E2F6 is widely expressed in human tissues and cell lines. Recent studies have demonstrated its involvement in developmental patterning and in the regulation of various genes implicated in chromatin remodelling. Despite a growing number of studies, nothing is really known concerning the E2F6 expression regulation. To understand how cells control E2F6 expression, we analysed the activity of the previously cloned promoter region of the human E2F6 gene. DNase I footprinting, gel electrophoreticmobility shift, transient transfection and site-directed mutagenesis experiments allowed the identification of two functional NRF-1/α-PAL (nuclear respiratory factor-1/α-palindrome-binding protein)-binding sites within the human E2F6 core promoter region, which are conserved in the mouse and rat E2F6 promoter region. Moreover, ChIP (chromatin immunoprecipitation) analysis demonstrated that overexpressed NRF-1/α-PAL is associated in vivo with the E2F6 promoter. Furthermore, overexpression of full-length NRF-1/α-PAL enhanced E2F6 promoter activity, whereas expression of its dominant-negative form reduced the promoter activity. Our results indicate that NRF-1/α-PAL is implicated in the regulation of basal E2F6 gene expression.
E2F6 is a member of the E2F transcription factor family that plays a key role in the regulation of cellular proliferation and differentiation via target genes involved in DNA replication, DNA repair, cell cycle control and apoptosis [1–3]. In mammals, the E2F family consists of seven E2F members (E2F1–E2F7) and two distantly related DP members (DP1–DP2), which form heterodimers to generate functional E2F complexes and regulate transcription from a consensus sequence TTTSSCGC [4,5]. E2Fs members can be grouped into four groups based on their structure, affinity for pRB (retinoblastoma susceptibility protein) family members (pRB, p107 and p130) and functions. E2F6, which lacks the pocket protein-binding domain and the acidic transactivation domain common to E2F1–E2F5 proteins, forms the third E2F subgroup [6–8]. While E2F1–5 proteins can mediate either activation or repression depending upon which proteins associate with their C-terminal domain, E2F6 and E2F7 are only known to mediate repression of E2F target genes. The transcriptionally repressive properties of E2F6 are mainly supported by its C-terminal repression domain , which binds components of the mammalian PcG (polycomb group proteins) complex and recruit histone deacetylase activity [9,10]. Overexpression of E2F6 suppresses the E2F activity on E2F reporter constructs and is able to repress the activity of synthetic reporter constructs when fused to the corresponding heterologous DNA-binding domain [7,8]. Moreover, ectopic expression of E2F6 leads to the accumulation of cells in S-phase, and delays re-entry of quiescent cells into the cell cycle [8,11]. In addition to potential roles in cell proliferation and quiescence depending of the cell type, E2F6 protein is critical for developmental patterning, as revealed by knock-out experiments. Mice lacking E2F6, similarly to some PcG mutant mice, display posterior homoeotic transformations of the axial skeleton . While these transformations arise generally when Hox genes are mis-expressed as a result of a lack of PcG-dependent repression, it was tempting to speculate that E2F6 contributes to Hox genes promoter regulation. Nevertheless, recent characterization of E2F6 target genes using a combination of ChIP (chromatin immunoprecipitation) and genomic micro-arrays failed to identify Hox genes, but identified numerous genes involved in tumour suppression and maintenance of chromatin structure (brca1, ctip, art27, hp1-alpha, rbap48) . Although considerable information is now available concerning the mechanism underlying E2F6 repression, little is known about the regulatory mechanisms that might affect E2F6 expression. We have therefore recently cloned and characterized the promoter region of the gene encoding human and mouse E2F6 [14,15]. To clarify the molecular mechanisms controlling E2F6 expression, we analysed more precisely the human E2F6 promoter activity and identified the NRF-1/α-PAL transcription factor as an E2F6 promoter regulator.
NRF-1/α-PAL (nuclear respiratory factor-1/α-palindrome-binding protein), named NRF-1 hereafter, was concomitantly characterized as an activator of the eukaryotic initiation factor 2α  and cytochrome c expression . It was subsequently found to act on many nuclear genes required for mitochondrial respiratory function (reviewed in ). This predominant role was confirmed by disrupting the NRF-1 gene in mice, which results in a peri-implantation lethal phenotype and a striking decrease in the mitochondrial DNA content in NRF-1−/− blastocysts . More recently, the regulation by NRF-1 of the chemokine receptor CXCR4 [20,21], the GluR2 subunit of the AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) subtypes of glutamate receptors , the CD155 human polio virus receptor  and the CD47 or IAP (integrin-associated protein)  has been demonstrated, suggesting a broad role for NRF-1 in the transcriptional modulation of genes implicated in various cellular functions.
In the present study, we reported the characterization of two NRF-1 sites spaced 16 bp apart within the E2F6 core promoter at positions −11/+1 and +18/+29 that are essential for basal promoter activity. These sites, close to the major transcription start sites, are conserved in the promoters of human, mouse and rat E2F6 genes. DNase I footprinting analysis and EMSA (electrophoretic mobility-shift assay) revealed that these sites are bound by NRF-1 present in nuclear extracts, and we demonstrated by ChIP that NRF-1 was associated with the regulatory region of the E2F6 gene in vivo. Moreover, integrity of both sites was found to be critical for efficient basal transcription of the E2F6 gene, as well as for NRF-1 activation or DN-NRF-1 (dominant-negative NRF-1) repression of E2F6.
Human bone osteosarcoma U2OS (A.T.C.C. number HTB-96) and mammary gland adenocarcinoma MDA-MB-468 (A.T.C.C. number HTB-132) cell cultures were routinely maintained in DMEM (Dulbecco's modified Eagle's medium) with 10% fetal calf serum and kept at 37 °C in a water-saturated 5% CO2 atmosphere.
DNase I footprinting
DNase I footprinting was performed using the Core Footprint kit according to manufacturer's recommendations (Promega). The −62 to +103 E2F6 promoter region subcloned in the PGL3-basic vector (Promega) was linearized by HindIII or Asp718 digestion and labelled with [γ-32P]ATP and T4 polynucleotide kinase (New England Biolabs). The single end-labelled DNA was then digested with Asp718 or HindIII and the 220-bp fragment purified on a native 6% polyacrylamide non-denaturating gel. The DNase I protection assays were performed using 30000 cpm of labelled probe in a 50 μl volume of binding reaction, containing 1 μg of poly[d(I-C)] and 10 μg of MDA-MB-468 nuclear extracts prepared according to the method of Dignam et al. . After a 30-min incubation on ice, 50 μl of a solution (room temperature −22 °C) of 5 mM CaCl2 and 10 mM MgCl2 was added to each reaction and incubated for 2 min at room temperature. DNase I (3 μl, 100 ng/ml) was then added and incubated for 3 min at room temperature. DNase I digestion was stopped by the addition of 90 μl of stop solution (200 mM NaCl, 30 mM EDTA and 1% SDS), and the DNA template was purified by phenol/chloroform (1:1) extraction and ethanol precipitation. Samples were analysed by electrophoresis through denaturing 6% polyacrylamide/urea sequencing gels with a 25 bp DNA ladder (Invitrogen) and an M13 forward sequencing reaction as a ladder.
Mutant E2F6 promoters were constructed from the parent plasmids (−1614/+290)pGL3 and (−31/+103)pGL3  using the QuikChange site-directed mutagenesis kit (Stratagene) and the oligonucleotides F6(−31/+8)Mut1, 5′-TGGGAGCGCTCCGGCAGCGGCGGTCATTCGCAGAGGGGG-3′, and F6(+9/+38)Mut2, 5′-GCGGTGTACTGCTCATTCGGGAAGATGGCG-3′.
NRF-1 and DN-NRF-1 expression vectors were constructed by PCR with Pfu polymerase (Stratagene) using the following oligonucleotides: 5′-FLAG NRF, 5′-GATCAGATCTGAGGAACACGGAGTGACCCAAACCGAAC-3′; 3′-NRF, 5′-GATCTCTAGATTACTGTTCCAATGTCACCACCTCCAC-3′; and 3′-DN-NRF, 5′-GATCTCTAGATTAGACTACAGTCTGTGATGGTACAAGATG-3′. The BglII/XbaI-digested PCR products were then ligated into the BamHI/XbaI sites of the pcDNA3.2-FLAG–N-term vector (Invitrogen) to obtain the plasmids pcDNA3-FLAG–NRF-1 and pcDNA3-FLAG–DN-NRF-1. All constructs were verified by sequencing.
Multiple alignments were performed using ClustalW (http://www.infobiogen.fr/) with genomic sequences for human E2F6 (GenBank® accession number AY551345), mouse E2F6 (GenBank® accession number AF393244) and rat E2F6 (LocusLink number 313978).
Transient transfections and promoter activity assays
Transfections were performed by the Ex-Gen 500 (Euromedex) procedure in 12-well plates. For co-transfection, 400 ng of reporter plasmid were transfected with increasing amounts of the NRF-1- or DN-NRF-1-expressing plasmids, supplemented with appropriate quantities of the empty vector to maintain constant amount of DNA. Cells were harvested 24 h later and luciferase assays were performed (Promega). All the assays were normalized to the β-galactosidase activity of a CMV (cytomegalovirus)–β-galactosidase vector (25 ng). Experiments were performed at least twice using two different plasmid preparations.
EMSAs were performed using IvT (in vitro-translated) proteins, total extracts of U2OS transfected cells or MDA-MB-468 nuclear extracts, as described previously . Briefly, extracts were incubated for 20 min in a final volume of 20 μl containing 20 mM Hepes, pH 7.9, 20% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol, 3 μg of sheared salmon sperm DNA, 50 mM NaCl and 50000 cpm of probe. The reactions were then loaded on to a 5% polyacrylamide gel in 0.5×TBE (Tris/borate/EDTA) and run for 4 h at 170 V (at 4 °C). When indicated, a 200-fold molar excess of unlabelled double-stranded oligonucleotides was added. The following oligonucleotides were employed in binding assays after hybridization to obtain the corresponding DNA duplex: NRF-1cons5′, 5′-ATGCTAGCCCGCATGCGCGCGCACCTT-3′ (mutated version: TTA) and NRF-1cons3′, 5′-AAGGTGCGCGCGCATGCGGGCTAGCAT-3′; F6(−31/+8)5′, 5′-TGGGAGCGCTCCGGCAGCGGCGGGCATGCGCAGAGGGGG-3′ (mutated version: TCATT) and F6(−31/+8)3′, 5′-CCCCCTCTGCGCATGCCCGCCGCTGCCGGAGCGCTCCCA-3′; F6(+9/+38)-5′, 5′-GCGGTGTACTGCGCATGCGGGAAGATGGCG-3′ (mutated version: TCATT) and F6(+9/+38)3′, 5′-CGCCATCTTCCCGCATGCGCAGTACACCGC-3′; MYCcons 5′, 5′-CCCCACCACGTGGTGCCTGACACGTG-3′ (mutated version: TTGA) and MYCcons 3′, 5′-CACGTGTCAGGCACCACGTGGTGGGG-3′; SP1cons 5′, 5′-ATTCGATCGGGGCGGGGCGAGC-3′ (mutated version: TT) and SP1cons 3′, 5′-GCTCGCCCCGCCCCGATCGAAT-3′; E2Fcons 5′, 5′-ATTTAAGTTTCGCGCCCTTTCTCAA-3′ (mutated version: AT) and E2Fcons 3′, 5′-TTGAGAAAGGGCGCGAAACTTAAAT-3′. Anti-FLAG-M2 (1 μl; Sigma) or 2 μl of anti-(cytochrome c) (PharMingen), as a nonspecific antibody, were added to binding reactions for super-shift experiments.
In vivo detection of E2F6 promoter-associated NRF-1 was performed by ChIP, as described previously , on U2OS cells overexpressing the FLAG–NRF-1 protein. The FLAG–NRF-1-containing complexes were immunoprecipitated with a polyclonal anti-FLAG antibody (Santa Cruz, D-8), as well as with a pool M2 and M5 anti-FLAG monoclonal antibodies (Sigma–Aldrich). The anti-DBD-Gal4 (DNA-binding domain of Gal4) monoclonal RK5C1 antibody (Santa Cruz) and the anti-ICAM-1 (intercellular cell-adhesion molecule 1) M-19 polyclonal antibody (Santa Cruz) were used to control the immunoprecipitation specificity. The following oligonucleotides were designed to amplify by PCR a 150 bp fragment encompassing the −39/+111 region of the human E2F6 promoter: 5′E2F6p, 5′-TCGGTGCGTGGGAGCGCTCCGGC-3′, and 3′E2F6p, 5′-CTCACGTGCCCGGGAGCTCCCGAC-3′. Specific primers were also designed to amplify a 146-bp fragment of the sixth exon of the human E2F6 gene as a negative control: 5′-E2F6ex6, 5′-CTTTGTCATCTGTTAACTC-3′, and 3′E2F6ex6, 5′-TCTGATCTTAGCAATTTTGG-3′. Finally, specific primers were designed to amplify a 146-bp fragment of the human FMR1 (fragile X mental retardation 1) promoter as a positive control: 5′FMR1p, 5′-CGAGGCAGTGCGACCTGTCAC-3′, and 3′FMR1p, 5′-CTCTTCAAGTGGCCTGGGAGC-3′.
DNase I footprinting analysis of the E2F6 proximal promoter reveals protection of two potential NRF-1 sites
We previously published the initial characterization of human and mouse E2F6 promoter [14,15]. Transient transfection studies allowed us to map a 134 bp human core promoter fragment located between −31 and +103 and a 130 bp mouse core promoter fragment located between −55 and +75 with regards to the transcription start site. These two GC-rich promoter regions devoid of TATA box or CAAT box led to multiple transcriptional start sites [14,15]. A comparison of the 200 bp promoter region surrounding the transcriptional start site of the human and mouse genes with the sequence of the rat E2F6 gene found in the databank (LocusLink 313978) reveals a high degree of overall sequence conservation (60% identity) along with several short stretches of absolute identity (Figure 1). Among them, the comparison allow us to identify putative NRF-1 sites with core sequences perfectly conserved. In order to determine if the phylogenetic conservation of the NRF-1 sites is consistent with a functional role in controlling E2F6 expression, we analysed the DNA–protein interactions of the minimal human E2F6 promoter by DNase I footprint analysis. Nuclear extracts from MDA-MB-468 cells were tested for their ability to protect end-labelled DNA promoter fragment (−31 to +103) from digestion with DNase I. As shown in Figure 2(A), four distinct regions (named FPI–FPIV) were found to be protected in a reproducible manner on both the sense and antisense strands by proteins present in the extracts.
Sequence alignment of the human, mouse and rat E2F6 promoter
Identification of protein-binding sites in the human E2F6 promoter by DNase I footprinting
Scanning of several databases for transcription factor binding motifs matching with the observed protected regions was negative. However, two protected regions, FPI and FPIII partially overlapped sequences which share obvious sequence similarity with the NRF-1 transcription factor binding site. The optimal NRF-1 binding site is (T/C)GCGCA(C/T)GCGC(A/G) , whereas the human E2F6 promoter harbours the elements CGGGCATGCGCA (nucleotides −11 to +1) and TGCGCATGCGGG (nucleotides +18 to +29). The results showed that bases −11 to −5 of the more upstream putative NRF-1 binding site were protected against DNase I digestion on the coding strand, whereas the −5 to +1 residues were protected on the complementary strand. On the other hand, the +20 to +24 residues of the second putative NRF-1 binding site were protected against DNase I digestion on both the coding and the non-coding strands, whereas the +25 to +29 residues of the site were protected only on the coding strand (Figure 2B).
NRF-1 binds NRF-1-binding sites on the human E2F6 promoter
To demonstrate that the DNase I footprint of the FPI and FPIII regions resulted from NRF-1 binding, we performed EMSA experiments. Double-stranded oligonucleotides spanning each putative NRF-1 binding site were therefore synthesized and radiolabelled. EMSAs were performed with IvT human recombinant FLAG–NRF-1 proteins. As expected, a major band of a protein–DNA complex was observed when the IvT NRF-1 proteins were incubated with the −31/+8 and the +9/+38 probes (Figure 3A). In both cases, the complex was competed by a 200-fold molar excess of an unlabelled double-stranded oligonucleotide containing the functional NRF-1 consensus site from the rat somatic cytochrome c promoter (−173 to −147) , but not by an oligonucleotide containing a mutated NRF-1 binding site. The binding on each probe of the IvT NRF-1 proteins was unaffected by competition with a non-specific consensus binding site oligonucleotide (E-box). Moreover, addition of anti-FLAG antibodies diminished the initial complex formation and led to the apparition of a supershifted complex, thus confirming the presence of FLAG–NRF1 in the complexes. Non-specific IvT E2F6–DP1 proteins were also added to confirm the specificity of the observed DNA–NRF-1 complex. So, these results indicate that the two putative NRF-1 binding sites present in the E2F6 core promoter can recruit NRF-1 proteins.
Binding to E2F6 promoter elements −31 to +8 and +9 to +38
To further demonstrate that these sequences are effectively recognized by cellular NRF-1 proteins, we performed EMSA with MDA-MB-468 and U2OS cell nuclear extracts. Similar results were obtained with both cell extracts and results presented thereafter are those obtained with U2OS cell nuclear extracts. As shown in Figure 3(B) (left panel), incubation of the (−31 to +8) probe with the U2OS cell nuclear extracts resulted in the formation of two distinct protein–DNA complexes. The slower migrating complex was eliminated by addition of a 200-fold excess of unlabelled oligonucleotides containing the rat cytochrome c NRF-1 consensus site or E2F6 −31/+8 probe, but not by a mutated version of these sites. Moreover, incubation of U2OS cell nuclear extracts with mutated −31/+8 probe, in which the core CATG of the NRF-1 site was replaced by AATT, results in a complete loss of the slower migrating complex, suggesting that the latter is a genuine NRF-1 complex. On the other hand, the faster migrating complex seems to be non-specific, since it was strongly competed by either the wild-type and the mutated NRF-1 unlabelled oligonucleotides, and was unaffected by mutation of the NRF-1 site on the −31/+8 probe. Similarly, incubation of the +9/+38 probe with nuclear extracts yields two protein–DNA complexes (Figure 3B, right panel). Competition with a 200-fold excess of intact, as well as mutated, NRF-binding site failed to abolish the faster migrating complex, indicating that it is probably a nonspecific complex. By contrast, the slower migrating complex contains NRF-1 proteins, whereas its formation is prevented by the addition of an unlabelled wild-type NRF-1 oligonucleotide, but is unaffected by the mutated one. Moreover, the binding is lost when the NRF-1 binding site was mutated in the +9/+38 probe.
The NRF-1 binding sequence in the E2F6 FPIII region (included in the +9/+38 probe) overlapped in part with an E2F binding site (Figure 2B). Moreover, both NRF-1 binding sequences present in the E2F6 promoter regions −31/+8 and +9/+38 include a non-canonical Myc–Max-binding site [CA(C/T)GCG] that may be targeted by c-Myc, as demonstrated for the cytochrome c promoter . In order to determine whether the complexes observed with nuclear extracts on the E2F6 probes contain these transcription factors, competition experiments were performed with consensus E2F, E-box or SP1 unlabelled oligonucleotides. In fact, none of these consensus oligonucleotides affect the binding observed on the E2F6–NRF-1 probes. Therefore, it seems that NRF-1 is the major component of the slower migrating complexes formed with U2OS nuclear extracts on −31/+8 and +9/+38 probes.
Specificity of NRF-1 binding was further confirmed by supershift experiments. Total cellular extracts (1 μg) from U2OS cells transfected with plasmids encoding the FLAG–NRF-1 protein or the FLAG–DN-NRF-1 [28,29] were tested for their DNA-binding activity on the E2F6–NRF-1 probes. As shown in Figure 3(C), this quantity of cellular extracts from cells transfected with an empty vector (A3–FLAG) was insufficient to detect the endogenous NRF-1-binding activity, as observed in Figure 3(B). In contrast, extracts from cells expressing FLAG–NRF-1 protein or FLAG—DN-NRF-1 led to the formation of a major binding complex with the −31/+8 probe (Figure 3B, left panel) or the +9/+38 probe (right panel). The observed complexes were supershifted by the anti-FLAG antibody, but not by the control anti-(cytochrome c) antibody assessing their specificity. Taken together, these results confirm that NRF-1 proteins associate with the NRF-1 binding sites on nucleotides −11 to +1 and nucleotides +18 to +29 of the minimal E2F6 promoter.
NRF-1 transactivates the human E2F6 promoter
To determine the functional significance of NRF-1 binding on the E2F6 promoter activity, we next performed transient transfection experiments of U2OS cells. The activity of the wild-type p(−1614/+290) and the minimal p(−31/+103) E2F6–luciferase reporter constructs were compared with the activity of their corresponding versions mutated on the (−11/+1)-NRF-1 binding site (M1), on the (+9/+29)-NRF-1 binding site (M2) or on both NRF-1 binding sites (M1+2) (Figure 4A). Results, expressed as fold activity relative to that obtained with the empty pGL3-Basic plasmid, indicated that the M1 and M2 mutations reduced respectively the p(−1614/+290) activity by about 50% and 30%. Mutation of the two sites led only to a 60% reduction in the p(−1614/+290) activity. Interestingly, the introduction of the same mutations in the minimal promoter construct p(−31/+103) resulted in a more pronounced effect on the basal promoter activity, with about 80% decrease in the promoter activity for each single mutation. Likewise, combined mutation resulted in a nearly complete loss of the p(−31/+103) transcriptional activity. These results indicate that NRF-1 binding sites at positions −11 to +1 and +18 to +29 of the E2F6 promoter are needed for optimal basal activity of the E2F6 promoter. Nevertheless, the more pronounced effect of mutation of these sites for the p(−31/+103) construct, as compared with p(−1614/+290), may indicate the presence of other important regulatory elements for the promoter activity. Similar results were obtained in Cos-1 and RK-13 cells (results not shown).
Functional analysis of the human E2F6 gene promoter
To investigate the ability of NRF-1 to potentiate E2F6 transcription, co-transfections of the NRF-1 expression vector (FLAG–NRF-1) were performed in U2OS cells with the wild-type E2F6 p(−1614/+290) or p(−31/+103) constructs, as well as with the corresponding mutated promoter constructs. As shown in Figure 4(B), NRF-1 co-transfection induced a 2-fold increase of the p(−1614/+290) and p(−31/+103) activity. In contrast, mutation of the NRF-1 elements in the same constructs completely eliminated the NRF-1 effect.
Further evidence of the requirement of NRF-1 for efficient E2F6 promoter activity were obtained by co-transfection of the FLAG–DN-NRF-1 expression vector. The p(−1614/+290) and p(−31/+103) constructs promoter activity is dose-dependently inhibited upon expression of FLAG–DN-NRF-1, with around 75% inhibition observed at 5 ng (Figure 4C). Our results clearly indicate that the NRF-1 elements present in the E2F6 promoter are at least partially responsible for its transcriptional activity and may be implicated in its regulation during physiological processes.
NRF-1 is associated in vivo with the E2F6 promoter
To confirm the involvement of NRF-1 in the E2F6 promoter control, we performed ChIP experiments upon U2OS cell transfection with the FLAG–NRF-1 expressing vector, as described previously . As shown in Figure 5, the cross-linked E2F6 promoter–FLAG–NRF-1 complexes immunoprecipitated with either a polyclonal anti-FLAG antibody (D8) or a mix of two monoclonal anti-FLAG antibodies (M2/M5) were detected by PCR amplification with oligonucleotides spanning the −39 to +111 region of the human E2F6 promoter. As a positive control, in vivo association of FLAG–NRF-1 with the FMR1 promoter, a well-characterized NRF-1 target gene , was also detected by PCR with specific primers on the same samples. PCR amplification using primers spanning the sixth exon of E2F6 gene, which does not possess NRF-1-binding sites, revealed a similar level of non-specific DNA contamination for all the samples, and provides a further control for the specificity of the interaction between NRF-1 and the E2F6 promoter. As a negative control, ChIP with non-specific antibodies directed against ICAM-1 or DBD-Gal4 resulted in the absence of signal for the E2F6 promoter, as well as for FMR1. These results demonstrated that NRF-1 binds the E2F6 promoter in vivo and provide further evidence for NRF-1 transcriptional control of E2F6 promoter activity.
NRF-1 association with the E2F6 promoter in vivo
We previously defined the murine and human E2F6 promoter region [14,15]. The present study has extended these previous results. Alignment of the mouse and human E2F6 core promoter sequences reveals a 66% identity within the 200 nt surrounding the start sites and 85% identity between the murine and rat sequences. DNase I footprint analysis was performed to localize the major regulatory elements present in the E2F6 proximal promoter region. Four major protected areas were identified (FPI–FPIV). Computer-based analysis of these sequences identified two potential binding sites for NRF-1 (FPI and FPIII), which differ from the optimal consensus NRF-1 binding site (T/C)GCGCA(C/T)GCGC(A/G)  by 1 bp. Although the perfect consensus is probably the best NRF-1 binding site, a large number of binding sequences in genes known to be regulated by NRF-1 revealed a more general consensus. The presence of other protected regions (FPII and FPIV) indicates additional protein–DNA interactions with the human E2F6 core promoter region. No clear similarity was found between these sequences and any genuine consensus binding sites, and further studies are required to determine which proteins could be implicated. Nevertheless, it should also be noted that these sequences are poorly conserved in the other mammalian genes, indicating probably a more marginal effect on E2F6 regulation. Conversely, the two NRF-1 sites are well conserved among human, murine and rat E2F6 core promoters, supporting the hypothesis that they play an essential role in E2F6 gene regulation. As expected, EMSA assays performed with mouse E2F6 probes, encompassing the potential NRF-1 binding sites (−37 to +1 and +2 to +39 relative to the transcription start site), showed NRF-1 binding to these sites (results not shown). With regard to the human gene, these two NRF-1 elements seem to be essential for optimal E2F6 gene expression, since their mutation dramatically decreased the promoter activity in U2OS cells. Nevertheless, a more pronounced mutation effect was observed for the p(−31/+103) construct than for p(−1614/+290), suggesting that other regulatory elements are localized between nucleotides −1614 and −31. Interestingly, overexpression of NRF-1 activated both E2F6 promoter constructs. Although the overall NRF-1 effect on E2F6 transcription is relatively low, it is in the range of that observed for the rat Tfam promoter and the human eIF-2α (eukaryotic translation-initiation factor-2α) promoter, two well-known NRF-1 target genes [16,31]. Moreover, E2F6 promoter activity is completely abolished upon transfection of DN-NRF-1, as previously described for other NRF-1 target genes [28,29]. The potential involvement of NRF-1 in the control of the E2F6 transcriptional activity is also reinforced by the fact that NRF-1 is associated in vivo with the E2F6 promoter as demonstrated by ChIP.
Therefore, these results indicate that endogenous NRF-1 is important for efficient E2F6 promoter activity. Interestingly, a large number of NRF-1 target genes, including E2F6, present common features, such as multiple transcription start sites, absence of canonical TATA or CCAAT boxes and GC-rich content [14,15,22,32]. Moreover, most of these genes encode products that are widely expressed. This is in agreement with the wide distribution of E2F6 that correlates well with NRF-1. In adult tissues, these genes are ubiquitously co-expressed in all tissues analysed and their highest expression is found in skeletal muscle [7,14,27,33]. Moreover, the NRF-1 gene was expressed during oogenesis and during early stages of embryogenesis . Similarly, Northern blot experiments performed on total RNA from various stages of mouse embryonic development revealed E2F6 expression at a high level from day 9 to day 18 post-coitum , whereas in situ hybridization showed a broad pattern of E2F6 expression at low level on E14.5 embryos .
The best defined biological role for NRF-1 is in the nuclear control of mitochondrial respiratory function . In addition, NRF-1 has also been implicated in other cellular functions. Efiok and Safer  demonstrated that overexpression of NRF-1 increased both protein synthesis and growth by upregulating the transcription of eIF-2α and eIF-2β genes, but retarded cell cycle progression by the repression of E2F1 gene transcription. Moreover, NRF-1 sites are found in genes that may be directly involved in cell cycle regulation (cdc2 and the guanine-nucleotide exchange factor RCC1) or are regulated by cell growth [DNA polymerase-α, GADD153 (growth-arrest and DNA-damage-inducible protein 153) and ornithine decarboxylase] . Nevertheless, despite a growing number of publications and potential implication of NRF-1 as a key regulator of numerous crucial cellular processes, nothing is really known concerning the regulation of its activity, except a potential enhanced NRF-1 DNA-binding capacity upon phosphorylation [36,37].
Our finding that NRF-1 directly regulates E2F6 complements the published observation that NRF-1 may participate in coordinating the regulation of global protein synthesis, growth and cell cycle [27,35]. To date, our results only suggest a role for NRF-1 in the control of basal E2F6 transcription and may partially explain its wide expression in tissues and cell lines. Analysis of E2F6 expression in NRF-1−/− mice would have been very interesting, but these animals died between embryonic days 3.5 and 6.5, and their blastocysts are unable to grow in vitro . Consistent with these observations, NRF-1 belongs to a family of developmentally expressed transcription factors. Invertebrate NRF-1 homologues, P3A2  and EWG (erect wing gene product) , as well as the zebrafish (Danio rerio) homologue nrf (not really finished) , have been implicated in embryonic or larval development. Interestingly, E2F6 knock-out experiments have revealed that E2F6 protein is critical for developmental patterning . Our results raises the possibility that the NRF-1 transcription factor could act, partially via E2F6, in human developmental patterning, similar to its invertebrate homologues.
We thank J. L. Baert for critical reading of the manuscript. This work has been carried out on the basis of grants awarded in part by the Centre National de la Recherche Scientifique, the Institut Pasteur de Lille, the Université des sciences et technologies de Lille 1, the Association pour la Recherche sur le Cancer and the Ligue Contre le Cancer (France).
DNA-binding domain of Gal4
eukaryotic translation-initiation factor
electrophoretic mobility-shift assay
fragile X mental retardation 1
intercellular cell-adhesion molecule 1
in vitro translated
nuclear respiratory factor-1/α-palindrome-binding protein
polycomb group proteins