-containing 7

  •  
  • GCLC

    glutamate–cysteine ligase catalytic

  •  
  • GCLM

    glutamate–cysteine ligase modifier

  •  
  • GSIS

    glucose-stimulated insulin secretion

  •  
  • GSK-3

    glycogen synthase kinase 3

  •  
  • GSS

    glutathione synthetase

  •  
  • GST

    glutathione S-transferase

  •  
  • HNE

    4-hydroxynonenal

  •  
  • HO-1

    haem oxygenase 1

  •  
  • Hrd1

    3-hydroxy-3-methylglutaryl-CoA reductase degradation protein 1

  •  
  • INM

    inner nuclear membrane

  •  
  • INMAD

    INM-associated degradation

  •  
  • IRE1

    inositol-requiring enzyme 1

  •  
  • JUNQ

    juxtanuclear quality-control compartment

  •  
  • KD

    knockdown

  •  
  • Keap1

    Kelch-like ECH-associated protein 1

  •  
  • KO

    knockout

  •  
  • LCR

    locus control region

  •  
  • LCR-F1

    LCR factor 1

  •  
  • LDS

    lithium dodecyl sulfate

  •  
  • MAPK

    mitogen-activated protein kinase

  •  
  • 3-MC

    3-methylcholanthrene

  •  
  • MEF

    mouse embryonic fibroblast

  •  
  • mORF

    main ORF

  •  
  • MRP

    multidrug-resistance-associated protein

  •  
  • mTORC1

    mammalian (or mechanistic) target of rapamycin complex 1

  •  
  • NASH

    non-alcoholic steatohepatitis

  •  
  • NE

    nuclear envelope

  •  
  • Neh1L (etc.)

    Nrf2–ECH homology 1-like (etc.)

  •  
  • NF-E2

    nuclear factor-erythroid 2

  •  
  • NF-κB

    nuclear factor κB

  •  
  • NHB

    N-terminal homology

  •  
  • NLS

    nuclear localization signal

  •  
  • NQO1

    NAD(P):quinone oxidoreductase 1

  •  
  • Nrf

    NF-E2 p45-related factor

  •  
  • NTD

    N-terminal domain

  •  
  • ONM

    outer nuclear membrane

  •  
  • Osx

    osterix

  •  
  • PDK4

    pyruvate dehydrogenase kinase 4

  •  
  • PERK

    PKR (dsRNA-dependent protein kinase)-like ER kinase

  •  
  • PGC

    PPARγ co-activator

  •  
  • PI3K

    phosphoinositide 3-kinase

  •  
  • PKA

    protein kinase A

  •  
  • PKC

    protein kinase C

  •  
  • PPAR

    peroxisome-proliferator-activated receptor

  •  
  • ROS

    reactive oxygen species

  •  
  • Skn-1

    skinhead-1

  •  
  • sMaf

    small Maf (musculoaponeurotic fibrosarcoma retrovirus transforming gene)

  •  
  • SNP

    single nucleotide polymorphism

  •  
  • SR

    serine-repeat

  •  
  • SRE

    sterol-regulatory element

  •  
  • SREBP

    sterol-response-element-binding protein

  •  
  • TAD

    transactivation domain

  •  
  • tBHQ

    t-butylhydroquinone

  •  
  • TCA

    tricarboxylic acid

  •  
  • TCF11

    transcription factor 11

  •  
  • TG

    thapsigargin

  •  
  • TGFβ

    transforming growth factor β

  •  
  • TM

    transmembrane region

  •  
  • TMc

    transmembrane C-terminal region

  •  
  • TMi

    transmembrane intermediate region

  •  
  • TMp

    proline-kinked transmembrane hinge region

  •  
  • TNFα

    tumour necrosis factor α

  •  
  • TRE

    TPA (PMA)-response element

  •  
  • β-TrCP

    β-transducin repeat-containing protein

  •  
  • TU

    tunicamycin

  •  
  • uORF

    upstream ORF

  •  
  • UPR

    unfolded protein response

  •  
  • VCP

    valosin-containing protein

References

References
1
Stamler
J.S.
Lamas
S.
Fang
F.C.
Nitrosylation: the prototypic redox-based signaling mechanism
Cell
2001
, vol. 
106
 (pg. 
675
-
683
)
[PubMed]
2
Fang
Y.Z.
Yang
S.
Wu
G.
Free radicals, antioxidants, and nutrition
Nutrition
2002
, vol. 
18
 (pg. 
872
-
879
)
[PubMed]
3
Kensler
T.W.
Wakabayashi
N.
Biswal
S.
Cell survival responses to environmental stresses via the Keap1–Nrf2–ARE pathway
Annu. Rev. Pharmacol. Toxicol.
2007
, vol. 
47
 (pg. 
89
-
116
)
[PubMed]
4
Haddad
J.J.
Antioxidant and prooxidant mechanisms in the regulation of redox(y)-sensitive transcription factors
Cell. Signal.
2002
, vol. 
14
 (pg. 
879
-
897
)
[PubMed]
5
Liu
H.
Colavitti
R.
Rovira
I.I.
Finkel
T.
Redox-dependent transcriptional regulation
Circ. Res.
2005
, vol. 
97
 (pg. 
967
-
974
)
[PubMed]
6
Rushmore
T.H.
Morton
M.R.
Pickett
C.B.
The antioxidant responsive element: activation by oxidative stress and identification of the DNA consensus sequence required for functional activity
J. Biol. Chem.
1991
, vol. 
266
 (pg. 
11632
-
11639
)
[PubMed]
7
Kwak
M.K.
Wakabayashi
N.
Itoh
K.
Motohashi
H.
Yamamoto
M.
Kensler
T.W.
Modulation of gene expression by cancer chemopreventive dithiolethiones through the Keap1–Nrf2 pathway: identification of novel gene clusters for cell survival
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
8135
-
8145
)
[PubMed]
8
Nguyen
T.
Yang
C.S.
Pickett
C.B.
The pathways and molecular mechanisms regulating Nrf2 activation in response to chemical stress
Free Radic. Biol. Med.
2004
, vol. 
37
 (pg. 
433
-
441
)
[PubMed]
9
Xiao
H.
Lu
F.
Stewart
D.
Zhang
Y.
Mechanisms underlying chemopreventive effects of flavonoids via multiple signaling nodes within Nrf2–ARE and AhR–XRE gene regulatory networks
Curr. Chem. Biol.
2013
, vol. 
7
 (pg. 
151
-
176
)
10
Sasaki
H.
Sato
H.
Kuriyama-Matsumura
K.
Sato
K.
Maebara
K.
Wang
H.
Tamba
M.
Itoh
K.
Yamamoto
M.
Bannai
S.
Electrophile response element-mediated induction of the cystine/glutamate exchange transporter gene expression
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
44765
-
44771
)
[PubMed]
11
Vollrath
V.
Wielandt
A.M.
Iruretagoyena
M.
Chianale
J.
Role of Nrf2 in the regulation of the Mrp2 (ABCC2) gene
Biochem. J.
2006
, vol. 
395
 (pg. 
599
-
609
)
[PubMed]
12
Andressoo
J.O.
Hoeijmakers
J.H.
Transcription-coupled repair and premature ageing
Mutat. Res.
2005
, vol. 
577
 (pg. 
179
-
194
)
[PubMed]
13
Barzilai
A.
Yamamoto
K.
DNA damage responses to oxidative stress
DNA Repair (Amst.)
2004
, vol. 
3
 (pg. 
1109
-
1115
)
[PubMed]
14
Li
J.
Lee
J.M.
Johnson
J.A.
Microarray analysis reveals an antioxidant responsive element-driven gene set involved in conferring protection from an oxidative stress-induced apoptosis in IMR-32 cells
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
388
-
394
)
[PubMed]
15
Giudice
A.
Montella
M.
Activation of the Nrf2–ARE signaling pathway: a promising strategy in cancer prevention
BioEssays
2006
, vol. 
28
 (pg. 
169
-
181
)
[PubMed]
16
Beckman
K.B.
Ames
B.N.
Oxidative decay of DNA
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
19633
-
19636
)
[PubMed]
17
Lombard
D.B.
Chua
K.F.
Mostoslavsky
R.
Franco
S.
Gostissa
M.
Alt
F.W.
DNA repair, genome stability, and aging
Cell
2005
, vol. 
120
 (pg. 
497
-
512
)
[PubMed]
18
Hussain
S.P.
Hofseth
L.J.
Harris
C.C.
Radical causes of cancer
Nat. Rev. Cancer
2003
, vol. 
3
 (pg. 
276
-
285
)
[PubMed]
19
Valko
M.
Izakovic
M.
Mazur
M.
Rhodes
C.J.
Telser
J.
Role of oxygen radicals in DNA damage and cancer incidence
Mol. Cell. Biochem.
2004
, vol. 
266
 (pg. 
37
-
56
)
[PubMed]
20
Finkel
T.
Holbrook
N.J.
Oxidants, oxidative stress and the biology of ageing
Nature
2000
, vol. 
408
 (pg. 
239
-
247
)
[PubMed]
21
Balaban
R.S.
Nemoto
S.
Finkel
T.
Mitochondria, oxidants, and aging
Cell
2005
, vol. 
120
 (pg. 
483
-
495
)
[PubMed]
22
Sykiotis
G.P.
Bohmann
D.
Stress-activated cap'n’collar transcription factors in aging and human disease
Sci. Signal.
2010
, vol. 
3
 pg. 
re3
 
[PubMed]
23
Kwak
M.K.
Kensler
T.W.
Targeting NRF2 signaling for cancer chemoprevention
Toxicol. Appl. Pharmacol.
2011
, vol. 
244
 (pg. 
66
-
76
)
24
Zhang
Y.
Molecular and Cellular Control of the Nrf1 Transcription Factor: an Integral Membrane Glycoprotein
2009
Saarbrücken
Vdm Verlag Dr Müller Publishing House
25
Higgins
L.G.
Kelleher
M.O.
Eggleston
I.M.
Itoh
K.
Yamamoto
M.
Hayes
J.D.
Transcription factor Nrf2 mediates an adaptive response to sulforaphane that protects fibroblasts in vitro against the cytotoxic effects of electrophiles, peroxides and redox-cycling agents
Toxicol. Appl. Pharmacol.
2009
, vol. 
237
 (pg. 
267
-
280
)
[PubMed]
26
Chan
K.
Lu
R.
Chang
J.C.
Kan
Y.W.
NRF2, a member of the NFE2 family of transcription factors, is not essential for murine erythropoiesis, growth, and development
Proc. Natl. Acad. Sci. U.S.A.
1996
, vol. 
93
 (pg. 
13943
-
13948
)
[PubMed]
27
Xu
C.
Huang
M.T.
Shen
G.
Yuan
X.
Lin
W.
Khor
T.O.
Conney
A.H.
Kong
A.N.
Inhibition of 7,12-dimethylbenz(a)anthracene-induced skin tumorigenesis in C57BL/6 mice by sulforaphane is mediated by nuclear factor E2-related factor 2
Cancer Res.
2006
, vol. 
66
 (pg. 
8293
-
8296
)
[PubMed]
28
Leung
L.
Kwong
M.
Hou
S.
Lee
C.
Chan
J.Y.
Deficiency of the Nrf1 and Nrf2 transcription factors results in early embryonic lethality and severe oxidative stress
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
48021
-
48029
)
[PubMed]
29
Farmer
S.C.
Sun
C.W.
Winnier
G.E.
Hogan
B.L.
Townes
T.M.
The bZIP transcription factor LCR-F1 is essential for mesoderm formation in mouse development
Genes Dev.
1997
, vol. 
11
 (pg. 
786
-
798
)
[PubMed]
30
Chan
J.Y.
Kwong
M.
Lu
R.
Chang
J.
Wang
B.
Yen
T.S.
Kan
Y.W.
Targeted disruption of the ubiquitous CNC–bZIP transcription factor, Nrf-1, results in anemia and embryonic lethality in mice
EMBO J.
1998
, vol. 
17
 (pg. 
1779
-
1787
)
[PubMed]
31
Kwong
M.
Kan
Y.W.
Chan
J.Y.
The CNC basic leucine zipper factor, Nrf1, is essential for cell survival in response to oxidative stress-inducing agents: role for Nrf1 in γ-gcsL and gss expression in mouse fibroblasts
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
37491
-
37498
)
[PubMed]
32
Xu
Z.
Chen
L.
Leung
L.
Yen
T.S.
Lee
C.
Chan
J.Y.
Liver-specific inactivation of the Nrf1 gene in adult mouse leads to nonalcoholic steatohepatitis and hepatic neoplasia
Proc. Natl. Acad. Sci. U.S.A.
2005
, vol. 
102
 (pg. 
4120
-
4125
)
[PubMed]
33
Ohtsuji
M.
Katsuoka
F.
Kobayashi
A.
Aburatani
H.
Hayes
J.D.
Yamamoto
M.
Nrf1 and Nrf2 play distinct roles in activation of antioxidant response element-dependent genes
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
33554
-
33562
)
[PubMed]
34
Steffen
J.
Seeger
M.
Koch
A.
Kruger
E.
Proteasomal degradation is transcriptionally controlled by TCF11 via an ERAD-dependent feedback loop
Mol. Cell
2010
, vol. 
40
 (pg. 
147
-
158
)
[PubMed]
35
Radhakrishnan
S.K.
Lee
C.S.
Young
P.
Beskow
A.
Chan
J.Y.
Deshaies
R.J.
Transcription factor Nrf1 mediates the proteasome recovery pathway after proteasome inhibition in mammalian cells
Mol. Cell
2010
, vol. 
38
 (pg. 
17
-
28
)
[PubMed]
36
Sha
Z.
Goldberg
A.L.
Proteasome-mediated processing of Nrf1 is essential for coordinate induction of all proteasome subunits and p97
Curr. Biol.
2014
, vol. 
24
 (pg. 
1573
-
1583
)
[PubMed]
37
Zhang
Y.
Ren
Y.
Li
S.
Hayes
J.D.
Transcription factor Nrf1 is topologically repartitioned across membranes to enable target gene transactivation through its acidic glucose-responsive domains
PLoS One
2014
, vol. 
9
 pg. 
e93458
 
[PubMed]
38
Zhang
Y.
Crouch
D.H.
Yamamoto
M.
Hayes
J.D.
Negative regulation of the Nrf1 transcription factor by its N-terminal domain is independent of Keap1: Nrf1, but not Nrf2, is targeted to the endoplasmic reticulum
Biochem. J.
2006
, vol. 
399
 (pg. 
373
-
385
)
[PubMed]
39
Wang
W.
Chan
J.Y.
Nrf1 is targeted to the endoplasmic reticulum membrane by an N-terminal transmembrane domain: inhibition of nuclear translocation and transacting function
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
19676
-
19687
)
[PubMed]
40
Zhang
Y.
Lucocq
J.M.
Yamamoto
M.
Hayes
J.D.
The NHB1 (N-terminal homology box 1) sequence in transcription factor Nrf1 is required to anchor it to the endoplasmic reticulum and also to enable its asparagine-glycosylation
Biochem. J.
2007
, vol. 
408
 (pg. 
161
-
172
)
[PubMed]
41
Zhang
Y.
Lucocq
J.M.
Hayes
J.D.
The Nrf1 CNC/bZIP protein is a nuclear envelope-bound transcription factor that is activated by t-butyl hydroquinone but not by endoplasmic reticulum stressors
Biochem. J.
2009
, vol. 
418
 (pg. 
293
-
310
)
[PubMed]
42
deBoer
E.
Antoniou
M.
Mignotte
V.
Wall
L.
Grosveld
F.
The human β-globin promoter; nuclear protein factors and erythroid specific induction of transcription
EMBO J.
1988
, vol. 
7
 (pg. 
4203
-
4212
)
[PubMed]
43
Curtin
P.T.
Liu
D.P.
Liu
W.
Chang
J.C.
Kan
Y.W.
Human β-globin gene expression in transgenic mice is enhanced by a distant DNase I hypersensitive site
Proc. Natl. Acad. Sci. U.S.A.
1989
, vol. 
86
 (pg. 
7082
-
7086
)
[PubMed]
44
Whitelaw
E.
Tsai
S.F.
Hogben
P.
Orkin
S.H.
Regulated expression of globin chains and the erythroid transcription factor GATA-1 during erythropoiesis in the developing mouse
Mol. Cell. Biol.
1990
, vol. 
10
 (pg. 
6596
-
6606
)
[PubMed]
45
Liu
D.
Chang
J.C.
Moi
P.
Liu
W.
Kan
Y.W.
Curtin
P.T.
Dissection of the enhancer activity of β-globin 5′ DNase I-hypersensitive site 2 in transgenic mice
Proc. Natl. Acad. Sci. U.S.A.
1992
, vol. 
89
 (pg. 
3899
-
3903
)
[PubMed]
46
Andrews
N.C.
Erdjument-Bromage
H.
Davidson
M.B.
Tempst
P.
Orkin
S.H.
Erythroid transcription factor NF-E2 is a haematopoietic-specific basic-leucine zipper protein
Nature
1993
, vol. 
362
 (pg. 
722
-
728
)
[PubMed]
47
Moi
P.
Chan
K.
Asunis
I.
Cao
A.
Kan
Y.W.
Isolation of NF-E2-related factor 2 (Nrf2), a NF-E2-like basic leucine zipper transcriptional activator that binds to the tandem NF-E2/AP1 repeat of the β-globin locus control region
Proc. Natl. Acad. Sci. U.S.A.
1994
, vol. 
91
 (pg. 
9926
-
9930
)
[PubMed]
48
Kotkow
K.J.
Orkin
S.H.
Complexity of the erythroid transcription factor NF-E2 as revealed by gene targeting of the mouse p18 NF-E2 locus
Proc. Natl. Acad. Sci. U.S.A.
1996
, vol. 
93
 (pg. 
3514
-
3518
)
[PubMed]
49
Bean
T.L.
Ney
P.A.
Multiple regions of p45 NF-E2 are required for β-globin gene expression in erythroid cells
Nucleic Acids Res.
1997
, vol. 
25
 (pg. 
2509
-
2515
)
[PubMed]
50
Andrews
N.C.
Kotkow
K.J.
Ney
P.A.
Erdjument-Bromage
H.
Tempst
P.
Orkin
S.H.
The ubiquitous subunit of erythroid transcription factor NF-E2 is a small basic-leucine zipper protein related to the v-maf oncogene
Proc. Natl. Acad. Sci. U.S.A.
1993
, vol. 
90
 (pg. 
11488
-
11492
)
[PubMed]
51
Andrews
N.C.
The NF-E2 transcription factor
Int. J. Biochem. Cell Biol.
1998
, vol. 
30
 (pg. 
429
-
432
)
[PubMed]
52
Igarashi
K.
Kataoka
K.
Itoh
K.
Hayashi
N.
Nishizawa
M.
Yamamoto
M.
Regulation of transcription by dimerization of erythroid factor NF-E2 p45 with small Maf proteins
Nature
1994
, vol. 
367
 (pg. 
568
-
572
)
[PubMed]
53
Chan
J.Y.
Han
X.L.
Kan
Y.W.
Cloning of Nrf1, an NF-E2-related transcription factor, by genetic selection in yeast
Proc. Natl. Acad. Sci. U.S.A.
1993
, vol. 
90
 (pg. 
11371
-
11375
)
[PubMed]
54
McKie
J.
Johnstone
K.
Mattei
M.G.
Scambler
P.
Cloning and mapping of murine Nfe2l1
Genomics
1995
, vol. 
25
 (pg. 
716
-
719
)
[PubMed]
55
Caterina
J.J.
Donze
D.
Sun
C.W.
Ciavatta
D.J.
Townes
T.M.
Cloning and functional characterization of LCR-F1: a bZIP transcription factor that activates erythroid-specific, human globin gene expression
Nucleic Acids Res.
1994
, vol. 
22
 (pg. 
2383
-
2391
)
[PubMed]
56
Luna
L.
Johnsen
O.
Skartlien
A.H.
Pedeutour
F.
Turc-Carel
C.
Prydz
H.
Kolsto
A.B.
Molecular cloning of a putative novel human bZIP transcription factor on chromosome 17q22
Genomics
1994
, vol. 
22
 (pg. 
553
-
562
)
[PubMed]
57
Husberg
C.
Murphy
P.
Martin
E.
Kolsto
A.B.
Two domains of the human bZIP transcription factor TCF11 are necessary for transactivation
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
17641
-
17652
)
[PubMed]
58
Zhang
Y.
Hayes
J.D.
The membrane-topogenic vectorial behaviour of Nrf1 controls its post-translational modification and transactivation activity
Sci. Rep.
2013
, vol. 
3
 pg. 
2006
 
[PubMed]
59
Zhang
Y.
Li
S.
Xiang
Y.
Qiu
L.
Zhao
H.
Hayes
J.D.
The selective post-translational processing of Nrf1 yields distinct isoforms that dictate its activity to differentially regulate target genes
Sci. Rep.
2015
, vol. 
5
 pg. 
e12983
 
60
Zhang
Y.
Lu
Q.
Li
S.
Xiang
Y.
Chen
J.
Ren
Y.
The C-terminal domain (CTD) of Nrf1 negatively regulates this full-length CNC–bZIP factor and its shorter form Nrf1β/LCR-F1; both are also inhibited by the small dominant-negative Nrf1γ/δ isoforms that down-regulates ARE-battery gene expression
PLoS One
2014
, vol. 
9
 pg. 
e109159
 
[PubMed]
61
Wang
W.
Kwok
A.M.
Chan
J.Y.
The p65 isoform of Nrf1 is a dominant negative inhibitor of ARE-mediated transcription
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
24670
-
24678
)
[PubMed]
62
Novotny
V.
Prieschl
E.E.
Csonga
R.
Fabjani
G.
Baumruker
T.
Nrf1 in a complex with fosB, c-jun, junD and ATF2 forms the AP1 component at the TNFα promoter in stimulated mast cells
Nucleic Acids Res.
1998
, vol. 
26
 (pg. 
5480
-
5485
)
[PubMed]
63
Prieschl
E.E.
Novotny
V.
Csonga
R.
Jaksche
D.
Elbe-Burger
A.
Thumb
W.
Auer
M.
Stingl
G.
Baumruker
T.
A novel splice variant of the transcription factor Nrf1 interacts with the TNFα promoter and stimulates transcription
Nucleic Acids Res.
1998
, vol. 
26
 (pg. 
2291
-
2297
)
[PubMed]
64
Schultz
M.A.
Hagan
S.S.
Datta
A.
Zhang
Y.
Freeman
M.L.
Sikka
S.C.
Abdel-Mageed
A.B.
Mondal
D.
Nrf1 and nrf2 transcription factors regulate androgen receptor transactivation in prostate cancer cells
PLoS One
2014
, vol. 
9
 pg. 
e87204
 
[PubMed]
65
Walker
A.K.
See
R.
Batchelder
C.
Kophengnavong
T.
Gronniger
J.T.
Shi
Y.
Blackwell
T.K.
A conserved transcription motif suggesting functional parallels between Caenorhabditis elegans SKN-1 and Cap'n’Collar-related basic leucine zipper proteins
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
22166
-
22171
)
[PubMed]
66
Kwong
E.K.
Kim
K.M.
Penalosa
P.J.
Chan
J.Y.
Characterization of Nrf1b, a novel isoform of the nuclear factor-erythroid-2 related transcription factor-1 that activates antioxidant response element-regulated genes
PLoS One
2012
, vol. 
7
 pg. 
e48404
 
[PubMed]
67
Chepelev
N.L.
Bennitz
J.D.
Huang
T.
McBride
S.
Willmore
W.G.
The Nrf1 CNC–bZIP protein is regulated by the proteasome and activated by hypoxia
PLoS One
2011
, vol. 
6
 pg. 
e29167
 
[PubMed]
68
Luna
L.
Skammelsrud
N.
Johnsen
O.
Abel
K.J.
Weber
B.L.
Prydz
H.
Kolsto
A.B.
Structural organization and mapping of the human TCF11 gene
Genomics
1995
, vol. 
27
 (pg. 
237
-
244
)
[PubMed]
69
McMahon
M.
Itoh
K.
Yamamoto
M.
Chanas
S.A.
Henderson
C.J.
McLellan
L.I.
Wolf
C.R.
Cavin
C.
Hayes
J.D.
The Cap'n’Collar basic leucine zipper transcription factor Nrf2 (NF-E2 p45-related factor 2) controls both constitutive and inducible expression of intestinal detoxification and glutathione biosynthetic enzymes
Cancer Res.
2001
, vol. 
61
 (pg. 
3299
-
3307
)
[PubMed]
70
Williams
L.M.
Timme-Laragy
A.R.
Goldstone
J.V.
McArthur
A.G.
Stegeman
J.J.
Smolowitz
R.M.
Hahn
M.E.
Developmental expression of the Nfe2-related factor (Nrf) transcription factor family in the zebrafish, Danio rerio
PLoS One
2013
, vol. 
8
 pg. 
e79574
 
[PubMed]
71
Virbasius
C.A.
Virbasius
J.V.
Scarpulla
R.C.
NRF-1, an activator involved in nuclear-mitochondrial interactions, utilizes a new DNA-binding domain conserved in a family of developmental regulators
Genes Dev.
1993
, vol. 
7
 (pg. 
2431
-
2445
)
[PubMed]
72
Becker
T.S.
Burgess
S.M.
Amsterdam
A.H.
Allende
M.L.
Hopkins
N.
not really finished is crucial for development of the zebrafish outer retina and encodes a transcription factor highly homologous to human nuclear respiratory factor-1 and avian initiation binding repressor
Development
1998
, vol. 
125
 (pg. 
4369
-
4378
)
[PubMed]
73
Perez-Garcia
A.
Snoeijers
S.S.
Joosten
M.H.
Goosen
T.
De Wit
P.J.
Expression of the Avirulence gene Avr9 of the fungal tomato pathogen Cladosporium fulvum is regulated by the global nitrogen response factor NRF1
Mol. Plant Microbe Interact.
2001
, vol. 
14
 (pg. 
316
-
325
)
[PubMed]
74
Murray
J.M.
Johnson
D.I.
Isolation and characterization of Nrf1p, a novel negative regulator of the Cdc42p GTPase in Schizosaccharomyces pombe
Genetics
2000
, vol. 
154
 (pg. 
155
-
165
)
[PubMed]
75
Zhang
Y.
Crouch
D.H.
Hayes
J.D.
Identification of functional domains in the p45 nuclear factor-erythroid 2-related transcription factor 1 (Nrf1)
The 10th SCBA International Symposium
2004
Beijing, China
pg. 
247
  
18–23 July 2004
76
Zhang
Y.
Hayes
J.D.
Identification of topological determinants in the N-terminal domain of transcription factor Nrf1 that control its orientation in the endoplasmic reticulum membrane
Biochem. J.
2010
, vol. 
430
 (pg. 
497
-
510
)
[PubMed]
77
Rupert
P.B.
Daughdrill
G.W.
Bowerman
B.
Matthews
B.W.
A new DNA-binding motif in the Skn-1 binding domain-DNA complex
Nat. Struct. Biol.
1998
, vol. 
5
 (pg. 
484
-
491
)
[PubMed]
78
Johnsen
O.
Murphy
P.
Prydz
H.
Kolsto
A.B.
Interaction of the CNC–bZIP factor TCF11/LCR-F1/Nrf1 with MafG: binding-site selection and regulation of transcription
Nucleic Acids Res.
1998
, vol. 
26
 (pg. 
512
-
520
)
[PubMed]
79
Itoh
K.
Wakabayashi
N.
Katoh
Y.
Ishii
T.
Igarashi
K.
Engel
J.D.
Yamamoto
M.
Keap1 represses nuclear activation of antioxidant responsive elements by Nrf2 through binding to the amino-terminal Neh2 domain
Genes Dev.
1999
, vol. 
13
 (pg. 
76
-
86
)
[PubMed]
80
Johnsen
O.
Skammelsrud
N.
Luna
L.
Nishizawa
M.
Prydz
H.
Kolsto
A.B.
Small Maf proteins interact with the human transcription factor TCF11/Nrf1/LCR-F1
Nucleic Acids Res.
1996
, vol. 
24
 (pg. 
4289
-
4297
)
[PubMed]
81
Venugopal
R.
Jaiswal
A.K.
Nrf1 and Nrf2 positively and c-Fos and Fra1 negatively regulate the human antioxidant response element-mediated expression of NAD(P)H:quinone oxidoreductase1 gene
Proc. Natl. Acad. Sci. U.S.A.
1996
, vol. 
93
 (pg. 
14960
-
14965
)
[PubMed]
82
Nioi
P.
McMahon
M.
Itoh
K.
Yamamoto
M.
Hayes
J.D.
Identification of a novel Nrf2-regulated antioxidant response element (ARE) in the mouse NAD(P)H:quinone oxidoreductase 1 gene: reassessment of the ARE consensus sequence
Biochem. J.
2003
, vol. 
374
 (pg. 
337
-
348
)
[PubMed]
83
Kimura
M.
Yamamoto
T.
Zhang
J.
Itoh
K.
Kyo
M.
Kamiya
T.
Aburatani
H.
Katsuoka
F.
Kurokawa
H.
Tanaka
T.
, et al. 
Molecular basis distinguishing the DNA binding profile of Nrf2–Maf heterodimer from that of Maf homodimer
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
33681
-
33690
)
[PubMed]
84
Kurokawa
H.
Motohashi
H.
Sueno
S.
Kimura
M.
Takagawa
H.
Kanno
Y.
Yamamoto
M.
Tanaka
T.
Structural basis of alternative DNA recognition by Maf transcription factors
Mol. Cell. Biol.
2009
, vol. 
29
 (pg. 
6232
-
6244
)
[PubMed]
85
Motohashi
H.
O'Connor
T.
Katsuoka
F.
Engel
J.D.
Yamamoto
M.
Integration and diversity of the regulatory network composed of Maf and CNC families of transcription factors
Gene
2002
, vol. 
294
 (pg. 
1
-
12
)
[PubMed]
86
Yamamoto
T.
Kyo
M.
Kamiya
T.
Tanaka
T.
Engel
J.D.
Motohashi
H.
Yamamoto
M.
Predictive base substitution rules that determine the binding and transcriptional specificity of Maf recognition elements
Genes Cells
2006
, vol. 
11
 (pg. 
575
-
591
)
[PubMed]
87
Venugopal
R.
Jaiswal
A.K.
Nrf2 and Nrf1 in association with Jun proteins regulate antioxidant response element-mediated expression and coordinated induction of genes encoding detoxifying enzymes
Oncogene
1998
, vol. 
17
 (pg. 
3145
-
3156
)
[PubMed]
88
Yang
H.
Magilnick
N.
Lee
C.
Kalmaz
D.
Ou
X.
Chan
J.Y.
Lu
S.C.
Nrf1 and Nrf2 regulate rat glutamate–cysteine ligase catalytic subunit transcription indirectly via NF-κB and AP-1
Mol. Cell. Biol.
2005
, vol. 
25
 (pg. 
5933
-
5946
)
[PubMed]
89
Newman
J.R.
Keating
A.E.
Comprehensive identification of human bZIP interactions with coiled-coil arrays
Science
2003
, vol. 
300
 (pg. 
2097
-
2101
)
[PubMed]
90
Husberg
C.
Murphy
P.
Bjorgo
E.
Kalland
K.H.
Kolsto
A.B.
Cellular localisation and nuclear export of the human bZIP transcription factor TCF11
Biochim. Biophys. Acta
2003
, vol. 
1640
 (pg. 
143
-
151
)
[PubMed]
91
Narayanan
K.
Ramachandran
A.
Peterson
M.C.
Hao
J.
Kolsto
A.B.
Friedman
A.D.
George
A.
The CCAAT enhancer-binding protein (C/EBP)β and Nrf1 interact to regulate dentin sialophosphoprotein (DSPP) gene expression during odontoblast differentiation
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
45423
-
45432
)
[PubMed]
92
Marini
M.G.
Chan
K.
Casula
L.
Kan
Y.W.
Cao
A.
Moi
P.
hMAF, a small human transcription factor that heterodimerizes specifically with Nrf1 and Nrf2
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
16490
-
16497
)
[PubMed]
93
Nioi
P.
Nguyen
T.
Sherratt
P.J.
Pickett
C.B.
The carboxy-terminal neh3 domain of Nrf2 is required for transcriptional activation
Mol. Cell. Biol.
2005
, vol. 
25
 (pg. 
10895
-
10906
)
[PubMed]
94
Tsuchiya
Y.
Taniguchi
H.
Ito
Y.
Morita
T.
Karim
M.R.
Ohtake
N.
Fukagai
K.
Ito
T.
Okamuro
S.
Iemura
S.
, et al. 
The CK2–Nrf1 axis controls the clearance of ubiquitinated proteins by regulating proteasome gene expression
Mol. Cell. Biol.
2013
, vol. 
33
 (pg. 
3461
-
3472
)
[PubMed]
95
Huang
C.F.
Wang
Y.C.
Tsao
D.A.
Tung
S.F.
Lin
Y.S.
Wu
C.W.
Antagonism between members of the CNC–bZIP family and the immediate-early protein IE2 of human cytomegalovirus
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
12313
-
12320
)
[PubMed]
96
Wu
J.L.
Lin
Y.S.
Yang
C.C.
Lin
Y.J.
Wu
S.F.
Lin
Y.T.
Huang
C.F.
MCRS2 represses the transactivation activities of Nrf1
BMC Cell Biol.
2009
, vol. 
10
 pg. 
9
 
[PubMed]
97
McMahon
M.
Itoh
K.
Yamamoto
M.
Hayes
J.D.
Keap1-dependent proteasomal degradation of transcription factor Nrf2 contributes to the negative regulation of antioxidant response element-driven gene expression
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
21592
-
21600
)
[PubMed]
98
Kobayashi
A.
Kang
M.I.
Okawa
H.
Ohtsuji
M.
Zenke
Y.
Chiba
T.
Igarashi
K.
Yamamoto
M.
Oxidative stress sensor Keap1 functions as an adaptor for Cul3-based E3 ligase to regulate proteasomal degradation of Nrf2
Mol. Cell. Biol.
2004
, vol. 
24
 (pg. 
7130
-
7139
)
[PubMed]
99
Tong
K.I.
Kobayashi
A.
Katsuoka
F.
Yamamoto
M.
Two-site substrate recognition model for the Keap1–Nrf2 system: a hinge and latch mechanism
Biol. Chem.
2006
, vol. 
387
 (pg. 
1311
-
1320
)
[PubMed]
100
Padmanabhan
B.
Tong
K.I.
Ohta
T.
Nakamura
Y.
Scharlock
M.
Ohtsuji
M.
Kang
M.I.
Kobayashi
A.
Yokoyama
S.
Yamamoto
M.
Structural basis for defects of Keap1 activity provoked by its point mutations in lung cancer
Mol. Cell
2006
, vol. 
21
 (pg. 
689
-
700
)
[PubMed]
101
McMahon
M.
Thomas
N.
Itoh
K.
Yamamoto
M.
Hayes
J.D.
Dimerization of substrate adaptors can facilitate cullin-mediated ubiquitylation of proteins by a “tethering” mechanism: a two-site interaction model for the Nrf2–Keap1 complex
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
24756
-
24768
)
[PubMed]
102
Lo
S.C.
Li
X.
Henzl
M.T.
Beamer
L.J.
Hannink
M.
Structure of the Keap1:Nrf2 interface provides mechanistic insight into Nrf2 signaling
EMBO J.
2006
, vol. 
25
 (pg. 
3605
-
3617
)
[PubMed]
103
Ogura
T.
Tong
K.I.
Mio
K.
Maruyama
Y.
Kurokawa
H.
Sato
C.
Yamamoto
M.
Keap1 is a forked-stem dimer structure with two large spheres enclosing the intervening, double glycine repeat, and C-terminal domains
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
2842
-
2847
)
[PubMed]
104
Sykiotis
G.P.
Bohmann
D.
Keap1/Nrf2 signaling regulates oxidative stress tolerance and lifespan in Drosophila
Dev. Cell
2008
, vol. 
14
 (pg. 
76
-
85
)
[PubMed]
105
Hochmuth
C.E.
Biteau
B.
Bohmann
D.
Jasper
H.
Redox regulation by Keap1 and Nrf2 controls intestinal stem cell proliferation in Drosophila
Cell Stem Cell
2011
, vol. 
8
 (pg. 
188
-
199
)
[PubMed]
106
Tsuchiya
Y.
Morita
T.
Kim
M.
Iemura
S.
Natsume
T.
Yamamoto
M.
Kobayashi
A.
Dual regulation of the transcriptional activity of Nrf1 by β-TrCP- and Hrd1-dependent degradation mechanisms
Mol. Cell. Biol.
2011
, vol. 
31
 (pg. 
4500
-
4512
)
[PubMed]
107
Biswas
M.
Phan
D.
Watanabe
M.
Chan
J.Y.
The Fbw7 tumor suppressor regulates nuclear factor E2 related factor 1 (Nrf1) transcription factor turnover through proteasome-mediated proteolysis
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
39282
-
39289
)
[PubMed]
108
Zhao
R.
Hou
Y.
Xue
P.
Woods
C.G.
Fu
J.
Feng
B.
Guan
D.
Sun
G.
Chan
J.Y.
Waalkes
M.P.
, et al. 
Long isoforms of NRF1 contribute to arsenic-induced antioxidant response in human keratinocytes
Environ. Health Perspect.
2011
, vol. 
119
 (pg. 
56
-
62
)
[PubMed]
109
Zhao
R.
Hou
Y.
Zhang
Q.
Woods
C.G.
Xue
P.
Fu
J.
Yarborough
K.
Guan
D.
Andersen
M.E.
Pi
J.
Cross-regulations among NRFs and KEAP1 and effects of their silencing on arsenic-induced antioxidant response and cytotoxicity in human keratinocytes
Environ. Health Perspect.
2012
, vol. 
120
 (pg. 
583
-
589
)
[PubMed]
110
McMahon
M.
Thomas
N.
Itoh
K.
Yamamoto
M.
Hayes
J.D.
Redox-regulated turnover of Nrf2 is determined by at least two separate protein domains, the redox-sensitive Neh2 degron and the redox-insensitive Neh6 degron
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
31556
-
31567
)
[PubMed]
111
Huang
H.C.
Nguyen
T.
Pickett
C.B.
Phosphorylation of Nrf2 at Ser-40 by protein kinase C regulates antioxidant response element-mediated transcription
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
42769
-
42774
)
[PubMed]
112
Katoh
Y.
Itoh
K.
Yoshida
E.
Miyagishi
M.
Fukamizu
A.
Yamamoto
M.
Two domains of Nrf2 cooperatively bind CBP, a CREB binding protein, and synergistically activate transcription
Genes Cells
2001
, vol. 
6
 (pg. 
857
-
868
)
[PubMed]
113
Zhang
J.
Hosoya
T.
Maruyama
A.
Nishikawa
K.
Maher
J.M.
Ohta
T.
Motohashi
H.
Fukamizu
A.
Shibahara
S.
Itoh
K.
, et al. 
Nrf2 Neh5 domain is differentially utilized in the transactivation of cytoprotective genes
Biochem. J.
2007
, vol. 
404
 (pg. 
459
-
466
)
[PubMed]
114
Kim
J.H.
Yu
S.
Chen
J.D.
Kong
A.N.
The nuclear cofactor RAC3/AIB1/SRC-3 enhances Nrf2 signaling by interacting with transactivation domains
Oncogene
2012
, vol. 
32
 (pg. 
514
-
527
)
[PubMed]
115
Wang
H.
Liu
K.
Geng
M.
Gao
P.
Wu
X.
Hai
Y.
Li
Y.
Li
Y.
Luo
L.
Hayes
J.D.
, et al. 
RXRα inhibits the NRF2–ARE signaling pathway through a direct interaction with the Neh7 domain of NRF2
Cancer Res.
2013
, vol. 
73
 (pg. 
3097
-
3108
)
[PubMed]
116
Biswas
M.
Kwong
E.K.
Park
E.
Nagra
P.
Chan
J.Y.
Glycogen synthase kinase 3 regulates expression of nuclear factor-erythroid-2 related transcription factor-1 (Nrf1) and inhibits pro-survival function of Nrf1
Exp. Cell Res.
2013
, vol. 
319
 (pg. 
1922
-
1931
)
[PubMed]
117
Chowdhry
S.
Zhang
Y.
McMahon
M.
Sutherland
C.
Cuadrado
A.
Hayes
J.D.
Nrf2 is controlled by two distinct β-TrCP recognition motifs in its Neh6 domain, one of which can be modulated by GSK-3 activity
Oncogene
2013
, vol. 
32
 (pg. 
3765
-
3781
)
[PubMed]
118
An
J.H.
Vranas
K.
Lucke
M.
Inoue
H.
Hisamoto
N.
Matsumoto
K.
Blackwell
T.K.
Regulation of the Caenorhabditis elegans oxidative stress defense protein SKN-1 by glycogen synthase kinase-3
Proc. Natl. Acad. Sci. U.S.A.
2005
, vol. 
102
 (pg. 
16275
-
16280
)
[PubMed]
119
Chen
J.
Liu
X.
Lu
F.
Liu
X.
Ru
Y.
Ren
Y.
Yao
L.
Zhang
Y.
Transcription factor Nrf1 is negatively regulated by its O-GlcNAcylation status
FEBS Lett.
2015
, vol. 
589
 (pg. 
2347
-
2358
)
[PubMed]
120
von Heijne
G.
Membrane-protein topology
Nat. Rev. Mol. Cell Biol.
2006
, vol. 
7
 (pg. 
909
-
918
)
[PubMed]
121
Dowhan
W.
Bogdanov
M.
Lipid-dependent membrane protein topogenesis
Annu. Rev. Biochem.
2009
, vol. 
78
 (pg. 
515
-
540
)
[PubMed]
122
Nouhi
Z.
Chevillard
G.
Derjuga
A.
Blank
V.
Endoplasmic reticulum association and N-linked glycosylation of the human Nrf3 transcription factor
FEBS Lett.
2007
, vol. 
581
 (pg. 
5401
-
5406
)
[PubMed]
123
Zhang
Y.
Kobayashi
A.
Yamamoto
M.
Hayes
J.D.
The Nrf3 transcription factor is a membrane-bound glycoprotein targeted to the endoplasmic reticulum through its N-terminal homology box 1 sequence
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
3195
-
3210
)
[PubMed]
124
Grimberg
K.B.
Beskow
A.
Lundin
D.
Davis
M.M.
Young
P.
Basic leucine zipper protein Cnc-C is a substrate and transcriptional regulator of the Drosophila 26S proteasome
Mol. Cell. Biol.
2011
, vol. 
31
 (pg. 
897
-
909
)
[PubMed]
125
Staab
T.A.
Griffen
T.C.
Corcoran
C.
Evgrafov
O.
Knowles
J.A.
Sieburth
D.
The conserved SKN-1/Nrf2 stress response pathway regulates synaptic function in Caenorhabditis elegans
PLoS Genet.
2013
, vol. 
9
 pg. 
e1003354
 
[PubMed]
126
Glover-Cutter
K.M.
Lin
S.
Blackwell
T.K.
Integration of the unfolded protein and oxidative stress responses through SKN-1/Nrf
PLoS Genet.
2013
, vol. 
9
 pg. 
e1003701
 
[PubMed]
127
Radhakrishnan
S.K.
den Besten
W.
Deshaies
R.J.
p97-dependent retrotranslocation and proteolytic processing govern formation of active Nrf1 upon proteasome inhibition
eLife
2014
, vol. 
3
 pg. 
e01856
 
[PubMed]
128
Johnson
P.R.
Swanson
R.
Rakhilina
L.
Hochstrasser
M.
Degradation signal masking by heterodimerization of MATα2 and MATa1 blocks their mutual destruction by the ubiquitin–proteasome pathway
Cell
1998
, vol. 
94
 (pg. 
217
-
227
)
[PubMed]
129
Zattas
D.
Hochstrasser
M.
Ubiquitin-dependent protein degradation at the yeast endoplasmic reticulum and nuclear envelope
Crit. Rev. Biochem. Mol. Biol.
2015
, vol. 
50
 (pg. 
1
-
17
)
[PubMed]
130
Kacena
M.A.
Gundberg
C.M.
Nelson
T.
Horowitz
M.C.
Loss of the transcription factor p45 NF-E2 results in a developmental arrest of megakaryocyte differentiation and the onset of a high bone mass phenotype
Bone
2005
, vol. 
36
 (pg. 
215
-
223
)
[PubMed]
131
Kobayashi
A.
Ito
E.
Toki
T.
Kogame
K.
Takahashi
S.
Igarashi
K.
Hayashi
N.
Yamamoto
M.
Molecular cloning and functional characterization of a new Cap'n’collar family transcription factor Nrf3
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
6443
-
6452
)
[PubMed]
132
Etchevers
H.C.
The cap ‘n’ collar family member NF-E2-related factor 3 (Nrf3) is expressed in mesodermal derivatives of the avian embryo
Int. J. Dev. Biol.
2005
, vol. 
49
 (pg. 
363
-
367
)
[PubMed]
133
Derjuga
A.
Gourley
T.S.
Holm
T.M.
Heng
H.H.
Shivdasani
R.A.
Ahmed
R.
Andrews
N.C.
Blank
V.
Complexity of CNC transcription factors as revealed by gene targeting of the Nrf3 locus
Mol. Cell. Biol.
2004
, vol. 
24
 (pg. 
3286
-
3294
)
[PubMed]
134
Murphy
P.
Kolsto
A.
Expression of the bZIP transcription factor TCF11 and its potential dimerization partners during development
Mech. Dev.
2000
, vol. 
97
 (pg. 
141
-
148
)
[PubMed]
135
Chen
L.
Kwong
M.
Lu
R.
Ginzinger
D.
Lee
C.
Leung
L.
Chan
J.Y.
Nrf1 is critical for redox balance and survival of liver cells during development
Mol. Cell. Biol.
2003
, vol. 
23
 (pg. 
4673
-
4686
)
[PubMed]
136
Oh
D.H.
Rigas
D.
Cho
A.
Chan
J.Y.
Deficiency in the nuclear-related factor erythroid 2 transcription factor (Nrf1) leads to genetic instability
FEBS J.
2012
, vol. 
279
 (pg. 
4121
-
4130
)
[PubMed]
137
Lee
C.S.
Ho
D.V.
Chan
J.Y.
Nuclear factor-erythroid 2-related factor 1 regulates expression of proteasome genes in hepatocytes and protects against endoplasmic reticulum stress and steatosis in mice
FEBS J.
2013
, vol. 
280
 (pg. 
3609
-
3620
)
[PubMed]
138
Tsujita
T.
Peirce
V.
Baird
L.
Matsuyama
Y.
Takaku
M.
Walsh
S.V.
Griffin
J.L.
Uruno
A.
Yamamoto
M.
Hayes
J.D.
Transcription factor Nrf1 negatively regulates the cystine/glutamate transporter and lipid-metabolizing enzymes
Mol. Cell. Biol.
2014
, vol. 
34
 (pg. 
3800
-
3816
)
[PubMed]
139
Masuoka
H.C.
Townes
T.M.
Targeted disruption of the activating transcription factor 4 gene results in severe fetal anemia in mice
Blood
2002
, vol. 
99
 (pg. 
736
-
745
)
[PubMed]
140
Shavit
J.A.
Motohashi
H.
Onodera
K.
Akasaka
J.
Yamamoto
M.
Engel
J.D.
Impaired megakaryopoiesis and behavioral defects in mafG-null mutant mice
Genes Dev.
1998
, vol. 
12
 (pg. 
2164
-
2174
)
[PubMed]
141
Onodera
K.
Shavit
J.A.
Motohashi
H.
Katsuoka
F.
Akasaka
J.E.
Engel
J.D.
Yamamoto
M.
Characterization of the murine mafF gene
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
21162
-
21169
)
[PubMed]
142
Yamazaki
H.
Katsuoka
F.
Motohashi
H.
Engel
J.D.
Yamamoto
M.
Embryonic lethality and fetal liver apoptosis in mice lacking all three small Maf proteins
Mol. Cell. Biol.
2012
, vol. 
32
 (pg. 
808
-
816
)
[PubMed]
143
Hilberg
F.
Aguzzi
A.
Howells
N.
Wagner
E.F.
c-jun is essential for normal mouse development and hepatogenesis
Nature
1993
, vol. 
365
 (pg. 
179
-
181
)
[PubMed]
144
Eferl
R.
Sibilia
M.
Hilberg
F.
Fuchsbichler
A.
Kufferath
I.
Guertl
B.
Zenz
R.
Wagner
E.F.
Zatloukal
K.
Functions of c-Jun in liver and heart development
J. Cell. Biol.
1999
, vol. 
145
 (pg. 
1049
-
1061
)
[PubMed]
145
Behrens
A.
Sibilia
M.
David
J.P.
Mohle-Steinlein
U.
Tronche
F.
Schutz
G.
Wagner
E.F.
Impaired postnatal hepatocyte proliferation and liver regeneration in mice lacking c-jun in the liver
EMBO J.
2002
, vol. 
21
 (pg. 
1782
-
1790
)
[PubMed]
146
Meixner
A.
Karreth
F.
Kenner
L.
Penninger
J.M.
Wagner
E.F.
Jun and JunD-dependent functions in cell proliferation and stress response
Cell Death Differ.
2010
, vol. 
17
 (pg. 
1409
-
1419
)
[PubMed]
147
Lee
C.S.
Lee
C.
Hu
T.
Nguyen
J.M.
Zhang
J.
Martin
M.V.
Vawter
M.P.
Huang
E.J.
Chan
J.Y.
Loss of nuclear factor E2-related factor 1 in the brain leads to dysregulation of proteasome gene expression and neurodegeneration
Proc. Natl. Acad. Sci. U.S.A.
2011
, vol. 
108
 (pg. 
8408
-
8413
)
[PubMed]
148
Hertel
M.
Braun
S.
Durka
S.
Alzheimer
C.
Werner
S.
Upregulation and activation of the Nrf-1 transcription factor in the lesioned hippocampus
Eur. J. Neurosci.
2002
, vol. 
15
 (pg. 
1707
-
1711
)
[PubMed]
149
Shivdasani
R.A.
Orkin
S.H.
Erythropoiesis and globin gene expression in mice lacking the transcription factor NF-E2
Proc. Natl. Acad. Sci. U.S.A.
1995
, vol. 
92
 (pg. 
8690
-
8694
)
[PubMed]
150
Itoh
K.
Chiba
T.
Takahashi
S.
Ishii
T.
Igarashi
K.
Katoh
Y.
Oyake
T.
Hayashi
N.
Satoh
K.
Hatayama
I.
, et al. 
An Nrf2/small Maf heterodimer mediates the induction of phase II detoxifying enzyme genes through antioxidant response elements
Biochem. Biophys. Res. Commun.
1997
, vol. 
236
 (pg. 
313
-
322
)
[PubMed]
151
Sun
J.
Hoshino
H.
Takaku
K.
Nakajima
O.
Muto
A.
Suzuki
H.
Tashiro
S.
Takahashi
S.
Shibahara
S.
Alam
J.
, et al. 
Hemoprotein Bach1 regulates enhancer availability of heme oxygenase-1 gene
EMBO J.
2002
, vol. 
21
 (pg. 
5216
-
5224
)
[PubMed]
152
Kobayashi
A.
Ohta
T.
Yamamoto
M.
Unique function of the Nrf2–Keap1 pathway in the inducible expression of antioxidant and detoxifying enzymes
Methods Enzymol.
2004
, vol. 
378
 (pg. 
273
-
286
)
[PubMed]
153
Muto
A.
Tashiro
S.
Nakajima
O.
Hoshino
H.
Takahashi
S.
Sakoda
E.
Ikebe
D.
Yamamoto
M.
Igarashi
K.
The transcriptional programme of antibody class switching involves the repressor Bach2
Nature
2004
, vol. 
429
 (pg. 
566
-
571
)
[PubMed]
154
Kobayashi
A.
Tsukide
T.
Miyasaka
T.
Morita
T.
Mizoroki
T.
Saito
Y.
Ihara
Y.
Takashima
A.
Noguchi
N.
Fukamizu
A.
, et al. 
Central nervous system-specific deletion of transcription factor Nrf1 causes progressive motor neuronal dysfunction
Genes Cells
2011
, vol. 
16
 (pg. 
692
-
703
)
[PubMed]
155
Katsuoka
F.
Motohashi
H.
Tamagawa
Y.
Kure
S.
Igarashi
K.
Engel
J.D.
Yamamoto
M.
Small Maf compound mutants display central nervous system neuronal degeneration, aberrant transcription, and Bach protein mislocalization coincident with myoclonus and abnormal startle response
Mol. Cell. Biol.
2003
, vol. 
23
 (pg. 
1163
-
1174
)
[PubMed]
156
Franceschi
R.T.
Iyer
B.S.
Cui
Y.
Effects of ascorbic acid on collagen matrix formation and osteoblast differentiation in murine MC3T3-E1 cells
J. Bone Miner. Res.
1994
, vol. 
9
 (pg. 
843
-
854
)
[PubMed]
157
Hadzir
S.N.
Ibrahim
S.N.
Abdul Wahab
R.M.
Zainol Abidin
I.Z.
Senafi
S.
Ariffin
Z.Z.
Abdul Razak
M.
Zainal Ariffin
S.H.
Ascorbic acid induces osteoblast differentiation of human suspension mononuclear cells
Cytotherapy
2014
, vol. 
16
 (pg. 
674
-
682
)
[PubMed]
158
Xing
W.
Singgih
A.
Kapoor
A.
Alarcon
C.M.
Baylink
D.J.
Mohan
S.
Nuclear factor-E2-related factor-1 mediates ascorbic acid induction of osterix expression via interaction with antioxidant-responsive element in bone cells
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
22052
-
22061
)
[PubMed]
159
Mohan
S.
Kapoor
A.
Singgih
A.
Zhang
Z.
Taylor
T.
Yu
H.
Chadwick
R.B.
Chung
Y.S.
Donahue
L.R.
Rosen
C.
, et al. 
Spontaneous fractures in the mouse mutant sfx are caused by deletion of the gulonolactone oxidase gene, causing vitamin C deficiency
J. Bone Miner. Res.
2005
, vol. 
20
 (pg. 
1597
-
1610
)
[PubMed]
160
Hinoi
E.
Fujimori
S.
Wang
L.
Hojo
H.
Uno
K.
Yoneda
Y.
Nrf2 negatively regulates osteoblast differentiation via interfering with Runx2-dependent transcriptional activation
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
18015
-
18024
)
[PubMed]
161
Hinoi
E.
Takarada
T.
Fujimori
S.
Wang
L.
Iemata
M.
Uno
K.
Yoneda
Y.
Nuclear factor E2 p45-related factor 2 negatively regulates chondrogenesis
Bone
2007
, vol. 
40
 (pg. 
337
-
344
)
[PubMed]
162
Kim
J.
Xing
W.
Wergedal
J.
Chan
J.Y.
Mohan
S.
Targeted disruption of nuclear factor erythroid-derived 2-like 1 in osteoblasts reduces bone size and bone formation in mice
Physiol. Genomics
2010
, vol. 
40
 (pg. 
100
-
110
)
[PubMed]
163
Chevillard
G.
Nouhi
Z.
Anna
D.
Paquet
M.
Blank
V.
Nrf3-deficient mice are not protected against acute lung and adipose tissue damages induced by butylated hydroxytoluene
FEBS Lett.
2010
, vol. 
584
 (pg. 
923
-
928
)
[PubMed]
164
Biswas
M.
Chan
J.Y.
Role of Nrf1 in antioxidant response element-mediated gene expression and beyond
Toxicol. Appl. Pharmacol.
2010
, vol. 
244
 (pg. 
16
-
20
)
[PubMed]
165
Hayes
J.D.
McMahon
M.
The double-edged sword of Nrf2: subversion of redox homeostasis during the evolution of cancer
Mol. Cell
2006
, vol. 
21
 (pg. 
732
-
734
)
[PubMed]
166
Koch
A.
Steffen
J.
Kruger
E.
TCF11 at the crossroads of oxidative stress and the ubiquitin proteasome system
Cell Cycle
2011
, vol. 
10
 (pg. 
1200
-
1207
)
[PubMed]
167
Hayes
J.D.
McMahon
M.
Chowdhry
S.
Dinkova-Kostova
A.T.
Cancer chemoprevention mechanisms mediated through the Keap1–Nrf2 pathway
Antioxid. Redox Signal.
2010
, vol. 
13
 (pg. 
1713
-
1748
)
[PubMed]
168
Hayes
J.D.
McMahon
M.
NRF2 and KEAP1 mutations: permanent activation of an adaptive response in cancer
Trends Biochem. Sci.
2009
, vol. 
34
 (pg. 
176
-
188
)
[PubMed]
169
Wakabayashi
N.
Itoh
K.
Wakabayashi
J.
Motohashi
H.
Noda
S.
Takahashi
S.
Imakado
S.
Kotsuji
T.
Otsuka
F.
Roop
D.R.
, et al. 
Keap1-null mutation leads to postnatal lethality due to constitutive Nrf2 activation
Nat. Genet.
2003
, vol. 
35
 (pg. 
238
-
245
)
[PubMed]
170
Myhrstad
M.C.
Husberg
C.
Murphy
P.
Nordstrom
O.
Blomhoff
R.
Moskaug
J.O.
Kolsto
A.B.
TCF11/Nrf1 overexpression increases the intracellular glutathione level and can transactivate the γ-glutamylcysteine synthetase (GCS) heavy subunit promoter
Biochim. Biophys. Acta
2001
, vol. 
1517
 (pg. 
212
-
219
)
[PubMed]
171
Hayes
J.D.
Strange
R.C.
Glutathione S-transferase polymorphisms and their biological consequences
Pharmacology
2000
, vol. 
61
 (pg. 
154
-
166
)
[PubMed]
172
Zhang
H.
Liu
H.
Iles
K.E.
Liu
R.M.
Postlethwait
E.M.
Laperche
Y.
Forman
H.J.
4-Hydroxynonenal induces rat γ-glutamyl transpeptidase through MAPK mediated EpRE/Nrf2 signaling
Am. J. Respir. Cell Mol. Biol.
2006
, vol. 
34
 (pg. 
174
-
181
)
[PubMed]
173
Angeloni
C.
Motori
E.
Fabbri
D.
Malaguti
M.
Leoncini
E.
Lorenzini
A.
Hrelia
S.
H2O2 preconditioning modulates phase II enzymes through p38 MAPK and PI3K/Akt activation
Am. J. Physiol. Heart. Circ. Physiol.
2011
, vol. 
300
 (pg. 
H2196
-
H2205
)
[PubMed]
174
Lee
E.K.
Kim
J.A.
Park
S.J.
Kim
J.K.
Heo
K.
Yang
K.M.
Son
T.G.
Low-dose radiation activates Nrf1/2 through reactive species and the Ca2+/ERK1/2 signaling pathway in human skin fibroblast cells
BMB Rep.
2013
, vol. 
46
 (pg. 
258
-
263
)
[PubMed]
175
Chepelev
N.L.
Enikanolaiye
M.I.
Chepelev
L.L.
Almohaisen
A.
Chen
Q.
Scoggan
K.A.
Coughlan
M.C.
Cao
X.L.
Jin
X.
Willmore
W.G.
Bisphenol A activates the Nrf1/2-antioxidant response element pathway in HEK 293 cells
Chem. Res. Toxicol.
2013
, vol. 
26
 (pg. 
498
-
506
)
[PubMed]
176
Chepelev
N.L.
Zhang
H.
Liu
H.
McBride
S.
Seal
A.J.
Morgan
T.E.
Finch
C.E.
Willmore
W.G.
Davies
K.J.
Forman
H.J.
Competition of nuclear factor-erythroid 2 factors related transcription factor isoforms, Nrf1 and Nrf2, in antioxidant enzyme induction
Redox Biol.
2013
, vol. 
1
 (pg. 
183
-
189
)
[PubMed]
177
Song
M.O.
Mattie
M.D.
Lee
C.H.
Freedman
J.H.
The role of Nrf1 and Nrf2 in the regulation of copper-responsive transcription
Exp. Cell Res.
2014
, vol. 
322
 (pg. 
39
-
50
)
[PubMed]
178
Siegel
D.
Bolton
E.M.
Burr
J.A.
Liebler
D.C.
Ross
D.
The reduction of α-tocopherolquinone by human NAD(P)H: quinone oxidoreductase: the role of α-tocopherolhydroquinone as a cellular antioxidant
Mol. Pharmacol.
1997
, vol. 
52
 (pg. 
300
-
305
)
[