Co-regulation of nuclear respiratory factor-1 by NFκB and CREB links LPS-induced inflammation to mitochondrial biogenesis

The nuclear respiratory factor-1 (NRF1) gene is activated by lipopolysaccharide (LPS), which might reflect TLR4-mediated mitigation of cellular inflammatory damage via initiation of mitochondrial biogenesis. To test this hypothesis, we examined NRF1 promoter regulation by NFκB, and identified interspecies-conserved κB-responsive promoter and intronic elements in the NRF1 locus. In mice, activation of Nrf1 and its downstream target, Tfam, by Escherichia coli was contingent on NFκB, and in LPS-treated hepatocytes, NFκB served as an NRF1 enhancer element in conjunction with NFκB promoter binding. Unexpectedly, optimal NRF1 promoter activity after LPS also required binding by the energy-state-dependent transcription factor CREB. EMSA and ChIP assays confirmed p65 and CREB binding to the NRF1 promoter and p65 binding to intron 1. Functionality for both transcription factors was validated by gene-knockdown studies. LPS regulation of NRF1 led to mtDNA-encoded gene expression and expansion of mtDNA copy number. In cells expressing plasmid constructs containing the NRF-1 promoter and GFP, LPS-dependent reporter activity was abolished by cis-acting κB-element mutations, and nuclear accumulation of NFκB and CREB demonstrated dependence on mitochondrial H2O2. These findings indicate that TLR4-dependent NFκB and CREB activation co-regulate the NRF1 promoter with NFκB intronic enhancement and redox-regulated nuclear translocation, leading to downstream target-gene expression, and identify NRF-1 as an early-phase component of the host antibacterial defenses.

NRF1 expression, similarly to that of NFB, is stimulated by exogenous factors and by endogenous physiological events (Bergeron et al., 2001;Piantadosi and Suliman, 2006;Scarpulla, 2008;Xia et al., 1997). NRF-1 is coordinately involved in the regulation of mitochondrial mass , and is strongly upregulated by LPS in wild-type but not in Tlr4 -/mice. In the latter case, the limited response delays the restoration of mtDNA copy number (Suliman et al., 2004;Suliman et al., 2005). Despite the hundreds of genes, including dozens of transcription factors potentially activated by NFB (Tian and Brasier, 2003), no direct interactions between classical NFB signaling and the transcriptional activation of mitochondrial biogenesis have been defined. Hence, we tested the hypothesis that NFB, acting through the LPS-receptor pathway, regulates NRF1 directly, leading to amplification of mitochondrial mRNA transcription and enrichment of mtDNA copy number. Our initial findings substantiated such a role for NFB; however, the full protection of mtDNA copy number after activation of the LPS receptor complex also required cooperative CREB-dependent regulation of NRF1.

NRF-1 and mtDNA transcription and replication via NFB activation in vivo
The effect of NFB activation on Nrf1 expression was characterized in the livers of mice injected with a single i.p. dose of heatinactivated E. coli (5ϫ10 6 c.f.u.) by measuring sequential Nrf1 mRNA levels. In wild-type mice, Nrf1 mRNA analysis by real time RT-PCR showed that hepatic mRNA levels increase significantly 6-24 hours after E. coli administration (Fig. 1A). To test whether NFB activation regulates NRF-1 production, mice were pre-treated with the irreversible IB kinase inhibitor, BAY11-7085, followed by E. coli. BAY11 significantly delayed and attenuated the increase in Nrf1 mRNA (Fig. 1A). The inhibitory effect of BAY11 on NFB was confirmed by suppression of E. coli-induced NOS2 expression (data not shown). To confirm that NFB participates in Nrf1 gene expression, we challenged p50 -/mice with heat-inactivated E. coli. The absence of p50 inhibited only the 6 hour Nrf1 gene expression (Fig. 1A), implicating p50 in initial Nrf1 induction and one or more other subunits in the complete early-phase response.
NRF-1 stimulates nuclear-encoded Tfam expression by binding to NRF-1-response elements in the promoter region (Virbasius and Scarpulla, 1994). Tfam is then imported into mitochondria and increases mtDNA transcription and replication (Scarpulla, 2002). The mRNA levels for Tfam and two mitochondrial-encoded proteins, COI and NDI, were analyzed by real time RT-PCR to determine whether Nrf1 induction by NFB activation causes Tfam transcription and mitochondrial-encoded target gene expression. In wild-type mouse liver, Tfam mRNA levels increased at 24 and 48 hours after E. coli administration, and the response was blocked by addition of BAY11 (Fig. 1B). Tfam expression was also delayed in p50 -/mice until 48 hours (Fig. 1B). mRNA levels of mtDNA-encoded COI (Mtco1) and NDI (Mtnd1) increased 24-48 hours after E. coli administration; this was inhibited in BAY11-treated mice and delayed in p50 -/mice ( Fig. 1C-D). We also examined downstream NFBmediated effects of NRF-1 induction by checking hepatic mtDNA copy number, and found an increased copy number at 48 and 72 hours after E. coli administration that was inhibited in wild-type mice by BAY11 and delayed in p50 -/mice (Fig. 1E).

Identification of B sites in the NRF-1 locus
Since NRF-1 expression was induced in an NFB-dependent manner, we explored how LPS and E. coli stimulate NRF1 gene expression. Despite mounting evidence that the immune system activates NRF1 (Piantadosi and Suliman, 2006;Suliman et al., 2003;Suliman et al., 2005), there are no reports of the gene having functional B-binding sites. We searched for NFB and CREB consensus binding sequences using web-based rVISTA to identify conserved sequences for specific transcription factors by linking them to the TRANSFAC database (Loots and Ovcharenko, 2004). Analysis of the mouse and human proximal 1.5kb of the NRF1 5ЈUTR (DNAsis and Genomatix) identified potential NFBresponse elements (BREs) within the conserved NRF1 5Ј-promoter sequence. A schematic of the NRF1 locus with expanded sequences is shown in Fig. 2A where the regions at -500 to -120 of the mouse and -920 to -150 of the human upstream of the NRF-1 transcription start site (TSS) bear sequences identified with a high likelihood for NFB binding by exhibiting 90-100% identity with the canonical NFB enhancer sequence, 5Ј-GGRRNNYYCC-3Ј (where R is a purine, Y is a pyrimidine and N is any nucleic acid). Comparative sequence analysis, effective for finding functional coding and noncoding elements in vertebrates (Loots et al., 2002), identified 11 NFB sites in the non-coding region; one in the promoter region is interspecies-conserved whereas three conserved sites are located in intron 1 at positions +902 and +972 relative to the human or mouse TSS. As the other NFB sites are not conserved between human and mouse, intron 1 might be particularly important for NFB regulation of NRF1 gene expression.
To establish NRF1 as an NFB target gene, we treated TLR4activatable human HepG2 and mouse HL-1 cardiac cells with LPS and TNF and measured NRF-1 expression by quantitative realtime RT-PCR. Cell dose-time experiments with LPS and TNF, singly and in combination, revealed maximal NRF1 mRNA induction by 10 ng/ml of each after 4-8 hours (data not shown). LPS upregulated NRF1 mRNA expression in both cell lines, with earlier induction in HepG2 than in HL-1 cells (Fig. 2C) and LPS+TNF gave the most rapid response. To examine the contribution of NFB to NRF1 gene activation by LPS and/or TNF, cells were transfected with control non-specific siRNA or siRNA targeting RelA/p65, and silencing efficacy was determined at 48 hours (Fig. 2D). After establishing an optimal protocol for each line, the cells were treated with LPS and/or TNF for 8 hours. Control siRNA did not affect basal or LPS-and TNFinduced NRF1 transcription, whereas knockdown of RelA/p65 reduced basal and LPS-stimulated NRF1 mRNA levels by about half (Fig. 2E). To confirm that NFB directly mediates LPS-and TNF-induced NRF1 expression, cells were treated for 6 hours with BAY11 before LPS or TNF, which blocked the increase in NRF1 mRNA and provided evidence that NFB mediates the response (Fig. 2E). Figs 1 and 2 thus indicate that NFB activation contributes to LPS-and LPS+TNF-induced NRF1 transcription in vivo and in HepG2 and HL-1 cells.
Given the potential NFB-binding sites identified in the NRF1 promoter and intron 1 regions, we used electrophoretic mobilityshift assay (EMSA) to investigate NFB binding specificity at these sites. Liver nuclear extracts from mice treated with E. coli were used in binding reactions with probes containing either NFB promoter (P) sequences ( Fig. 3A) or intron (I) sequences (Fig. 3B). In Fig. 3C, when promoter region probes P1-P4 were used in the analysis, the signal for the DNA-protein complex was shifted in liver extract from E. coli-treated mice. Signal intensity was highest with P2, and this labeling was essentially absent in liver nuclear extracts from untreated mice (lane C). Furthermore, when a mutant probe (Mu) for the NFB consensus sequence was used with liver extract from E. coli-treated mice, no DNA-protein complex was observed (Fig. 3C). When nuclear probes for the intronic region (I1-I4) were used, a shift in the DNA-protein signal from E. coli-treated mice was observed (Fig. 3D). The signal intensity was highest with I1, and was greatly attenuated in nuclear extracts from untreated mice (Fig. 3D, lane C). Supershift experiments with anti-RelA or anti-cRel antibody showed that the P2-binding factor was primarily p65, and to a lesser extent, cRel, but no RelB supershift was detected (Fig. 3E). This indicated a specific NFB interaction with the predicted binding site in the Nrf1 promoter.
NFB subunit nuclear translocation was also checked at different times. Liver nuclei of E. coli-treated mice showed p65 translocation by 2 hours and cRel translocation by 6 hours (Fig. 3F). E. coli induced a minor p50 translocation at 2 hours and almost no RelB translocation.

NFB binding to Nrf1 promoter in vivo
ChIP assays were used to examine whether NFB bound directly to putative B sites identified in the Nrf1 promoter and intronic regions in mouse liver after E. coli challenge. ChIP was used with anti-NFB p65 and primer sets designed to detect the promoter and intron 1 of Nrf1 (Fig. 4A,B). Since recruitment of transcription factors precedes the maximal transcription rate (Ryser et al., 2007;Ryser et al., 2004), we monitored p65 occupancy of these Nrf1 regions at 2 hours and 6 hours after E. coli and found that p65/RelA bound most strongly at 6 hours to promoter regions R1 to R3 (Fig.   4C) and to intron 1 at R7, just downstream of exon 1 (Fig. 4D). The ChIP indicated moderate p65 promoter occupancy after E. coli treatment, but more active binding at intron 1 (Fig. 4D). Since anti-cRel antibody had shown binding to P2, we assessed the occupancy of cRel at the Nrf1 promoter, and cRel binding at position -94 to -298 was found at 6 and 24 hours (Fig. 4E). Occupancy of the Nrf1 promoter by p65 was also accompanied by DNA PolII recruitment, which is indicative of initiation of transcription (Fig.  4F).

CREB binding to Nrf1 promoter and enhanced transcription
The Nrf1 promoter region revealed CRE-binding sites at three locations: -714, -829 and -968, respectively. CREB binding was evaluated in nuclear protein extracts from control and E. colistimulated livers with unique oligonucleotides for the three promoter CRE motifs (Fig. 5). The EMSA indicated the formation of a single DNA-protein complex (Fig. 5A, arrow) with changes in the intensity of the complex in response to E. coli, with the highest intensity for the -829 tcTGACACCAtg sequence (Fig.  5A). Since transcriptional activation by CREB occurs when Ser133 is phosphorylated, EMSA was used with supershift assays for CREB and phosphorylated CREB to test binding at the Nrf1 promoter CRE sites. We found an oligonucleotide-protein complex supershift in liver nuclear extracts from mice treated with E. coli 2568 Journal of Cell Science 123 (15) Fig. 3. Gel-shift analyses using oligonucleotide probes for NFB in the Nrf1 locus. The probes are listed below the maps of the (A) 5Ј-proximal region and (B) intron 1 region. A mutant (Mu) probe was also used. Nuclear extracts from wild-type mice with or without E. coli administration were prepared from fresh liver and used for gel-shift analysis with different NFB consensus probes from Nrf1. Arrows indicate NFB complexes. (C)Gel-shift experiments using NFB-motif containing probes; lanes 1-4 are nuclear extract from livers of wild-type mice 2 hours after addition of E. coli using 32 P-labeled oligonucleotides derived from the Nrf1 promoter (labeled P1-P4, respectively). A mutant oligonucleotide was used in lane 5, and lane 6 is nuclear extract from control mouse liver using oligonucleotide P2. (D)Gel-shift on mouse liver nuclear extracts at 2 hours after E. coli exposure using the intronic region oligonucleotides I1-I4. (E)EMSA of wild-type mouse liver nuclear protein 2 hours after E. coli exposure. Nuclear protein was incubated with 32 P-labeled NFB recognition site (P2) and with serum or polyclonal anti-RelB, anti-cRel or anti-p65. The supershift complex is indicated by SS. The cRel and p65 supershifts are typical of three experiments. (F)Representative nuclear western blots of NFB subunits in wild-type mouse liver after E. coli treatment. TBP was detected as a loading control.
( Fig. 5B). Unlabeled CRE oligonucleotide in molar excess outcompeted labeled oligonucleotide for binding, indicating specificity of the CRE complex.
A further ChIP assay in liver revealed significant CREB binding to the Nrf1 promoter at two different positions (-652 to -748, -812 to -982) (Fig. 5C). Interestingly, the two sites were activated at different times in E. coli-treated mice and the -829 CREB site showed more binding activity at 6 hours.
To confirm NRF1 as a transcriptional target of NFB p65 and CREB, one or both were silenced in HepG2 cells. We have reported NRF-1 silencing in these cells at 24-48 hours, and effective CREB silencing is demonstrated in Fig. 5D. The transfected cells were treated with LPS and TNF for 8 hours and NRF1 expression was measured by real-time RT-PCR. These data showed that NRF1 induction was suppressed by silencing CREB and p65 (Fig. 5E).

LPS induces a functional Nrf1 promoter
We mapped the upstream Nrf1 promoter region to expand our knowledge of potential interactions between basal transcriptional core promoter elements and specific DNA-binding and accessory transcription factors. For instance, Nrf1 ARE consensus sequences bind NF-E2-related factor 2 (Nfe2l2) to enhance its expression in the heart (Piantadosi et al., 2008), but in inflammation, the presence of complex cognate TLR signaling would imply the involvement of several transcription factors in transcriptional regulation of Nrf1.
We expected that GFP expression from the basal promoter would be altered depending on the presence of regulatory sequence(s) in this region. For these assays, HepG2 cells were used for ease of transfection and high LPS responsiveness. After transfection with the vectors, HepG2 cells were treated for 8 hours with LPS, and GFP activity was measured 24 hours later by fluorescence microscopy or by microplate fluorometer. Our results with these constructs are shown in Fig. 6B (cell images) and Fig. 6C (histograms). In preliminary studies, the basal NRF1 promoter transcription efficiency was ~1.06-fold that of control promoterfree vector (GFP0; not shown). After LPS exposure in cells transfected with GFP1, green fluorescence intensity increased 11fold compared with untreated cells (Fig. 6B,C), whereas fluorescence increased ~7-fold in cells transfected with GFP2 vector compared with untreated cells (Fig. 6B,C), suggesting that the -500 to +40 region of NRF1 augments transcription by the basal promoter. GFP1 transfection revealed an enhanced NRF1 promoter transcription efficiency compared with cells transfected with plasmid GFP2, thus implicating additional regulatory motifs between -1000 and -500 in the sequence. The -1000 to -500 region also had CREB consensus sequences at -968, -829, -716 and -714; however, deletion of -700 to -52 as in GFP3 caused a 63.5% decrease in GFP fluorescence intensity compared with GFP1, indicating that CREB cooperates with NFB in the 1040 bp fragment to drive optimal NRF1 promoter activity.
To assess the effect of NFB binding on NRF1 promoter activity, transient transfection analysis with the GFP1 construct was carried out in HepG2 cells with and without BAY11 treatment. GFP1transfected cells treated with BAY11 followed by exposure to LPS showed a ~90% lower promoter transcription efficiency compared with control LPS-treated cells (Fig. 6C). Cooperation between NFB and CREB was seen in HepG2 cells co-transfected with GFP1 and p65 siRNA or CREB siRNA. Co-transfection before LPS treatment with p65 or CREB siRNA decreased transcriptional efficiency at the NRF1 promoter by 52% and 48%, respectively. Thus, NFB p65 and CREB are both required for optimal NRF1 expression after LPS treatment. BAY11 also blocked the effects of CREB, and conversely, CREB siRNA reduced the effect of CREB and decreased the stimulatory effects of NFB. This crosstalk indicates that NFB and CREB cooperate for maximal LPSinduced NRF1 promoter activation.

Intronic NFkB enhancement of NRF1 gene expression
To determine the role of the intronic B binding sites in NRF1, we constructed a GFP vector consisting of a promoter sequence derived from the 5Ј-proximal NRF1 region (from -1000 to +40 bp or -500 to +40 bp relative to the TSS) and an enhancer sequence derived from the first intron (+902 to +980 bp of intron 1 relative to the TSS). This 78 bp fragment with the enhancer sequence, apart from the NFB site, includes no other known transcription factor elements. A mutation of the NFB site was introduced into construct GFP7. As shown in Fig. 7A-C, the GFP reporter vector with 5Јproximal region (GFP 4) and the 78 bp NFB-binding region had 58-fold higher fluorescence intensity after LPS treatment compared with control vector without the promoter or enhancer (GFP0) and 14-fold compared with untreated controls. Transfection of GFP5 vector (-500 to +40) with the intronic enhancer site resulted in ã 7.7-fold induction in GFP fluorescence intensity after LPS treatment. However, deletion of the 648 bp between -700 and -52 to form GFP6, containing only the CREB site, caused the GFP fluorescence intensity to drop by 50% after LPS (Fig. 7C). We further examined the effects of BAY11 on the enhancer activity of NFB using a vector with both the 5Ј-proximal NFB region and enhancer (GFP5). Here BAY11 alone did decrease GFP activity compared with the control (Fig. 7C). Experiments using a vector containing mutant intronic B sites (GFP7) decreased the LPS effect on GFP activity by 50% compared with results with GFP5 (Fig. 7C), confirming the involvement of the intronic kB site in NRF1 gene expression. These results indicate that intronic NFB functions as a cis-acting enhancer element on the NRF1 locus. such as TNF, stimulate mitochondrial ROS production, we asked whether mitochondrial ROS influence NFB-or CREBmediated NRF1 gene expression directly or alternatively via cytoplasmic Ca 2+ signaling. Superoxide and H 2 O 2 egress via Complex III, for instance, might serve as a messenger for NFB activation directly, or perhaps in conjunction with Ca 2+ (Sen et al., 1996). Therefore we used a mitochondrial-targeted catalase vector to investigate the role of mitochondrial H 2 O 2 production in activation of NFB and CREB. Normal mouse HL-1 atrial cardiomyocytes were transfected with mtCAT or empty vector for 48 hours, exposed to LPS+TNF for 4 hours, and nuclear extracts were analyzed for NFB and CREB. The presence of mCAT inhibited the LPS+TNF-dependent accumulation of nuclear NFB subunits by ~50-80% and phosphorylated CREB by approximately two thirds (Fig. 8A,B), implicating mitochondrial H 2 O 2 release as a control point.
Since mitochondrial H 2 O 2 might be involved in stress-induced Ca 2+ signaling, the effects of Ca 2+ mobilization on nuclear translocation of NFB and phosphorylated CREB were evaluated using EGTA-AM, a potent cellular Ca 2+ chelator, and thapsigargin (TG), an inhibitor of Ca 2+ -ATPase in the ER. Since most Ca 2+ chelators are short acting, HL-1 cells were pre-treated with EGTA-AM (20 M) for 30 minutes followed by LPS and TNF for 2 hours and nuclear extracts probed for NFB subunits by western blot. LPS-and TNF-induced nuclear translocation of p65, p50 and cRel was decreased, but not prevented by EGTA-AM (Fig. 8A), indicating a small Ca 2+ effect. EGTA-AM did however block the nuclear accumulation of phosphorylated CREB (Fig. 8B).
Thapsigargin produces rapid Ca 2+ efflux from the ER lumen (Thastrup et al., 1990), and in contrast to slow-acting ER stresseliciting agents (Pahl and Baeuerle, 1995), rapidly activates NFB (Pahl et al., 1996). If ER Ca 2+ release is involved in NFB activation after treatment with LPS and TNF, it would be important to know whether it interacts with mitochondrial H 2 O 2 generation. HL-1 cells stimulated with TG (30 nM for 30 minutes) did activate NFB, but mCAT transfection nearly completely blocked nuclear accumulation of p65 and cRel (Fig. 8A), implying that ER Ca 2+ release stimulates mitochondrial H 2 O 2 , which ultimately mediates 2571 NFB and CREB-induced NRF1 activation Fig. 6. Functional cis-acting elements in the NRF1 promoter. NFB and CREB function at the NRF1 promoter was established in HepG2 cells transfected with plasmid pGlow (GFP0) containing no promoter region (negative control) or plasmids harboring regions -500 to +40 (GFP2) or a chimeric promoter for -1000 to -700 plus -52 to +40 bp (GFP3) of the NFR1 gene. To assess additive effects of the two elements spanning -1000 to +40 of NRF-1, transfection was also carried out using promoter -1000 to +40 (GFP1). (A)Schematic representations of various plasmids used in the transfection assays are labeled on the left (drawings are not to scale). Plasmids pCMV-GFP and GFP0 plasmid were used as positive and negative controls, respectively. (B)HepG2 cells transfected with the GFP plasmids. The activity of each construct was measured before and 24 hours after LPS. Green indicates GFP expression driven by the NRF1 promoter construct by fluorescence microscopy. The blue fluorescence is nuclear staining with DAPI. (C)Relative activity of each construct measured before and after 24 hours of LPS treatment. The activity of construct GFP1 was measured as fluorescence intensity after treatment of cells with the NFB inhibitor BAY11 or after co-transfection with p65 or CREB siRNA. Comparative fluorescence intensity represents means ± s.e. of three independent studies performed with plasmids in triplicate (*P<0.05).
NFB activation. TG also stimulated nuclear accumulation of phosphorylated CREB, and this effect was partially abrogated by mCAT (Fig. 8B).
Confocal microscopy demonstrated the LPS-and TNF-mediated nuclear translocation of p65 and phosphorylated CREB in HL-1 cells, which was inhibited by mCAT (Fig. 8C). Ca 2+ chelation by EGTA-AM did not block LPS+TNF-induced nuclear p65, but did block nuclear accumulation of phosphorylated CREB; the latter together with the mCAT data implies an interaction between mitochondrial H 2 O 2 and ER Ca 2+ .

Discussion
A direct link between the innate inflammatory response mediated by NFB signaling and the transcriptional activation of mitochondrial biogenesis has not been reported previously. This work has identified active intragenic NFB-responsive cis-elements both in the promoter and in the first intron of NRF1, a pivotal transcription factor for mitochondrial biogenesis, which enables mitochondrial gene expression during the inflammatory response. These B elements act as transcriptional enhancers that allow the LPS-TLR4 cognate and the early-phase cytokine TNF, to activate NRF1. Also new is the aspect of synchronization between the classical NFB pathway and the bZIP transcription factor CREB, which is integral to the regulation of pyruvate, glycogen and fatty acid metabolism (Zhang et al., 2005). This unusual mechanism of NRF1 regulation drives NRF-1 target-gene expression during the TLR4-mediated immune response.
Because TLR4 activates NFB (da Silva Correia and Ulevitch, 2002; Li and Verma, 2002) and can be accompanied by CREB activation (Illario et al., 2008;Martin et al., 2005), it was necessary to address direct transcription factor binding to the NRF1 locus as an explanation for LPS-mediated NRF-1 expression (Suliman et al., 2003). A bioinformatics analysis revealed B sites spanning the NRF1 locus in both the 5Ј-proximal (promoter) and intronic regions. Three B sites within intron 1 are conserved across species, and we considered these potentially crucial for NRF1 expression because interspecies-conserved regions often correspond to DNaseI HS sites (Kang and Im, 2005), which could be targets for binding of transcription factors. Therefore, we focused our analysis initially on NFB.
By EMSA, NFB was demonstrated to function as an enhancer element within the NRF1 locus through significant p65 and weaker cRel binding at the NRF1 promoter. ChIP analysis confirmed that the NRF1 promoter site of interest is occupied primarily by p65. GFP-reporter assays using NRF1 promoter deletion and NFBmutation constructs in HepG2 cells demonstrated that NFB contributes to LPS-induced reporter activity in the mouse Nrf1 promoter. The enhancer operates efficiently with a promoter derived from the 5Ј-proximal region, but not with a minimal promoter. Because the 5Ј-proximal region includes B-responsive elements (Saraiva et al., 2005), it cooperates with these elements to exert its enhancing effects. The discovery of a working NFB enhancer element at intron 1 in the NRF1 gene locus is quite intriguing, because binding of NFB to intronic elements is rare (Charital et al., 2009).
We confirmed the NFB intronic functionality by GFP-reporter assays by co-transfecting HepG2 cells with NRF-1 reporter constructs and siRNA to knock down CREB or p65 after stimulation 2572 Journal of Cell Science 123 (15) Plasmids pCMV-GFP and GFP0 plasmid were used as positive and negative controls, respectively. The reporter constructs are GFP0 (negative control), pCMV-GFP (positive control); GFP4, containing the 1040 bp region at the 5Ј site (-1000 to +40 bp of exon 1) plus the 78 bp region of intron 1; GFP5, containing the 540 bp region at the 5Ј site (-500 to +40) plus the 78 bp region of intron 1; GFP6, containing the 78 bp of intron 1 region of the NFB sites plus the chimeric promoter region (-352 to +40 bp of exon 1), and GFP7, containing the mutation at intronic NFB sites within 78 bp plus the region from plasmid GFP5. (B)HepG2 cells transfected with the respective plasmids. The activity of each construct was measured before and after 24 hours of LPS treatment. Green fluorescence represents GFP expression driven by NRF1 promoter and intronic constructs detected by fluorescence microscopy. The blue fluorescence is nuclear staining with DAPI. (C)Ratio of observed fluorescence intensity to the basal level for GFP0 or to untreated controls. The relative activity of each construct was measured before and 24 hours after LPS treatment. GFP4 activity was measured after treatment of cells with BAY11. Relative fluorescence intensity represents mean ± s.e. of three studies performed with plasmids in triplicate (*P<0.01 vs GFP1 and P<0.05 vs BAY11).
with LPS. These assays indicated that full NRF1 induction requires not only NFB, but also CREB, and that these two transcription factors interact because NFB silencing blocks the CREB effect and vice versa. It is thus possible that NFB not only activates CREB, but that CREB participates in LPS-mediated NFB activation for NRF1 expression.
Stress-related mtDNA damage leads to sustained mitochondrial ROS production (Suliman et al., 2005) that alters the expression of an array of genes associated with apoptotic phenotype and cell survival (Suliman et al., 2007b). Exposure to LPS alone, or LPS and TNF, in HL-1 cells activates NFB and CREB, while stimulating the production of H 2 O 2 , a known regulator of NFB nuclear translocation (Beg et al., 1993). In these cells, NFB activation required mitochondrial H 2 O 2 whereas Ca 2+ had a minimal role. Mitochondrial-targeted catalase (mCAT) but not the Ca 2+ chelator EGTA-AM blocked NFB activation by the LPS signal, whereas after Ca 2+ mobilization by TG, EGTA-AM, and mCAT blocked nuclear accumulation of p65 and cRel. Ca 2+ chelation fully abrogated TG-mediated accumulation of phosphorylated CREB, but mCAT also partially blocked accumulation of phosphorylated CREB, placing importance on mitochondrial H 2 O 2 in both NFB and Ca 2+ -mediated CREB activation. Thus, the evidence in heart cells implicates only mitochondrial H 2 O 2 production in the LPS-and TNF-induced nuclear translocation of NFB and mitochondrial H 2 O 2 -associated Ca 2+ mobilization in the nuclear translocation of pCREB. Both conditions are consistent with the redox regulation of NRF1 (Piantadosi and Suliman, 2006); however, the sites of intramitochondrial ROS production were not determined in this case. Moreover, the conditional participation of other redox mechanisms involving NFB, for instance via NO and CREB (Dhakshinamoorthy et al., 2007), has not been excluded. Under comparable NFB-activating conditions, IK activation and accelerated loss of IB inhibitory protein is frequently reported (Kamata et al., 2002).
Loss-of-function experiments indicated that LPS and TNF induce NRF1 transcription through NFB-promoter and intronicenhancement elements, thus providing an explanation for the prompt increases in NRF-1, an established integrator of nuclearmitochondrial communication (Scarpulla, 2002), during the early phase of the antibacterial host response. Similarly, LPS and TNF increase NRF-1-regulated Tfam gene expression and subsequent increases in mtDNA-encoded COI and NDI mRNA, reflecting mitochondrial transcriptional activity befitting reports that LPS exposure doubles COI mRNA content in mouse cells . Following NRF-1 production, Tfam levels and mtDNA copy number also increase, which is necessary to support the capacity for oxidative phosphorylation (Scarpulla, 2008) and which might reduce oxidative stress (Stirone et al., 2005).
NRF1 induction through NFB and CREB binding to intragenic DNA with enhancement of NRF-1-regulated Tfam transcription addresses the Tfam-regulated mitochondrial gene transcription and early mtDNA replication found in LPS-injured animals (Suliman et al., 2003), as well as the late restitution of mtDNA copy number after LPS exposure in Tlr4-null mice (Suliman et al., 2005). Tfam is required for efficient mitochondrial promoter recognition by DNA polg, and is necessary for maintenance of oxidative phosphorylation as well as for mtDNA replication during mitochondrial proliferation (Scarpulla, 2008). The NFB-CREB signal integration for NRF1 function also reflects a rapid optimization of mitochondrial gene expression coincident with metabolic and inflammatory stressor influences on host defense and repair through classical NFB activation, apparently without dependence on non-canonical mechanisms (Bakkar et al., 2008).
In conclusion, NRF1 promoter co-regulation by NFB and pCREB combinatorial interactions enhanced by p65 binding to NRF1 intron 1 and with essential regulation by mitochondrial H 2 O 2 represents the first known mechanism by which LPS-receptormediated signaling engages directly in the transcriptional control of mitochondrial biogenesis. The implications for rapid mitochondrial turnover and quality control are vital to cell survival during inflammatory pathogenesis, and interference with this mechanism would place cells at consequential risk for increased apoptosis and/or necrosis, particularly during exaggerated or prolonged activation of host antibacterial defenses. In some experiments, cells were treated with Ca 2+ chelator (20M EGTA-AM) or with 30 nM thapsigargin (TG), or both for 30 minutes followed by LPS and TNF for 2 hours. Nuclear extracts were immunoblotted with anti-p65, anti-p50, anti-cRel or anti-TBP. (B)Nuclear western blots of pCREB and total CREB protein in cells with or without mCAT transfection 48 hours before LPS and TNF treatment. In some experiments, cells were treated with 20M EGTA-AM or with 30 nM TG or both for 30 minutes followed by LPS and TNF for 2 hours. Nuclear extracts were immunoblotted with antibodies against phosphorylated CREB, CREB or TBP. (C)Confocal images of control HL-1 cells with or without mCAT transfection 48 hours before treatment with LPS and TNF. In some experiments, cells were treated with 20M EGTA-AM for 30 minutes followed by LPS and TNF for 2 hours. Cells were stained with anti-p65 (green) or with anti-pCREB (red) antibodies and nuclei were visualized with DAPI (blue).

Materials
Antibodies against p65, cRel, RelB (Cell Signaling), CREB, pCREB and TBP were obtained from Santa Cruz Biotechnology. All secondary and fluorescent antibodies were from Invitrogen. The mCAT vector was developed and characterized in our laboratory (Suliman et al., 2007a;Suliman et al., 2007b). Small interfering (si) RNA oligonucleotides were from Ambion. Mouse recombinant tumor necrosis factor- (TNF), EGTA-AM and E. coli LPS were from Sigma. Thapsigargin and BAY11-7085 were obtained from BioMol (Plymouth Meeting, PA).

Mice
The studies were pre-approved by Duke Institutional Animal Care and Use Committee; male C57BL/6 and p50 -/mice were obtained from Jackson Labs and used at 12-16 weeks of age. E. coli (serotype 086a:K61, ATCC, Rockville, MD) was non-disruptively heat-inactivated (Suliman et al., 2005). Heat-killed bacteria were diluted with sterile 0.9% NaCl to a concentration of 5ϫ10 6 /ml and single 0.5 ml doses injected i.p. into mice. At the appropriate times, the livers were harvested and snap-frozen. For NFB inhibition, mice were injected with BAY11-7085 (20 mg/kg i.p. in 1% DMSO diluted with 0.9% NaCl). BAY11 was administered twice to each animal, once 6 hours before E. coli injection and then 6 hours after E. coli.

Cell studies
HepG2 (human hepatocellular carcinoma) cells purchased from ATCC were maintained in RMPI 1640 medium (Hyclone) containing 10% fetal bovine serum, 2 mM glutamine, 100 U/ml penicillin and 100 g/ml streptomycin. Murine atrial HL-1 cells, a generous gift from William C. Claycomb (LSU Medical Center, New Orleans, LA), were cultured in Claycomb medium with 10% FBS, 100 M norepinephrine and 4 mM L-glutamine in gelatin-or fibronectin-coated flasks or plates. Cells were cultured at 37°C with 5% CO 2 and 95% air. Cells were transfected with scrambled (negative control) or targeted siRNA using FuGene HD transfection reagent (Roche) and transfection efficiencies of 65-80% and gene suppression achieved. For pharmacological inhibition of NFB in cells challenged with LPS+TNF, BAY11-7082 (5 g/ml) was used.

Bioinformatics profiling
Mouse and human NRF1 loci were aligned and the extent of DNA sequence homology computed with the web-based Regulatory Visualization Tools for Alignment (rVISTA; www.gsd.lbl.gov/vista) (Loots and Ovcharenko, 2004;Loots et al., 2002). Promoter analysis was performed with consensus sequences for transcription factor binding located with DNASIS (Hitachi Software; Alameda, CA) and confirmed with MatInspector (Genomatix Software; München, Germany). Putative NFB-or CREB-binding sites of 92-100% homology were identified in the NRF1 promoter and intronic region (Ensembl Gene ID ENSMUSG00000058440).

EMSA and ChIP
Nuclear and cytoplasmic extracts were prepared as described (Suliman et al., 2003;Suliman et al., 2005) and nuclear extracts were incubated with [ 32 P]dCTP-labeled, double-stranded probes containing an NFB consensus site from the NRF1 promoter or the putative CRE-binding sites. For competition experiments, unlabeled oligonucleotide was added 10 minutes before addition of labeled probe. For positive controls 32 P-labeled probe derived from consensus for CREB or NFB oligonucleotides (Promega) were used. For supershift experiments, 1 g of Cell Signaling antibody to p65 (3034), cRel (4727), Rel B (4954) or Santa Cruz Biotechnology anti-CREB (sc-186X), pCREB (sc-7978X) or rabbit IgG was added to the binding reactions. Reaction products were resolved on non-denaturing 5% polyacrylamide gels, which were dried and exposed to X-ray film.
ChIP assays were performed with ChIP-IT kits (Active Motif, Carlsbad, CA) using the manufacturer's protocol (Suliman et al., 2007a). Approximately 200 mg mouse liver was diced on ice, suspended in PBS, and fixed in 1% (v/v) formaldehyde at room temperature to crosslink proteins to DNA. Reactions were stopped with 0.125 M glycine and the samples centrifuged first at 3000 r.p.m. for 10 minutes and then at 7500 r.p.m. for 5 minutes, and the final pellet was suspended in ChIP buffer. Genomic DNA was sheared by sonication on ice and lysates were tumbled overnight at 4°C with salmon sperm DNA and protein-A-agarose with anti-RNA polymerase II (sc-9001X, Santa Cruz), anti-p65 (Cell Signaling) or normal rabbit IgG. Complexes were precipitated, washed serially, and eluted with fresh elution buffer (1% SDS and 100 mM NaHCO 3 ). Na + concentration was adjusted to 200 mM with NaCl followed by incubation at 37°C to reverse protein-DNA crosslinks. DNA was purified with a PCR purification kit (Qiagen) and then amplified across the targeted mouse Nrf1 promoter regions containing B-binding sites. Liver extracts from Wt or BAY11 treated mice were evaluated for NFB enrichment by quantitative real-time RT-PCR (qPCR) in 20 l mixtures of SYBR Green master mix and 0.1 M primers. NFB enrichment was evaluated using primers for upstream regions of the mouse Nrf1 gene (sequences provided on request). PCR primer efficiency was determined using input DNA and adjusted accordingly. DNA in each ChIP sample was normalized to the corresponding input chromatin (C t ) and enrichment defined as change in C t in treated versus untreated control samples (C t ), relative to baseline control (IgG).

Reporter constructs
Starting with mouse liver genomic DNA, the 5Ј-proximal region 1.5 kbp upstream of the TSS on the NRF1 locus (-1500 to +40) was amplified under standard PCR conditions and cloned into the reporter construct pGlow-TOPO (Invitrogen). The 3Ј primer at +40 was 5Ј-GCGAGCGCTGCCGCCTCTGCC-3Ј and 5Ј primer, 5Ј-ACTTGGAAGGCAAGGACAGG-3Ј. To identify the regulatory sequences within -1500 to +40 region of the NRF-1, several fragments harboring 5Ј-end deletions of this region were generated by PCR using plasmid -1500/+40GFP as a template and region-specific primers and Pfu polymerase (Stratagene). Three PCR-amplified fragments spanning -500 to +40, -1000 to +40, and a chimera of -1000 to -700 plus -52 to +40 bp region of the NRF1 promoter upstream of the TSS were cloned in plasmid pGlow-TOPO. The constructs generated were designated p1040-Nrf1glow (GFP1), p540-Nrf1glow (GFP2) and p393-Nrf1glow (GFP3). Transformants were screened for the presence of the insert by restriction enzyme digestion and confirmed by DNA sequencing.
To test for the intronic enhancer, a fragment of intron 1 (78 bp fragment including NFB sites) was amplified with Pfu polymerase using specific primers. The primers for intronic region amplification were the sense primer, 5Ј-CCTGGGTTCTT -CCTGGGACT-3Ј and the antisense primer, 5Ј-GTTCCTTCTTAAGGGGTGCGCC-3Ј. To generate p1040-Nrf1glow +78-intron1 (GFP4), p540-Nrf1glow+78-intron 1 (GFP5) and (p393-Nrf1glow+78-intron1 (GFP6), we cloned the intron 1 fragment (+902/+980) of the Nrf1 gene from mouse genomic DNA and then fused it downstream of the promoter driving exon 1 by the fusion PCR method and using GFP1, GFP2 and GFP3 as a template. Mutations in the intronic sequences were created by PCR-site-directed mutagenesis using plasmid GFP5 as a template. Amplified fragments were ligated directly to the 5Ј-proximal regions upstream of the TSS prepared in the pGlow-TOPO vector. The transformants were screened for the presence of the insert by restriction enzyme digestion and confirmed by DNA sequencing.

Transient transfection and GFP reporter assays
HepG2 cells (5ϫ10 4 /well) were plated in 24-well plates and incubated overnight. NRF1 promoter GFP constructs (1.0 g) or CMV-GFP plasmid (0.1 g, as positive control) were transfected into HepG2 cells using FuGene (Roche) according to the manufacturer's instructions. The Glow-TOPO empty vector was used as a negative control (GFP0). 48 hours after transfection, cells were examined by fluorescence microscopy on a Nikon fluorescence microscope and filter with an excitation range of 480-510 nm. For fluorometric assays, cells were harvested in PBS at different times after transfection. Whole-cell pellets were resuspended in 50 l PBS and transferred to a 96-well plate containing black wells (Labsystems). Fluorescence intensity was measured in a spectrofluorometer (Genius, Tecan) using excitation at 380 nm and emission at 510 nm corrected for background activity for the pGlow-TOPO vector alone and for cell number between experiments. All transfection data are means ± s.e. of three experiments performed in triplicate. For RNA interference, reporters and plasmids and/or siRNA (100 nM) were co-transfected into HepG2 cells with FuGene. 48 hours later, additional complete medium containing the appropriate pharmacological agent(s) was added. 24 hours after treatment, cells were washed once with PBS and the intensity of the fluorescence was measured.

Protein immunoblotting
Western blots were performed on liver and cell lysates and nuclear extracts after separation of protein by SDS-PAGE. After transfer, primary and secondary antibodies were applied and the signals developed with ECL. Band images were quantified digitally from the mid-dynamic range, and the data expressed relative to a stable reference (Piantadosi et al., 2008). For immunochemistry, cells were grown in onewell chamber slides to ~70% confluence to avoid hypoxia. After treatments, cells were washed in PBS, fixed in 2% paraformaldehyde, and washed in 1% Triton X-100 for 15 minutes at room temperature. Cells were labeled with primary antibodies against p65 (1:200) and pCREB (1:100) (Suliman et al., 2007b).

MtDNA copy number and respiratory proteins
MtDNA was determined by SYBR green quantitative PCR (qPCR). Fluorescence intensities were recorded and analyzed on an ABI Prism 7000 sequence-detector system (Applied Biosystems). MtDNA-encoded cytochrome c oxidase subunit I (Mtco1) and NADH dehydrogenase subunit I (Mtnd1) mRNA were quantified by RT-PCR and normalized to nuclear-encoded 18S rRNA (Suliman et al., 2007a;Suliman et al., 2007b).

Real-time PCR
qPCR was performed on an ABI PRISM 7000 Sequence Detection System with TaqMan gene expression and premix assays (Applied Biosystems). 18S rRNA served as an endogenous control. Quantification of gene expression was determined using the comparative threshold cycle C T and RQ method.

Statistics
Grouped data are expressed as the means ± s.e. for n4-6 replicates. Statistical significance was tested with the unpaired Student's t-test or two-way analysis of variance using commercial software. Differences at P<0.05 were considered significant.