Cell culture and microscopy.

Mice with loxP recombination sites that flank exons that encode the Gabpa ETS domain (floxed Gabpa [Gabpa^fl/fl]) were previously described (16). MEFs were prepared from Gabpa^fl/fl or Gabpa^+/+ embryos at embryonic day 12.5 (E12.5) to E14.5 and maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Grand Island, NY) supplemented with 10% fetal bovine serum (Invitrogen), except where serum starvation is explicitly described. Retroviruses were packaged in the helper-free Phoenix packaging cell line (ATCC, Manassas, VA). For retroviral infection, MEFs at passage 2 were infected with either pBABE-Puro (empty virus) or pBABE-Cre for 48 h and selected for 3 to 5 days in medium containing 2.5 μg/ml puromycin (Sigma-Aldrich, St. Louis, MO). Electron microscopy and fluorescence microscopy were supported by Rhode Island Hospital Core Services.

Semiquantitative RT-PCR and real-time PCR.

Total RNA was prepared with Qiagen (Valencia, CA) RNeasy and reverse transcribed with the Invitrogen Superscript II RT system and poly(dT), and serial 2-fold dilutions of cDNA were subjected to PCR within the linear range of the assay (25 to 35 cycles, depending on the transcript). PCR products were resolved on 1.5% agarose gel, visualized by ethidium bromide, and scanned with an Amersham Typhoon image scanner (Amersham, Piscataway, NJ). Real-time PCR was performed with a Qiagen real-time PCR master kit on a real-time thermal cycler from Stratagene. The sequences of the primers used for reverse transcription (RT)-PCR, real-time PCR, or genomic PCR will be provided upon request.

Analysis of apoptosis.

The Apo-ONE homogeneous caspase 3/7 assay was performed according to the manufacturer's protocol (Promega Corporation, Madison, WI). Staurosporine (Sigma-Aldrich) was applied to cells at 1 μM for 24 to 48 h to induce apoptosis. Intracellular ATP detection was performed according to the manufacturer's protocol (Roche Applied Science).

Mitochondrial quantification and membrane potential.

Mitochondria in wild-type and Gabpα-null MEFs were counted in three separate fields. Mitochondrial mass was evaluated by staining cells with 50 nM NAO (nonyl acridine orange, a membrane potential-insensitive mitochondrial dye; Molecular Probes, Grand Island, NY) or MitoTracker Green FM (Invitrogen) for 30 to 60 min at 37°C (2). Mitochondrial membrane potential was measured by the incubation of MEFs with the MitoTracker probe 5,5′,6,6′-tetrachloro-1,19,3,39-tetraethybenzimidazol carbocyanine iodide (JC-1; Molecular Probes) or MitoTracker Red (Invitrogen) at 5 μg/ml at 37°C for 30 to 60 min (19), followed by flow cytometry or fluorescence microscopy. As a control, wild-type MEFs were incubated with dimethyl sulfoxide or 100 nM valinomycin (a potassium ionophore that dissipates the membrane potential) for 24 h before JC-1 staining.

Western blotting.

Cells were washed in phosphate-buffered saline, and total cellular proteins were prepared by the addition of Laemmli sample buffer, followed by incubation at 100°C for 5 min. Extracts that corresponded to 5 × 10⁵ cells were electrophoresed in 8 to 10% polyacrylamide gel, transferred to Amersham Hybond-P membrane, and probed with antibodies to GABPα, β-actin, procyclic acidic repetitive protein or poly(ADP-ribose) polymerase (PARP), Tomm20, Tomm70 (Santa Cruz Biotechnology), Vdac1 (Cell Signaling), and TFB1M (a gift of Gerald Shadel, Yale University, New Haven, CT).

Oxygen consumption and glycolysis.

Log-phase cells were grown in 24-well XF24 assay plates (Seahorse Bioscience) at 37°C with 5% CO₂ for 24 h before analysis. Cells were then incubated with assay medium and transferred to a non-CO₂ incubator for 1 h. Oxygen consumption and glycolysis were measured with a Seahorse XF24 analyzer with a Cell Mito Stress Test kit and a glycolysis stress test kit according to the manufacturer's protocols.

ChIP.

Chromatin was immunoprecipitated with antiserum against IgG and GABPα. Chromatin immunoprecipitation (ChIP) was performed as previously described (16).

Pulse-labeling of mitochondrial protein translation.

Labeling of mitochondrial translation in MEFs was performed as previously described (7). Briefly, MEFs were preincubated with methionine-free DMEM for 15 min, followed by methionine-free DMEM for labeling containing 100 μCi/ml ³⁵S (MP Biomedicals, Santa Ana, CA) and 100 μg/ml emetine (Sigma-Aldrich, St. Louis, MO), and chased for 10 min in regular DMEM. Total cellular protein was run on 4 to 20% polyacrylamide gradient gels, which were dried for 1 h and autoradiographed.

Disruption of Gabpα in mice and MEFs.

We used homologous recombination to generate mice in which loxP recombination sites flank exons that encode the Gabpa ets DNA-binding domain (Fig. 1A). Heterozygous floxed (Gabpa^fl/+) mice were bred with mice that bear the cytomegalovirus-Cre transgene and express Cre recombinase in all of their tissues. The resultant Gabpa^+/− mice were phenotypically normal, but intercrossing of these hemizygous mice generated no nullizygous mice among 35 pups. This finding is consistent with previous reports that Gabpα nullizygosity causes an embryonic lethal defect (16, 17).

We prepared MEFs from wild-type (Gabpa^+/+) or homozygous floxed (Gabpa^fl/fl) embryos and infected them with either pBABE-Cre, which expresses Cre recombinase, or the empty control retrovirus pBABE-Puro. Cre efficiently deleted Gabpa (Fig. 1B), the Gabpa transcript was absent (Fig. 1C), and Gabpα protein was not detected (Fig. 1D) in Gabpa^fl/fl MEFs (referred to here as Gabpa^−/− or Gabpα-null cells).

Gabpα^−/− MEFs have a reduced mitochondrial cell mass.

We sought to determine if deletion of Gabpa affected the morphology or the total cellular content of mitochondria. Electron microscopy demonstrated no discernible abnormalities in mitochondrial morphology (Fig. 2A), but there was a statistically significant reduction of mitochondrial content in Gabpα-null MEFs (Gabpa^fl/fl MEFs infected with pBABE-Cre) compared with that of control cells (Gabpa^fl/fl MEFs infected with pBABE-Puro) (Fig. 2B) (P < 0.02).

In order to validate the degree of mitochondrial loss in Gabpα-null MEFs, we incubated cells with NAO, which binds cardiolipin and is retained in mitochondria without regard to their energetic state or membrane potential. NAO staining demonstrated a 40% reduction in mitochondrial mass in Gabpα-null cells, compared to that in control MEFs (Fig. 2C, left) (P < 0.04). We further confirmed this result by staining cells with MitoTracker Green FM, a fluorescent dye that stains live cells and localizes to mitochondria regardless of their membrane potential (Fig. 2C, right) (P < 0.05). As an independent technique for determining the degree of mitochondrial loss, we measured the ratio of mtDNA genes to nuclear genes. The ratio of Cox I (an mtDNA gene) to the gene for actin B (a nuclear gene) was reduced by more than one-third in Gabpα-null cells (Fig. 2D) (P < 0.01). As another indicator of mitochondrial mass, we measured the levels of Vdac1, a gene that encodes a voltage-dependent anion channel protein that is a major component of the mitochondrial outer membrane. Immunoblotting of protein extract from control and knockout (KO) MEFs with antibodies against Vdac1 and β-actin demonstrated a reduction of Vdac1 protein in KO MEFs (Fig. 2E). Thus, these five independent methods demonstrate that Gabpa deletion reduces the cellular content of mitochondria by approximately one-third.

Mitochondrial membrane potential is not altered in Gabpα^−/− MEFs.

We stained cells with the cationic dye JC-1 to determine if Gabpα loss impairs the mitochondrial membrane potential. Emission of both orange and green fluorescence by flow cytometry following JC-1 staining indicates mitochondrial membrane potential integrity. Treatment of wild-type MEFs with the antibiotic valinomycin served as a positive control for mitochondrial membrane potential disruption (Fig. 3A). However, we observed no loss of mitochondrial membrane potential in Gabpα-null MEFs, compared to control MEFs (Fig. 3A). As a complementary method, we stained MEFs with MitoTracker Red FM, a fluorescent dye that stains live cells dependent upon their mitochondrial membrane potential. Similar to the result of JC-1 staining, there was no decrease in MitoTracker Red-stained KO MEFs, compared to the control MEFs (Fig. 3B). Mitochondrial membrane potential integrity was confirmed by fluorescence microscopy following JC-1 or MitoTracker Red staining (data not shown). Thus, Gabpa disruption does not impair mitochondrial membrane potential.

Reduced energy production in Gabpα^−/− MEFs.

Generation of ATP by oxidative phosphorylation and electron transport is the major source of energy generation in the eukaryotic cell. We sought to determine if Gabpa disruption significantly affects cellular energy production. We measured ATP production by luciferase-associated fluorescence, which requires intracellular ATP as an energy source. As a positive control, addition of 10% serum to the growth medium of serum-starved, wild-type MEFs more than doubled ATP production (Fig. 3C, left panel). Treatment with oligomycin (an inhibitor of ATP synthase that is required for oxidative phosphorylation of ADP) or valinomycin abrogated the serum-induced increase in luciferase activity.

We examined ATP production by Gabpα-null MEFs and control cells under conditions of serum starvation and following serum treatment (Fig. 3C, right panel). ATP production by control Gabpa^fl/fl MEFs infected with pBABE-Puro more than doubled following the addition of serum. Gabpα MEFs exhibited 30 to 40% less ATP production than control MEFs under both serum-starved and serum-treated conditions. Thus, Gabpα^−/− MEFs exhibit reduced ATP production at levels that are commensurate with the reduction in mitochondrial mass.

Reduced oxygen consumption in Gabpα^−/− MEFs.

Because cellular ATP is derived from both mitochondrial oxidative phosphorylation and cytoplasmic glycolysis, reduced ATP production in KO MEFs could potentially be due to altered glycolysis. We evaluated both mitochondrial oxygen consumption and the cytoplasmic glycolysis of control and KO MEFs. As shown in Fig. 4A, there was a 30% reduction in the basal oxygen consumption rate (OCR) of KO MEFs (P < 0.001), but the maximal OCRs of control and KO MEFs were comparable. Meanwhile, control and KO MEFs demonstrated similar glycolysis capacities under either resting (basal) or stressed (maximal) conditions (Fig. 4B). Thus, the reduced ATP production was clearly due to decreased oxidative phosphorylation in the mitochondria of Gabpα^−/− cells.

No increase in apoptosis in Gabpα^−/− MEFs.

Mitochondria are key effectors of programmed cell death through caspase-stimulated release of cytochrome c. Gabpa deletion was reported to increase apoptosis in hematopoietic stem cells (20), but we previously observed no significant increase in apoptosis in Gabpα^−/− MEFs, as measured by annexin V and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling assays (21). We measured caspase 3/7 activity to further assess apoptosis in Gabpα-null cells. As positive controls, wild-type MEFs treated with staurosporine activated caspase 3/7 (Fig. 5A). However, Gabpα^−/− MEFs did not exhibit greater activation of caspase 3/7 than control MEFs.

We measured caspase-mediated PARP cleavage to further assess apoptosis. As expected, staurosporine treatment of wild-type MEFs increased the level of the 89-kDa cleaved form of PARP, relative to that of 113-kDa PARP (Fig. 5B). However, there was no increase in PARP cleavage in Gabpα-null MEFs, compared to that in controls. We conclude that Gabpa deletion does not increase apoptosis under normal glucose growth conditions.

Mitochondrial gene expression in Gabpα^−/− MEFs.

GABP has since been reported to transcriptionally activate or bind the promoters of more than a dozen mitochondrial genes, including the cytochrome c oxidase (COX), F1 ATP synthase β, ATP synthase coupling factor 6 (CF6), and Tfam genes (2). We performed quantitative RT-PCR to assess the expression of putative GABP mitochondrial target genes following Gabpa deletion. Among the components of electron transport and oxidative phosphorylation that we examined (Fig. 6A), the cytochrome c oxidase subunit IV isoform ii (COX IVii; P < 0.005) and COX Vb (P < 0.03) mRNA levels were significantly reduced and the COXIVi mRNA level was increased (P < 0.03) following Gabpa disruption, but there was no significant alteration of F1 ATP synthase β, ATP synthase CF6, or cytochrome c mRNA levels compared to those of control MEFs.

We examined the expression of the mRNAs of transcription factors and cofactors that are known to regulate mitochondrial gene expression or biogenesis. There was no significant change in the Tfam, Tfb2m, Nrf1, c-Myc, Pgc-1a, or Prc mRNA level (Fig. 6B). However, the expression of Tfb1m mRNA was decreased by more than 70% in Gabpα-null cells (P < 0.005). The reduction of Tfb1m protein in Gabpα-null MEFs was confirmed by immunoblotting (Fig. 6C). Both Gabpα and Gabpβ bound to the promoters of F1 ATPase β and ATPase CF6 in control cells, but neither component of the GABP complex bound in Gabpα-null cells (Fig. 6D); this is consistent with our previous observation that GABPβ cannot bind to ETS sites in the Cox Vb promoter in the absence of GABPα (18). We conclude that only some of its putative mitochondrial gene targets are affected by Gabpa disruption but that the ribosomal methyltransferase Tfb1m is dependent on Gabp for expression.

TFB1M is a limiting factor that controls mitochondrial protein synthesis (7, 8), and reduced Tfb1m expression after Gabpa deletion might account for the decreased mitochondrial biogenesis in Gabpα-null MEFs. We examined total mitochondrial protein synthesis by ³⁵S pulse-labeling and observed decreased synthesis of all mitochondrial proteins in Gabpα-null MEFs; the reduction in mitochondrial proteins was global and not limited to any specific mitochondrial protein (Fig. 6E). We also examined the possibility that protein transport defects account for reduced mitochondrial biogenesis by examining the mitochondrial membrane proteins Tomm20 and Tomm70, which are crucial for mitochondrial protein transport across the mitochondrial membrane. Neither Tomm20 nor Tomm70 protein expression was affected by Gabpa deletion, as shown in Fig. 6F. Together with the previous report that decreased Tfb1m resulted in decreased mitochondrial protein translation and mitochondrial mass (7), we conclude that Gabp regulates mitochondrial biogenesis through its control of the key mitochondrial factor Tfb1m.