Lipopolysaccharide Stimulates Mitochondrial Biogenesis via Activation of Nuclear Respiratory Factor-1*

Exposure to bacterial lipopolysaccharide (LPS) in vivo damages mitochondrial DNA (mtDNA) and interferes with mitochondrial transcription and oxidative phosphorylation (OXPHOS). Because this damage accompanies oxidative stress and is reversible, we postulated that LPS stimulates mtDNA replication and mitochondrial biogenesis via expression of factors responsive to reactive oxygen species, i.e. nuclear respiratory factor-1 (NRF-1) and mitochondrial transcription factor-A. In testing this hypothesis in rat liver, we found that LPS induces NRF-1 protein expression and activity accompanied by mRNA expression for mitochondrial transcription factor-A, mtDNA polymerase γ, NRF-2, and single-stranded DNA-binding protein. These events restored the loss in mtDNA copy number and OXPHOS gene expression caused by LPS and increased hepatocyte mitotic index, nuclear cyclin D1 translocation, and phosphorylation of pro-survival kinase, Akt. Thus, NRF-1 was implicated in oxidant-mediated mitochondrial biogenesis to provide OXPHOS for proliferation. This implication was tested in novel mtDNA-deficient cells generated from rat hepatoma cells that overexpress NRF-1. Depletion of mtDNA (ρo clones) diminished oxidant production and caused loss of NRF-1 expression and growth delay. NRF-1 expression and growth were restored by exogenous oxidant exposure indicating that oxidative stress stimulates biogenesis in part via NRF-1 activation and corresponding to recovery events after LPS-induced liver damage.

Mitochondria play crucial roles both in cell survival and death, and mitochondrial damage may initiate apoptosis, necrosis, or both. Mitochondria rely on an intrinsic genome (mtDNA) that is replicated and transcribed semi-autonomously but whose maintenance requires nuclear factors. The 16-kbp circular mtDNA contains limited genetic information relative to nuclear DNA; it encodes two rRNAs, 22 tRNAs, and 13 polypeptides of the electron transport chain (ETC). Mitochondrial transcription is regulated by mitochondrial transcription factor A (mtTFA) that contains domains for high mobility group-box proteins critical for DNA promoter binding and transcriptional activation. Nuclear high mobility group-box proteins enhance transcription and structural organization of chromatin and exhibit dual DNA binding ability (14,15). mtTFA initiates transcription by binding to mtDNA upstream of control elements of both heavy and light strand promoters (16). The mtTFA knockout mouse shows mtDNA depletion and lack of OXPHOS, and dies early in embryogenesis, establishing the role of mtTFA in coordinating mitochondria-nuclear genomic activity and regulation of mtDNA copy number (17).
Regulation of mtTFA transcription is not well understood, but loss of mitochondrial transcription inhibits cell growth because of lack of energy for cell replication. Two redox-responsive transcription factors, nuclear respiratory factors-1 and -2 (NRF-1 and NRF-2), have been implicated in its expression (18,19), as well as that of several ETC proteins (18,20). NRF-1 is phosphorylated, translocates to the nucleus, and binds to DNA as a homodimer. Because NRF-1 is encoded in the nucleus but controls synthesis of ETC components, it is a candidate for linking nuclear and mitochondrial genome activity during repair of damage to both mitochondria and cells. A role for NRF-1 in the duality of mtTFA function in mtDNA replication and transcription has not been explored in the context of LPS damage, which led to the present investigation. The implication is that elucidating a role for NRF-1 in communication between nucleus and mitochondrion after LPS-induced mitochondrial damage could provide new insight into the molecular mechanisms underlying a major clinical problem, organ failure of severe sepsis.

Animal Procedures
Animal protocols were approved by the Duke University Animal Care and Use Committee. Male Sprague-Dawley rats weighing 300 -400 g were injected with a single dose of intraperitoneal LPS (Escherichia coli, 055 B5; Difco, Inc.) dissolved in 1 ml of sterile pyrogen-free 0.9% sodium chloride at a sub-lethal dose (1 mg/kg) chosen for significant but reversible hepatic injury (4). Control animals were injected with an equal volume of 0.9% sodium chloride. Animals were killed 6, 24, and 48 h after injection, the abdomen was opened, and the livers were excised quickly and placed in cold isolation buffer. Fresh liver (4 g) was used to isolate mitochondria, and the remainder was snap-frozen and stored at Ϫ80°C. For microscopy livers were flushed with phosphate-buffered saline, pH 7.0, and perfusion-fixed with 4% paraformaldehyde. The livers were removed, immersed in fixative for 2 h, and stored in 2% paraformaldehyde at 4°C until paraffin embedding, sectioning, and staining for light microscopy.

Mitochondrial Isolation and Cell Fractionation
Highly purified, tightly coupled liver mitochondria were prepared as reported by discontinuous gradient centrifugation (1)(2)(3)(4). Nuclei were obtained by centrifugation of liver homogenate at 1000 ϫ g for 20 min and further purified by centrifugation through 1.75 M sucrose. Protein content was determined by BCA method (Pierce, Rockford, IL) using bovine serum albumin as a standard.

␥GCS mRNA Expression Using Northern Blot
RNA was extracted with TRIzol reagent (Invitrogen), separated on 1.2% agarose gels, stained, and photographed. RNA was transferred to nylon membranes (Nytran Super Charge; Schleicher & Schuell) and UV cross-linked. PCR products were radiolabeled with [␣-32 P]dCTP using random primers and Ready-to-Go DNA labeling beads (Amersham Biosciences). Northern hybridization was done using QuickHyb (Stratagene). A ␥GCS (catalytic subunit) cDNA probe of 390 bp was used corresponding to nucleotides 79 to 468 of the rat sequence (21). After hybridization, membranes were washed twice in 0.2% SSC (0.15 M NaCl/0.015 M sodium citrate) containing 0.1% SDS at room temperature for 15 min then in 0.1% SSC/0.1% SDS at 60°C for 45 min. The membranes were exposed to x-ray film (Eastman Kodak Co.) to develop autoradiograms.

Genomic and Mitochondrial DNA Extraction
Genomic DNA was extracted with a DNA extraction kit (DNA Isolator PS Kit; Wako Pure Chemical Industries, Ltd., Tokyo, Japan), and DNA concentration was measured by optical density. mtDNA was prepared from rat liver mitochondria by NaI isolation (mtDNA Extractor WB kit; Wako Pure Chemical Industries, Ltd.) and stored at Ϫ20°C.

Mitochondrial DNA Copy Number
Synthesis of mtDNA Mutant Fragments-A DNA fragment of 590 bp of cytochrome b was amplified by PCR using 14314 FW primer, 5Ј-CATCAGTCACCCACATCTGC-3Ј, and 14904RV, 5Ј-GGTTAGCGGGT-GTATAATTG-3Ј, and cloned into pGEM-T vector (Promega, Madison, WI). The cloned DNA fragment was truncated by 175 bp using BsmFI to generate an internal standard competitor of 415 bp. The competitor DNA was amplified in E. coli, extracted with phenol/chloroform followed by alcohol precipitation, separated into aliquots, and stored at Ϫ20°C until use.
Copy Number Determination by Competitive PCR-mtDNA copy number was quantified by competitive PCR based on co-amplification of target template with a known amount of competitor to account for variables affecting PCR amplification and assure a constant target/ competitor ratio during amplification (22). The rat cytochrome b-specific primers amplify 590 bp of the target mtDNA and 415 bp of the competitor DNA. Triplicate samples containing 2 ϫ 10 3 copies (unless otherwise indicated) of competitor and 50 ng of target mtDNA were co-amplified in a thermal cycler (PE Applied Biosystems, Foster City, CA) at different cycle numbers optimized for visible product on ethidium bromide (EB)-stained gels in the PCR exponential phase. PCR products were separated by electrophoresis in 2% agarose. The intensities of the EB-stained bands were measured by densitometry, and the ratio of mtDNA-specific product was plotted against competitor density on a log scale. Copy number was determined from the x intercept or the point at which the ratio of the 590-and 415-bp products was 1 (log 0). A correction factor of 1.4 was used to account for differences in molecular weights.

Nuclear and Mitochondrial mRNA Expression
Cytoplasmic RNA was extracted with TRIzol (Invitrogen), and 1 g from each sample was reverse-transcribed (in a total volume of 20 l) using Moloney murine leukemia virus reverse transcriptase (180 units; Promega) in a reaction buffer containing random hexamer primers, dNTPs, and ribonuclease inhibitor RNasin (Promega). Gene transcripts were amplified in triplicate using gene specific primers (Table I). 18 S rRNA was used to control for variation in efficiency of RNA extraction, reverse transcription, and PCR for nuclear and mitochondrial RNA gene expression. Quantification of amplified mRNA was done by densitometry normalized to the 18 S rRNA mRNA signal density for each sample using image analysis software (Bio-Rad).

Cell Lines and mtDNA Depletion
Rat hepatoma cells obtained from American Type Culture Collection (H4IIE; Manassas, VA) were characterized in Dulbecco's modified Eagle's medium containing 10% fetal calf serum at 37°C with 5% CO 2 . To deplete cells of mtDNA ( o cells) normal H4IIE cells were cultured for 60 or more passages in the presence of 20 ng/ml EB in Dulbecco's modified Eagle's medium supplemented with 4.5 mg/ml glucose, 50 mg/ml uridine, and 100 mg/ml pyruvate to compensate for loss of respiration. Experiments were performed on cells 24 h post-confluence. One h before adding test agents cells were switched to phenol red-free Dulbecco's modified Eagle's medium with 0.5% fetal bovine serum (Invitrogen). Oxidant exposure of H4IIE (wild type) and mtDNA-deficient (⌬H4IIE) cells was produced with tertiary butyl hydroperoxide (t-BH). The t-BH (10 -100 M) diluted in serum-free medium was added to confluent monolayers in 6-well plates. Viability was assessed after trypsin release, staining with 0.4% trypan blue dye, and counting non-viable and total cells on a hemocytometer. PI3K inhibitor LY294002 was used at 50 M. All measurements were performed in duplicate, and the experiments were repeated three times.

Southern Analysis of mtDNA
Total DNA from H4IIE and respiration-deficient H4IIE clones was isolated, and 10-g samples were digested to completion and separated by agarose electrophoresis. DNA was transferred to a nylon membrane as described (24) and hybridized to the cytochrome b cloned probe labeled using Ready-to-Go DNA labeling beads (Amersham Biosciences) and [␣-32 P]dCTP (Amersham Biosciences).

Mitochondrial Membrane Potential and ROS Production
Mitochondrial membrane potential was compared in wild type and ⌬H4IIE cells using a cell permeant cationic fluorescent indicator, rhodamine-123 (5 M; Molecular Probes), which is sequestered by active mitochondria in proportion to membrane potential (23). ROS production was estimated in cells immediately after 3 h of exposure to t-BH using dihydrorhodamine-123 (Molecular Probes, Eugene, OR). Dihydrorhodamine-123 accumulates primarily in mitochondria and is oxidized by ROS to rhodamine-123 (23). Rhodamine-123 fluorescence (excitation wavelength, 490 nm; emission, 535 nm) was measured on a microplate reader (Safire; Tecan US, Research Triangle Park, NC), and signal intensity was expressed qualitatively in relative units.

Electrophoretic Mobility Shift Assay (EMSA)
EMSA for NRF-1 was performed with [␥-32 P]ATP polynucleotide kinase-labeled oligonucleotides annealed to double-strand oligonucleotides. Binding reactions were carried out on 20 g of nuclear extract from liver or hepatoma cell nuclei. Binding assays were performed in a volume of 20 l with 5 g of bovine serum albumin and 1 g of sonicated salmon sperm DNA. Preliminary experiments were performed to determine the range of nuclear protein that generated a linear signal response for the gel shift on autoradiographs. Control binding experiments were performed under conditions of mutant NRF-1 probe. For supershifts, the reactions included 1 l of undiluted goat anti-NRF-1 antiserum. For competition, excess unlabeled oligonucleotide was incubated with nuclear extract before adding labeled oligonucleotide. After 20 min at room temperature, reaction mixtures were loaded on prerun 5% native polyacrylamide for 2.5 h at 10 V/cm, and the gels were dried and exposed to x-ray film. The DNA oligonucleotide sequences used for EMSA were as follows (NRF-1 recognition sites are underlined): human mtTFA NRF-1A, 5Ј-CGCTCTCCCGCGCCTGCGCCAATT-3Ј; NRF-1B, 5Ј-GGGCGGAATTGGCGCAGGCGCGGG-3Ј. Control oligonucleotides were synthesized as follows (mutated regions underlined): muNRF-1C, 5Ј-CGCTCTCCGAAAATGAAAACAATT-3Ј; muNRF-1D, 5Ј-GGGC-GGAATTGTTTTCATTTTCGG.

Statistics
Data were expressed as mean Ϯ S.E. Statistics were performed by ANOVA followed by Tukey's post hoc comparison using computer software. A p Յ 0.05 was considered significant. n values refer to independent replicates. Regression analysis was performed using Statview (SAS, Version 5.0.1).

RESULTS
Oxidative Stress and Hepatic Cell and Mitochondrial Injury after LPS-LPS depletes hepatic glutathione (GSH) and increases glutathione disulfide (GSSG/GSH) even in highly coupled mitochondria (4); this stress occurred within 6 h and recovered by 48 h (not shown). GSH recovery was associated with increased expression of the regulatory heavy catalytic subunit of ␥GCS (24). The mRNA for ␥GCS heavy chain in- Competitive PCR demonstrated transient decreases in mtDNA copy number coincident with GSH depletion/recovery (Fig. 1B). mtDNA copy number decreased significantly 6 h after LPS, stabilized at 24 h, and had recovered by 48 h. This coincided with an increased hepatocyte mitotic index (Fig. 1C).
Nuclear and Mitochondrial mRNA-Mitochondrial transcription was checked using gene-specific primers for mitochondrial mRNA for four mitochondrial proteins ( Fig. 2A), COX I, ND1, and ATPase 6 referenced to nuclear 18 S rRNA. At 24 h significant decreases were found in ATPase 6, COX I, and ND1 transcripts (22 to 27%), which recovered at 48 h. Because mitochondrial mRNA is transcribed as single polycistronic molecules from mtDNA heavy chain, similar COX I, ND1, and ATPase 6 mRNA contents indicate near steady-state conditions. Message for nuclear-encoded Complex IV subunits (cytochrome c oxidase, COX IV, COX V), and cytochrome c were analyzed by RT-PCR normalized to the stable nuclear-encoded 18 S rRNA (Fig. 2B). COX IV and cytochrome c mRNA doubled 24 h after LPS whereas COX V mRNA did not change relative to 18 S rRNA. Mitochondrial DNA Replication-Mitochondrial biogenesis requires mtDNA replication under nuclear regulation. To investigate mtDNA regulation, we examined three nuclear genes encoding proteins for mtDNA replication (Fig. 3). DNA polymerase ␥ (Pol␥) replicates mtDNA, mitochondrial singlestranded DNA-binding protein (mtSSB) serves DNA replication, repair, and recombination, and mtTFA is the only currently known activator of mammalian mitochondrial transcription. Pol␥, mtSSB, and mtTFA transcripts increased significantly by 6 h after LPS, further at 24 h, and by 48 h had declined toward pre-injury baseline.
LPS Induced Protein Binding of NRF-1-The transcription factors NRF-1 and NRF-2 are involved in regulating expression of nuclear genes encoding major mitochondrial proteins that regulate mtDNA transcription and replication (18). We investigated the effect of LPS on NRF-1 activation by EMSA. An increase in NRF-1 binding was evident at 6 h after LPS exposure. Approximately 70% of the increase in NRF-1 protein binding was lost 24 h after LPS exposure (Fig. 4A). To confirm specificity of NRF-1 binding, we performed supershift assays with anti-NRF-1 antibody, and a mutated oligonucleotide sequence for NRF-1 interfered with protein binding (Fig. 4B).
The murine NRF-1 peptide sequence revealed an Akt phos-phorylation motif (RXRXX(S/T)) at position 104 -109. Therefore, we assessed Akt activation by Western blot using anti-Akt and anti-phospho-Akt antibodies. To activate Akt fully, phosphorylation of two sites is necessary, one in the activation domain, and the other in the COOH-terminal hydrophobic motif (25). Phospho-Akt increased significantly 6 h after LPS treatment and declined by 24 h but remained above baseline (Fig. 4C). LPS Induced Expression of NRF-2-We analyzed changes in NRF-2 mRNA by RT-PCR in liver after LPS. NRF-2 is also a nuclear regulatory protein for mitochondria; its mRNA was expressed at low levels in control liver (Fig. 4D), but transient NRF-2 mRNA induction was found within 6 h of LPS injection followed by a decline over the next 24 h (Fig. 4D). Up-regulation of NRF-2 expression is novel, and probable transcriptional regulation of this gene has not been demonstrated before.
Cell Growth Regulatory Proteins-Cell cycle activation evaluated with a PCNA marker was increased significantly (p Ͻ 0.05) by LPS at 24 and 48 h (Fig. 5A). To determine whether G 1 cyclin expression accompanied recovery from LPS injury, cyclin D1 and cyclin-dependent kinase-4 (Cdk4) expression were checked by Western analysis in liver nuclei and cytoplasm (Fig.  5B). At 24 h, cytoplasm and nuclear cyclin D1 protein levels were stable and nearly undetectable, but G 1 cyclin was found after 48 h in both cytoplasm and nuclei (Fig. 5B). In comparison, Cdk4 constitutively present in control cytoplasm was equivocal in the nucleus. Cytoplasmic Cdk4 was unaffected by LPS, but the nucleoprotein was expressed significantly by 48 h (p Ͻ 0.05) (Fig. 5B). Nuclear translocation of D1 and Cdk4 indicate cell cycle progression.
Respiration-deficient Cells-To assess the relationship between mitochondria and hepatic cell proliferation we selected a rat hepatoma (H4IIE) cell line that constitutively expresses NRF-1 and mtTFA and generated mtDNA depleted clones. Southern analysis of DNA from parental (H4IIE) and EB derivative cells identified two clones (⌬H4IIE and 1⌬H4IIE) as mtDNA-deficient cells (Ͼ98% reduction of the mtDNA content) (Fig. 6A). These cells grew slowly, depended on glycolysis, and required pyruvate and uridine (not shown). Both clones responded similarly, but the reported data were obtained from clone ⌬H4IIE. To assess nuclear genes that control mitochondrial biogenesis, we examined mRNA for transcription factors NRF-1 and mtTFA in H4IIE and ⌬H4IIE cells. Fig. 6B shows expression of NRF-1 and mtTFA relative to 18 S rRNA mRNA by RT-PCR in control and mtDNA-depleted cells. Depletion of mtDNA significantly down-regulated biogenesis gene expression in ⌬H4IIE cells compared with H4IIE cells (Fig. 6C).
Deletion of mtDNA decreased mitochondrial transmembrane potential and ROS generation relative to control cells as determined using fluorescence techniques (Fig. 7). The ⌬H4IIE cells, like H4IIE cells, took up the fluorescent dye rhodamine-123 initially but lost it rapidly, whereas H4IIE cells retained the dye (Fig. 7A). The effect of mtDNA depletion on ROS generation was assessed by oxidation of dihydrorhodamine (Fig. 7B). In comparing parent (H4IIE) with mtDNA-deficient (⌬H4IIE) cells, significantly less fluorescence was detected in ⌬H4IIE cells, but t-BH exposure caused both cell types to accumulate fluorescence signal in the mitochondria.
Strong oxidants such as t-BH affect cell growth and viability by both direct oxidation and through secondary ROS generation (26). In H4IIE and ⌬H4IIE cells, viability was decreased significantly in a concentration-dependent manner after exposure to t-BH at concentrations of 50 M or above for 24 h (not shown). However, lower concentrations of t-BH had either no effect or stimulated growth. To avoid cytotoxicity, proliferation studies were performed using 15 M t-BH, which enhanced ⌬H4IIE cell growth (and survival) significantly, restoring it almost to the level of control cells. H4IIE cells exhibited slight increases in growth rate at 24 to 48 h with t-BH (Fig. 7C).
Translocation of mtDNA Regulatory Proteins-Translocation and binding of transcription factors NRF-1 and -2 induce mt-TFA expression in response to oxidative damage of mtDNA (27,28). Therefore, we probed nuclear extracts of H4IIE and ⌬H4IIE cells with anti-NRF-1 by Western analysis after exposure to t-BH. The t-BH increased NRF-1 expression in both cell lines within 1 h and was pronounced at 3 h. NRF-1 expression extended through 6 h in ⌬H4IIE but decreased in control cells (Fig 8A).
Activation of Akt-Akt is activated by phosphorylation, and oxidant-activated PI3K binds with high affinity and specificity to the Akt PH domain. We assessed Akt by Western analysis after oxidant treatment of cells grown in serum-free medium for 12 h to reduce basal phosphorylated Akt. Using antibody that detects only serine-473 phospho-Akt, serum control showed Akt phosphorylation within 15 min, a maximal increase after 1 h that was sustained for 6 h (not shown). To determine whether t-BH promotes Akt phosphorylation, cells were incubated with t-BH in serum-deficient medium for 1 to 3 h, and phospho-Akt accumulation was compared in ⌬H4IIE and H4IIE cells. To evaluate the involvement of PI3K in t-BHinduced Akt phosphorylation, cells were pretreated with 50 M PI3K-specific inhibitor LY294002 for 1 h. Subsequent immunoblots demonstrated that LY294002 blocked t-BH-mediated Akt phosphorylation, indicating the involvement of PI3K in t-BH-induced Akt phosphorylation (Fig. 8B).
Role of PI3K in NRF-1 Activation-To determine whether PI3K was involved in activating NRF-1 binding, multiple experiments were carried out to determine the effects of PI3K inhibitors on NRF-1 binding activity. H4IIE and ⌬H4IIE cells incubated with t-BH and 50 M LY294002 for 3 h showed NRF-1 activity suppression by lack of DNA binding on EMSA. A representative gel with supershift analysis (Fig. 8C) demonstrates that PI3K is necessary for oxidant to activate NRF-1 especially in o hepatoma cells.

FIG. 4. LPS induced expression and nuclear binding of NRF-1 and Akt in liver.
A, DNA binding specificity of NRF-1 in liver nuclear extract. NRF-1 32 P-labeled consensus oligonucleotides were incubated with 20 g of nuclear protein after optimizing the binding conditions. Lanes 1 and 2 indicate reactions with radiolabeled oligo alone. Lanes 3-8 represent reactions with liver nuclear extract; arrow denotes NRF-1 complex. Nonspecific complex (NS) is also indicated. B, supershift analysis of NRF-1 binding activity. Nuclear proteins were isolated from rat livers treated with LPS for 6 h. Before gel-shift assay, protein extracts were incubated with 1 l of NRF-1 antibody. The position of specific bands is indicated by the arrow; the supershifted (SS) band is indicated by the arrow. The specificity of NRF-1 binding activity was confirmed by the absence of NRF-1 complex band in the negative control (muNRF-1 probe). C, induction of P-Akt in the liver after LPS. Western blots were performed using antibody to P-Akt (top). Anti-total Akt antibody serves as internal control (bottom). D, mRNA expression of NRF-2. Messenger RNA levels of NRF-2 were determined by semiquantitative RT-PCR according to "Materials and Methods" (top) and normalized to 18 S rRNA (bottom). Data are representative of four studies.

DISCUSSION
This study provides evidence that LPS stimulates the expression of genes regulating mitochondrial biogenesis in mammalian liver. Early effects of LPS transiently deplete mitochondrial GSH and decrease mtDNA copy number in vivo. The ordering of the recovery events implies biogenesis, which precedes completion of hepatic cell proliferation according to the line of reasoning that follows.
Mitochondria produce ROS (1, 7), and LPS causes mitochondrial oxidative stress (shown by GSH depletion). Mitochondrial GSH content correlates with mtDNA oxidation (28,29), which pinpoints mitochondria as both sources and targets of ROS generated after LPS. An oxidant stress response was confirmed by increases in the expression of the rate-limiting enzyme for de novo glutathione synthesis (␥GCS), which should attenuate mtDNA damage. LPS-induced cytokine expression stimulates ROS production by mitochondria e.g. via TNF-␣, and GSH depletion renders mitochondria susceptible to oxidative damage. GSH synthesis after ␥GCS transcription and translation contributes to recovery of GSH levels, thereby limiting further oxidative damage.
LPS-induced loss of hepatic mtDNA copy number was detectable within 6 h, but by 48 h, mtDNA copy number had recovered. Although mitochondria can function and even divide at low mtDNA copy number (30), copy restitution is required for adequate mitochondria-encoded OXPHOS protein expression and nascent organelle function during proliferation. Low mtDNA copy number interferes with maintenance of mtDNA necessary to sustain energy production. Low copy number has been implicated in mitochondrial diseases and zidovudine-induced myopathy (31) and adversely affects mitochondrial function and gene expression (32).
Mitochondria are not autonomous with respect to mtDNA replication; the nucleus coordinates nuclear and mitochondrial gene expression to meet the need for OXPHOS. For example, thyroid hormone increases mitochondrial mRNA in liver and muscle in hypothyroid rats (33,34) and increases steady-state nuclear mitochondrial F 1 -ATPase and adenine nucleotide translocase transcripts (35). In this study, mtDNA template damage after LPS significantly decreased mitochondrial mRNA, whereas selected nuclear-encoded mRNA for OXPHOS proteins increased, e.g. cytochrome c and COX IV. In concert, nuclear mRNA for regulatory proteins of mtDNA transcription increased including mtSSB, mtTFA, and DNA Pol␥. These proteins maintain and replicate the mitochondrial genome, and mtTFA is pivotal in both processes (36). mtTFA level varies with mtDNA copy number in mitochondriopathies (36,37), and mtTFA gene disruption demonstrates that mtTFA is essential for biogenesis and embryogenesis (17).
The increase in mRNA expression (ϳ5-fold) for mtTFA supports increased protein and mtTFA-DNA binding activity with LPS (4), consistent with activation of the mtTFA promoter. Transcription of mtTFA is regulated in part by NRF-1 and by NRF-2 (18); the increase in NRF-1 protein and binding to the cis element here are consistent with this notion. The promoter regions of nuclear genes involved in mitochondrial biogenesis (mtTFA, mtSSB, DNA Pol␥) reveal consensus motifs recognized by NRF-1 (and/or NRF-2) implicating them in nuclearmitochondria regulation (38). Regulation by NRF-1 and NRF-2 are also revealed in studies of subunit 2 of succinate dehydrogenase (39). It is not yet clear, however, why some mitochondrial genes appear to be regulated by one or both factors, whereas others do not have NRF response elements.
New mitosis induced by LPS is detectable by histology only after evidence of biogenesis has been established (4). Biogenesis maintains the volume density of mitochondria, and proliferation requires additional energy. The liver compensates for necrosis with cell division, and expression of PCNA and G 1 /S cyclin (cyclin D1 and Cdk4) was detected within 48 h after LPS injury. Cyclin D1 mediates progression from the quiescent gap 0 (G 0 ) phase of the cycle through G 1 (40), and cyclin D1 and Cdk presence in the nucleus after LPS indicates binding associated with mitosis-regulating retinoblastoma kinase activity (41). Regulation of cyclin D1/Cdk4 complex formation and translocation varies among cells (41), but in late G 1 and during S phase PCNA (cyclin A) binds DNA polymerase-␦ (42,43), and its expression is sensitive to proliferation (44) and consistent with liver regeneration (45). Though the mechanisms coordinating biogenesis and cell division are unknown, NRF-1 activation suggests a role for redox signaling by ROS.
LPS induces ROS generation by liver mitochondria associated with TNF-␣ production during pre-replication, which although potentially dangerous, also promotes cell survival and proliferation (40 -45). TNF-␣ regulates PI3K/Akt (46), and Akt inhibits apoptosis and stimulates cell growth (47,48). Akt contains a central motif with Ser or Thr specificity, and the NH 2 terminus includes pleckstrin homology essential for lipidprotein or protein-protein interactions. On activation, Akt induces anti-apoptotic effects through phosphorylation of Bad (49) and caspase-9 (50). Akt stimulates cyclin D1, which promotes cell growth (51) and protects murine hepatocytes from TNF-␣ and Fas (CD95)-mediated apoptosis (52). A key implication of Akt activation by LPS is that it may prevent apo- ptosis, e.g. mediated by TNF, but not necrosis. Akt activation would thus promote replication in surviving hepatocytes. LPS also activates PI3K/Akt in human monocytes to limit TNF-␣ induction (53).
The in vivo study fully supported the concept that mitochondrial biogenesis increases after LPS injury but did not provide a specific mechanism. There is evidence that ROS regulate nuclear genes (54) and that mtDNA replication and transcription depend on nuclear-encoded proteins some of which are redox-sensitive, e.g. NRF-1 (55). Stimulation of these oxidantsensitive transcription factors by LPS raised the possibility that mitochondrial ROS regulated biogenesis. To investigate this possibility, we generated o clones from rat hepatoma cells that constitutively express NRF-1 and mtTFA. These cells were found to have diminished mitochondrial ROS production and grew slowly compared with controls in accordance with studies of human o cells (56). After mtDNA knockout (⌬H4IIE cells), nuclear NRF-1 translocation decreased, but upon exogenous oxidant exposure, NRF-1 translocation and growth returned to control levels. These findings clearly demonstrated that mitochondrial ROS generation influences the redox regulation of NRF-1. ⌬H4IIE cells also showed a lower expression of mtTFA mRNA similar to other o cells (37).
In addition to ROS, modulation of mitochondrial gene expression by other nuclear factors is possible. For example, communication between mitochondria and nucleus may involve ATP/ADP and depend on mitochondrial ATP generation (57). We could not exclude this possibility, because mtDNA depletion impairs OXPHOS, and ATP turnover should be lower in o than control cells. That oxidant treatment increased NRF-1 translocation and cell growth in ⌬H4IIE cells, however, indicates the importance of oxidative signaling, because the o cell still responds to it.
We also found that Akt expression, which is regulated by oxidants (58), was increased by t-BH via PI3K activation in the o cell. Both kinases are integral to cell growth, survival, and transformation. Because Akt is redox-sensitive, PI3K activation of Akt was a candidate for NRF-1 phosphorylation and activation. Inhibition of PI3K suppressed NRF-1 binding activity and Akt phosphorylation by t-BH, thus demonstrating involvement of PI3K/Akt in activating nuclear NRF-1-binding proteins. Although PI3K/Akt activation preceded NRF-1 acti-

FIG. 6. Effect of mtDNA depletion on biogenesis in rat hepatoma cells.
A, Southern analysis of DNA of parental H4IIE cell and mtDNA-depleted clone. DNA was digested with XbaI and hybridized with labeled probe for rat cytochrome b. Arrow indicates predicted 1.5-kb band present in parental (lanes 1 and 2) but not in o cell lines. B, depletion of mtDNA and regulation of mitochondrial biogenesis. Differences in mtTFA and NRF-1 detection by RT-PCR are relative to 18 S rRNA in rat liver and rat ⌬H4IIE and H4IIE cells (lanes 1, 2, and 3, respectively). Comparisons of mtTFA and NRF-1 signals relative to 18 S rRNA by densitometry are shown in C. Densitometry values are mean Ϯ S.E. of four studies (*, p Ͻ 0.05).

FIG. 7.
A, effect of mtDNA depletion on transmembrane potential. Retention of Rhodamine-123 was measured in H4IIE and ⌬H4IIE cells. Cells were treated with rhodamine-123 at 5 g/ml, and fluorescence was recorded after 10 and 40 min. B, mitochondrial ROS production after mtDNA depletion. ROS were detected by dihydrorhodamine fluorescence in H4IIE and ⌬H4IIE cells under control conditions (C) and 1 h after 15 M t-BH treatment (tBH) (*, p Ͻ 0.05 relative to C; †, p Ͻ 0.05, C relative to ⌬H4IIE). C, growth curves of H4IIE and ⌬H4IIE cells. Cells were grown for 48 h, trypsinized, and stained with 0.4% trypan blue, and viability was determined by counting in a hemocytometer. Differences in growth kinetics and viability between the two cell types were statistically significant by ANOVA (*p Յ 0.05), and ⌬H4IIE cell growth recovered with t-BH (p Ͻ 0.05). vation, whether Akt phosphorylates NRF-1 directly or requires related kinase activity remains to be established. However, we identified an Akt phosphorylation motif (RXRXX(S/T)) in the murine NRF-1 sequence (position 104 -109), and NRF-1 phosphorylation enhances its DNA binding activity (59). Thus, NRF-1 might control ETC protein expression in response both to extracellular signals and to intracellular [ATP]. Additionally, ROS originating in mitochondria, cytoplasm, or outside the cell could activate Akt and promote NRF translocation. Hence, PI3K-Akt activation may regulate both mitochondrial biogenesis and cell proliferation after oxidative stress.
In summary, we report for the first time that hepatic oxidative stress induced by LPS increases NRF-1 nuclear activity and mtTFA gene expression and stimulates mtDNA replication, pro-survival kinase activation, and cell proliferation in vivo. Moreover, o cells generated from hepatoma cells that constitutively express NRF-1 and mtTFA reveal that loss of mitochondrial ROS generation is associated with loss of NRF-1 expression and growth delay, which can be restored by oxidants. The conclusion is that ROS signaling is necessary for hepatic mitochondrial biogenesis accompanying cell proliferation, which is linked to activation of pro-survival PI3K-Akt pathways.