CREB-1α Is Recruited to and Mediates Upregulation of the Cytochrome c Promoter during Enhanced Mitochondrial Biogenesis Accompanying Skeletal Muscle Differentiation

Abstract

To further understand pathways coordinating the expression of nuclear genes encoding mitochondrial proteins, we studied mitochondrial biogenesis during differentiation of myoblasts to myotubes. This energy-demanding process was accompanied by a fivefold increase of ATP turnover, covered by an eightfold increase of mitochondrial activity. While no change in mitochondrial DNA copy number was observed, mRNAs as well as proteins for nucleus-encoded cytochrome c, cytochrome c oxidase subunit IV, and mitochondrial transcription factor A (TFAM) increased, together with total cellular RNA and protein levels. Detailed analysis of the cytochrome c promoter by luciferase reporter, binding affinity, and electrophoretic mobility shift assays as well as mutagenesis studies revealed a critical role for cyclic AMP responsive element binding protein 1 (CREB-1) for promoter activation. Expression of two CREB-1 isoforms was observed by using specific antibodies and quantitative reverse transcription-PCR, and a shift from phosphorylated CREB-1Δ in myoblasts to phosphorylated CREB-1α protein in myotubes was shown, while mRNA ratios remained unchanged. Chromatin immunoprecipitation assays confirmed preferential binding of CREB-1α in situ to the cytochrome c promoter in myotubes. Overexpression of constitutively active and dominant-negative forms supported the key role of CREB-1 in regulating the expression of genes encoding mitochondrial proteins during myogenesis and probably also in other situations of enhanced mitochondrial biogenesis.

In mammals, mitochondria are composed of at least 1,000 proteins, including components of the inner membrane electron transport and oxidative phosphorylation system (OXPHOS), metabolite carriers, matrix enzymes, subunits of the protein import machineries, factors necessary for replication and expression of the small mitochondrial DNA (mtDNA) genome, and components of the mitochondrial protein biosynthesis machinery (5). To synthesize these proteins in a reasonably economical way, it is essential to orchestrate the expression of their genes, which are predominantly located on nuclear chromosomes, and coordinate it with the expression of mtDNA. As both ATP demand and mitochondrial content are very different in the various cell types of the body and can change even in terminally differentiated cells, these regulatory mechanisms must operate during developmental programs as well as in adaptation processes in the adult. Indeed, cells are able to adjust energy metabolism by altering the architecture and dynamics of the mitochondrial reticulum (10), by modifying its enzyme equipment and/or the level of proton leak, or by adjusting total mitochondrial respiratory capacity when changes in energy demand persist for long periods (23). Among the factors known to strongly stimulate mitochondrial biogenesis in vivo, the most prominent examples are high levels of thyroid (67) and glucocorticoid (55, 66) hormones and also conditions like endurance exercise of muscle (1) and cold adaptation in brown fat tissue (31).

While transcription of the two polycistronic transcripts containing the few genes carried by mtDNA is most likely regulated by the nucleus-encoded mitochondrial transcription factors TFAM and TFBM (15, 17, 20, 41, 58), coordination of nuclear genes encoding mitochondrial proteins (NEM genes) is much more complex and a network of regulatory pathways has been described. Promoter studies of such NEM genes indicated some frequent and recurrent features in the regulatory cascades (for a review, see reference 30). Many of the analyzed promoters of NEM genes contain binding sites not only for one or both of the nuclear respiratory factors NRF-1 and NRF-2 (or its mouse homolog GABP) but also for SP-1, estrogen-related receptor alpha, and members of the peroxisome proliferator-activated receptor/retinoid X receptor family (23, 54). Another common element in the promoters of NEM genes is the cyclic AMP (cAMP) responsive element (CRE) recognized by proteins of the CREB-1 transcription factor family, which are activated through phosphorylation by various protein kinases (8). CREB-1 is involved not only in signaling cascades transmitting external signals (neurotransmitters and hormones) to the nucleus via G-protein-coupled membrane receptors and a second messenger (cAMP) but is also a central target for a retrograde communication pathway signaling mitochondrial dysfunction to the nucleus, which involves no external signals but elevated intracellular Ca²⁺ levels (4).

The coordination of NEM gene expression seems to be governed by the coactivators PGC-1α, PGC-1β, and PGC-1-related coactivator, as these proteins were found to interact with and enhance the effects of the transcription factors mentioned above (30). However, none of these transcription factors and coactivators alone appears to be sufficient to regulate the entire set of genes encoding mitochondrial proteins during organelle biogenesis.

Skeletal muscle is one of the tissues with the strongest levels of dependence on mitochondrial function, as shown by the severe impacts of mitochondrial diseases on muscle performance in patients (61). In addition, mitochondrial dysregulation was demonstrated in muscle of patients suffering from type II diabetes (39); however, it is still unclear whether this is the cause or the consequence of insulin resistance. Thus, skeletal muscle is an attractive tissue for analyzing in depth the regulation of mitochondrial biogenesis. Transgenic, muscle-specific overexpression of the coactivator PGC-1α or PGC-1β in mice induces an impressive switch toward oxidative-type muscle fibers containing large amounts of mitochondria (3, 38). However, PGC-1α^−/− null mice still contain normal numbers of these fibers, and although mitochondrial gene expression and activity were blunted, mitochondrial fractional volume was unchanged in one study and only slightly decreased in another (2, 37). Also, PGC-1β^−/− null mice (36) or mice expressing a nonfunctional PGC-1β protein (64) show reduced oxidative capacity in muscle and other organs; however, these tissues still contain mitochondria. Thus, PGC-1α and PGC-1β may be very important factors during adaptation processes, but they are obviously not essential for mitochondrial biogenesis (2).

During embryonic development in vivo, muscle differentiation is accompanied by a pronounced rise in mitochondrial content (43), while in adult muscle, mitochondrial biogenesis is induced when its energy demand is enhanced via chronic nerve stimulation or endurance exercise (46). Also, during differentiation in vitro, the activities of various mitochondrial enzymes dramatically increase (1, 44) and the rather glycolytic energy production in proliferating myoblasts shifts to oxidative phosphorylation in terminally differentiated myotubes (35). However, it is unclear whether this enhanced mitochondrial content reflects an adaptation to the rising energy demand or whether it is due to a genetically controlled developmental program anticipating the future energetic costs of contraction and relaxation. As mitochondrial function is absolutely essential for successful differentiation and fusion of myoblasts (26) and dysfunction of mitochondria suppresses myogenin expression (33, 49), a direct interdependence of the muscle developmental program and the network regulating mitochondrial genes can be proposed.

To further elucidate the mechanisms regulating mitochondrial biogenesis in muscle, we comprehensively assessed mitochondrial content and oxidative capacity during differentiation of mouse myoblasts to myotubes in cell culture. These cells do not express PGC-1 proteins and are thus a perfect model for studying the basal mechanisms independent of these coactivators. The changes in energy metabolism and upregulation of selected NEM proteins were monitored, and their promoters, especially the cytochrome c (Cyt c) promoter, were analyzed by deletion and mutation studies as well as by transcription factor binding assays. Chromatin immunoprecipitation (ChIP) experiments and assays using constitutively active and dominant-negative constructs clearly show that CREB-1 is necessary and sufficient to drive the Cyt c promoter and that the preferential recruitment of phosphorylated CREB-1α in situ mediates upregulation of this promoter during mitochondrial biogenesis.

MATERIALS AND METHODS

Cell culture and analysis of proteins, RNA, and mtDNA.

C2F3 mouse myoblasts (12), a subclone of the C2 line able to form large, multinucleated myotubes, were propagated at a low density in Dulbecco's modified Eagle's medium plus 10% fetal calf serum supplemented with penicillin-streptomycin, 1 mM Na-pyruvate, and 1× Dulbecco's modified Eagle's medium nonessential amino acids (Invitrogen, San Diego, CA) at 37°C and 5% CO₂. To induce differentiation, cells were grown to confluence and fetal calf serum was replaced by 2% horse serum. Changes in lactate and glucose content in the cell culture medium were analyzed using an automated blood analyzer (EML 105; Radiometer, Copenhagen, Denmark) and normalized for cellular protein. Determination of total DNA, RNA, and protein contents is described in the supplemental material. The values were used to calculate the amounts of protein and RNA per DNA and used later to express the levels of specific proteins and mRNAs per nucleus in myoblasts and myotubes. Extraction of protein, RNA, and DNA for blotting as well as blotting procedures is described in the supplemental material.

Phosphatase treatment of nitrocellulose membrane.

Nuclear extracts from confluent cells were prepared according to a modified standard protocol (22), sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was performed, and the proteins were transferred to a nitrocellulose membrane. After being blocked for 1 hour, the membrane was incubated overnight at 4°C in phosphatase buffer (1% bovine serum albumin, 0.1% Triton X-100, 2 mM MnCl₂, 5 mM dithiothreitol, 1,200 U/ml lambda-phosphatase [Upstate, Charlottesville, VA] in Tris-buffered saline) with constant agitation and then washed once with phosphate-buffered saline (PBS)-0.1% Tween 20 and with Tris-buffered saline-0.1% Tween 20 several times. The membrane was probed with antibodies as described in the supplemental material.

Cloning procedures.

Promoter sequences of Cyt c and Cyt c oxidase subunit IV (COXIV) were amplified from rat genomic DNA using specific primers with suitable restriction sites and cloned into the pGL3 basic vector (Promega, Mannheim, Germany). The TFAM promoter construct was amplified using a plasmid containing the human promoter (a kind donation from S. Ohta, Kawasaki, Japan) as a template. Deletion constructs were made using endogenous restriction sites or PCR products. Mutations were introduced by PCR using mismatch primers and, when necessary, the mega-primer method (51). The constructs are described in the supplemental material.

RT-PCR/real-time RT-PCR.

Total RNA was extracted from logarithmically growing myoblasts, 100% confluent myoblasts, or differentiated myotubes with Trizol (Invitrogen, San Diego, CA) according to the manufacturer's recommendations. Reverse transcriptions (RTs) were made using a QuantiTect RT kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The mCREB-103 forward (5′ ACA TTA GCC CAG GTA TCC ATG CCA G 3′) and mCREB-396 reverse (5′ GGC CTC CTT GAA AGG ATT TCC CTT CG 3′) primers were used to amplify two products specific for the CREB-1α (293 bp) and CREB-1Δ (251 bp) transcripts. Samples from the RT-PCRs were collected at several consecutive cycles in order to determine the exponential phase of product accumulation.

Real-time RT-PCR was performed in triplicate for every sample with the ABI Prism 7900HT sequence detection system (Applied Biosystems, Foster City, CA). The CREB-1α forward (5′ CAG TCT CCA CAA GTC CAA ACA GTT 3′) and CREB-1α reverse (5′ TGA AAT CTG AGT TCC GGA GAA AA 3′) primers were used with the TaqMan probe CREB-1α (6-carbofluorescein-AGT CTT CCT GTA AGG ACT TA-MGBNFQ) to amplify the CREB-1α isoform. Since it is not possible to amplify CREB-1Δ alone, both isoforms were detected by the CREB-1α&Δ forward (5′ GGA ATC TGG AGC AGA CAA CCA 3′) and CREB-1α&Δ reverse (5′ CTG GGC TTG AAC TGT CAT TTG TT 3′) primers with the TaqMan probe CREB-1α&Δ (6-carbofluorescein-TGT AAC AGA AGC TGA AAA T-MGBNFQ). Glyceraldehyde phosphate dehydrogenase (GAPDH) was amplified as an endogenous control gene in the same reaction plate (assay identification no. Mm99999915_g1; Applied Biosystems). The averaged cycle threshold (C_T) values of each reaction derived from the target gene, determined with ABI PRISM 7500 software, were normalized to GAPDH levels. ΔC_T was calculated using the following equation: ΔC_T = C_T of target − C_T of GAPDH.

Reporter gene assays.

Myoblasts were seeded into 12-well plates and cotransfected with a firefly luciferase expression vector (pGL3 basic; Promega, Mannheim, Germany) driven by the promoter sequence to be analyzed and a Renilla luciferase expression vector driven by a viral promoter used to correct for transfection efficiency using Metafectene (Biontex, Martinsried/Planegg, Germany). Twenty-four hours after transfection, subconfluent and proliferating myoblasts were scraped off and lysed in cold lysis buffer (25 mM Gly-Gly, 15 mM MgSO₄, 4 mM EGTA, 0.2% Tween 20, 1 mM dithiothreitol, pH 7.8). Confluent myoblasts were analyzed 48 h after transfection, and myotubes were harvested and lysed after 3 days of differentiation, i.e., 5 days after transfection. Cell debris was removed by centrifugation, and luciferase activities in the supernatant were determined as described in the supplemental material.

Modulation of endogenous CREB-1 level and function.

When indicated, the reporter plasmids were cotransfected with plasmids encoding constitutively active (C2/CREB) and/or dominant-negative (A-CREB) CREB constructs (62) into myoblasts. Nonconfluent myoblasts were analyzed 24 h after transfection and confluent myoblasts 72 h after transfection. As positive controls, firefly luciferase vectors containing the rat α-inhibin promoter with functional CRE sites (45) were used.

EMSA.

The electrophoretic mobility shift assay (EMSA) was performed with nuclear protein extracts of myoblasts and myotubes and ³²P-end-labeled, double-stranded oligonucleotide probes corresponding to rat Cyt c promoter sequences containing either the wild-type or a mutated sequence. Preparation of nuclear protein extracts, binding reactions, and analysis of protein-DNA complexes are described in detail in the supplemental material.

ChIP assay.

C2F3 myoblasts, confluent cells, or myotubes were cross-linked by addition of 1% formaldehyde for 10 min. Cross-linking was stopped by 125 mM glycine. Cells were washed with ice-cold PBS (170 mM NaCl, 33 mM KCl, 40 mM Na₂HPO₄, and 18 mM KH₂PO₄ in PBS, pH 7.2), scraped into 1 ml PBS containing Roche complete protease inhibitor cocktail, and collected by centrifugation. The cell pellets were resuspended in lysis buffer {5 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)], pH 8.0 [KOH], 85 mM KCl, 0.5% NP-40} containing protease inhibitors and incubated for 10 min at 4°C. Nuclei were collected by centrifugation and resuspended in nuclear lysis buffer (50 mM Tris, pH 8.1, 10 mM EDTA, 1% SDS, protease inhibitor cocktail) and incubated for 10 min at 4°C. The DNA was sonicated to get fragments of 200 to 500 bp, cooled on ice, and centrifuged. The supernatant was diluted fivefold with ChIP dilution buffer (16.7 mM Tris, pH 8.1, 167 mM NaCl, 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, protease inhibitor cocktail). To reduce nonspecific background, the chromatin was incubated with 80 μl of protein A-Sepharose (Amersham Biosciences, Buckinghamshire, United Kingdom) for 30 min at 4°C with agitation and centrifuged, and the supernatant fraction was collected. Twenty percent of the total supernatant was saved as a total input control. The chromatin solutions were incubated overnight at 4°C with 5 μl anti-CREB (no. 06-863; Upstate), 5 μl anti-P133-CREB (no. 06-519; Upstate), or 5 μl anti-CREB-1α (sc-58; Santa Cruz) antibodies. The immune complexes were collected by incubation with 60 μl of protein A-Sepharose for 1 h at 4°C. The beads were consecutively washed with low-salt-concentration wash buffer (20 mM Tris, pH 8.1, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 2 mM EDTA) at 4°C, high-salt-concentration wash buffer containing 500 mM NaCl, LiCl wash buffer (10 mM Tris, pH 8.0, 0.25 M LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA), and Tris-EDTA buffer (10 mM Tris, pH 8.0, and 1 mM EDTA) at 4°C. Immunocomplexes were eluted from the beads with 250 μl elution buffer (1% SDS, 0.1 M NaHCO₃) at room temperature. Cross-links were reversed by addition of NaCl (final concentration, 0.3 M), and RNA was removed by addition of RNase A (10 μg/sample) for 4 h at 65°C. The DNA was precipitated by adding 2.5 volumes of ethanol; resuspended in 100 μl of a solution containing 40 mM Tris, pH 6.5, 10 mM EDTA, and 20 μg proteinase K; and incubated for 1 h at 45°C. The DNA was extracted by QIAquick spin columns. Immunoprecipitated and purified DNA was analyzed by PCR using the primers 5′-GTT ACC TGA GCC GAG CCA CAC-3′ and 5′-TGA CGT AAC CGC ACC TCA TTG G-3′, which are specific for mouse Cyt c promoter and were used to amplify a promoter fragment containing NRF-1 (bp −156/−145) and CREB-1 (bp −262/−255 and −110/−103) recognition sites. PCR products were electrophoresed on agarose gels and visualized by ethidium bromide staining.

RESULTS

During embryonic development in vivo, proliferating myoblasts exit the cell cycle, differentiate, and fuse to multinucleated, contractile skeletal muscle cells. In cell culture, the stimulus for differentiation is confluence of the cells, enhanced by the removal of growth factors normally supplied by fetal serum. After 3 to 5 days under differentiating conditions, routinely 60 to 70% of C2F3 myoblasts had formed syncytial myotubes (see Fig. S1A in the supplemental material). This is an anabolic process accompanied by a marked increase in cell volume, thus going along with a dramatic increase in total RNA and protein. Indeed, we found that myoblasts contained 91 ± 5 ng RNA and 79 ± 7 μg protein per mononucleated cell, while myotubes contained 898 ± 61 ng RNA and to 297 ± 19 μg protein per nucleus. In order to allow useful comparisons, all further data were normalized for DNA as the appropriate denominator, i.e., to the number of nuclei present in the preparation.

Increased O₂ consumption and total ATP turnover during muscle differentiation.

Anaerobic energy production was determined by measuring lactate release (Table 1). Confluent cells and myotubes produced about two times more lactate and consumed about two times more glucose than myoblasts. As most of the glucose is converted to lactate, oxidative phosphorylation obviously consumes additional substrates, such as amino or fatty acids or pyruvate provided by the culture medium.

TABLE 1.

Glucose consumption and lactate release in murine C2F3 myoblasts and myotubes

Cell type	Rate (fmol/nucleus/min) of:
	Glucose consumption	Lactate release
Myoblast	8.13 ± 2.6	18.5 ± 7.9
Confluent	21.2 ± 3.3	30.0 ± 4.7
Myotube	17.8 ± 2.9	32.3 ± 18.8

Myoblasts showed an oxygen uptake of 4.5 ± 2.2 fmol/min per cell under coupled conditions, a typical value found for proliferating cells (11). Oxygen consumption was eightfold higher in myotubes (see Fig. S1B in the supplemental material), and Cyt c oxidase activity was increased ninefold (see Fig. S1C in the supplemental material). CCCP was applied (see Fig. S1B in the supplemental material) to allow measurement of maximal oxygen consumption, which was five times higher than that in the coupled state in myoblasts, while it was only moderately stimulated further in myotubes. This was probably due to diffusion limitation of oxygen into these large cells. However, in order to exclude different coupling of the respiratory chain as the reason for this difference, the respiratory control ratios in isolated mitochondria from myoblasts, confluent cells, and myotubes were determined but were found to be similar (see Fig. S1D in the supplemental material).

The total ATP production rate calculated from these data indicates that differentiation is accompanied by at least a fivefold increase in energy demand (Table 2), which cannot be covered sufficiently by anaerobic glycolysis, thus requiring an increase in mitochondrial biogenesis. In differentiated myotubes, ATP production is covered mostly by OXPHOS (87%), in contrast to ATP production in myoblasts (59%). As oxygen consumption in myotubes might have been underestimated due to diffusion limitation and since force development was blocked by 2,3-butanedione 2-monoxime during measurement of oxygen uptake, the total ATP production rate and the contribution of OXPHOS may be even higher.

TABLE 2.

Oxidative, anaerobic, and total ATP production rates in C2F3 myoblasts, confluent cells, and myotubes^a

Cell type	Production rate (fmol/nucleus/min) for indicated ATP type
	Oxidativeb	Anaerobic	Total
Myoblast	27 ± 13 (59)	18.5 ± 7.9	45.5
Confluent	81 ± 12 (72)	30.0 ± 4.7	111.0
Myotube	226 ± 44 (87)	32.3 ± 18.8	258.3

^aRates were obtained from data shown in Fig. S1B in the supplemental material and Table 1. A P/O ratio of 3 for coupled oxidative phosphorylation and an ATP/lactate ratio of 1 were used.

^bValues in parentheses are percentages of oxidative compared to total ATP production rates.

Activated NEM gene expression in myotubes.

Cyt c, TFAM, and COXIV were chosen as representative NEM proteins, and their expression was analyzed first by Western and Northern blotting (Fig. 1 and 2). When equal amounts of protein were loaded, as was usually done, no major differences in the abundances of Cyt c, TFAM, COXIV and the mtDNA-encoded subunit COXI were seen (Fig. 1A), while as expected the sarcomeric protein myosin heavy chain and the muscle differentiation marker myogenin were induced in myotubes. PGC-1 protein was not detectable in myoblasts or myotubes (data not shown), as also reported by others (34). The cytoskeletal protein β-tubulin, usually used as a loading control, was also induced during differentiation, while levels of cyclin-dependent kinase 4, an important driver of the cell cycle, dropped (Fig. 1A). However, this analysis does not take into account the dramatic cell growth accompanied by massive protein synthesis during differentiation. In order to demonstrate the actual increase in mitochondrial proteins per nuclear gene copy, protein derived from equal numbers of nuclei was loaded (Fig. 1B and C), and the abundances of Cyt c, TFAM, and COXIV were found to increase 9, 4, and 12 times, respectively.

FIG. 1.

Western blot analysis of mitochondrial proteins and markers of differentiation in myoblasts and myotubes. Equal amounts of proteins (A) or protein amounts corresponding to equal numbers of cells or nuclei (B) were loaded from three or four independent samples for each group, respectively, and the membrane was probed for the proteins as indicated. (C) Quantitative data derived from the results shown in panel B and expressed as means ± standard deviations (n = 4). * indicates a significant difference at P values of <0.05, and ** indicates a significant difference at P values of <0.01, between myoblast and myotube values obtained by Student's t test.

FIG. 2.

Abundance of mRNAs for Cyt c, TFAM, COXIV, and COXIII as well as mtDNA copy number in myoblasts and myotubes. (A) Northern blot analysis of NEM gene expression in the poly(A)⁺ fraction. (B) Quantification of results obtained from a Northern blot containing total RNA and normalized for 18S rRNA. Results are expressed as means ± standard deviations (n = 4). * indicates a significant difference at P values of <0.05), and *** indicates a significant difference at P values of <0.001, between myoblast values and myotube values obtained by Student's t test. Southern (C) and Northern (D) blot analyses of mtDNA and mitochondrially encoded COXIII mRNA. Four independent samples were loaded for each group, and results were normalized for 18S ribosomal DNA or 18S rRNA.

Similarly, no major differences in the abundances of Cyt c, TFAM, and COXIV mRNA were seen on a Northern blot containing the complete poly(A)⁺ RNA from equal amounts of total RNA (Fig. 2A). Three mRNA species were observed for Cyt c, as reported earlier (24). However, when these mRNAs were normalized to 18S rRNA levels on blots containing total RNA (not shown) and the 10-fold increase in total RNA per nucleus during differentiation was taken into account, levels of these mRNAs were found to be enhanced between 10- and 15-fold (Fig. 2B). The copy number of mtDNA per nucleus, determined as the ratio of mtDNA to the multicopy nuclear 18S rRNA gene, slightly decreased during differentiation (Fig. 2C) while the mRNA for the mtDNA-encoded subunit III of COX increased 2.5-fold (Fig. 2D).

In conclusion, mRNAs transcribed from NEM genes as well as from mtDNA and the corresponding proteins are strongly increased, together with total RNA and total cellular protein, during muscle differentiation, while mtDNA levels slightly decrease.

cis-acting elements responsible for NEM gene expression upon differentiation.

To analyze the mechanisms responsible for this impressive upregulation of NEM genes, the activities of the corresponding promoters were analyzed. Early myoblasts were transfected with reporter plasmids, and luciferase activities were measured after 1 (myoblasts), 2 (confluent), or 5 (myotubes) days. In pilot experiments using the protein synthesis inhibitor cycloheximide, the half-lives of both the firefly and Renilla luciferases were determined to be 5 to 6 h in these cells (data not shown), ensuring that measured luciferase enzyme activities actually reflect the promoter activity at the time of measurement rather than an accumulation of the luciferase proteins over time.

For Cyt c, the rat promoter sequence from bp −631 to +135 was chosen, a fragment larger than the functional promoter described in previous studies of other cell types (24). The rat COXIV promoter was cloned from bp −670 to +31. As studies were available for the human but not the rat or mouse TFAM promoter and as the human promoter was recently shown to be fully functional in transgenic mice (15), a human sequence from bp −414 to +43 was chosen for analysis (63).

All three full-length promoter constructs showed clear activations during differentiation (Fig. 3A). By systematic deletion of promoter sequences from the 5′ end, regions necessary for high basal promoter activity in myoblasts as well as promoter structures responsible for gene activation during myotube formation could be identified. Putative protein binding sites potentially involved in promoter regulation (Fig. 3A) were either obtained from previously published studies or predicted by in silico analysis with Genomatix MatInspector software (Genomatix) (9), as exemplified in detail for the Cyt c promoter in Fig. S2 in the supplemental material.

FIG. 3.

Activities of the Cyt c, TFAM, and COXIV promoters analyzed with a dual luciferase reporter gene assay. (A) Promoter activation during differentiation is given as the ratio between the normalized firefly/Renilla luciferase values measured from myoblasts and myotubes. Only a few transcription factor binding sites considered important for promoter activity in myoblasts and myotubes are shown (see detailed scheme in Fig. S2 in the supplemental material). The recently described new consensus sequence element found in approximately 50% of NEM genes (22) is marked by a white rhombus. (B) Wild-type (Cyt c-111wt) and mutated (Cyt c-111mut) Cyt c promoter activity measured in myoblasts and myotubes. The sequence of the CRE element is illustrated by bold letters. Results are expressed as means ± standard deviations (n = 12).

In the Cyt c promoter, a truncation of about 400 bp of the upstream sequence neither impaired basal activity in myoblasts nor reduced activation during differentiation (Fig. 3A, upper panel). The deleted promoter fragment contains three SP-1 sites as well as a putative E box recognized by transcription factors of the MyoD family responsible for muscle-specific gene expression. Deletion of the sequence from bp −210 to −145 strongly decreased promoter activity in myoblasts, emphasizing the importance of the NRF-1 binding site for basal activity (16). However, this construct, starting at bp −145, still showed an 11-fold activation in myotubes, indicating that a promoter element other than the NRF-1 binding site is responsible for activation during differentiation. Removal of a further 34 bp, containing a tandem CAAT box motif, increased basal promoter activity dramatically but did not markedly alter the activation factor. Deletion of the promoter sequence down to bp −68, i.e., deletion of the downstream CRE site as well as a new sequence element common to many NEM genes recently described by us (22), completely abolished basal promoter activity. As the remaining sequence (bp −68 to +135) still contains downstream SP-1 sites, these data clearly show that SP-1 is not directly involved in the upregulation of Cyt c expression during muscle differentiation.

A deletion of the central region of the TFAM promoter from bp −255 to −117 did not impair promoter activity in myoblasts or the activation factor (Fig. 3A, middle panel). A truncation of the first 80 bp containing the new sequence element decreased promoter activation to 50%, which was not further affected by a deletion of the region from bp −332 to −117, containing two upstream CREB sites. The resulting promoter still contains four SP-1 and two NRF binding sites previously shown to be functional (65). No E-box-like motives were found in the TFAM promoter.

Basal COXIV promoter activity in myoblasts remained rather unchanged after being reduced to the core promoter sequence between bp −126 and +31 (Fig. 3A, lower panel), emphasizing the importance of this region, localized close to the transcription start site. However, the simultaneous deletion of the two putative upstream CRE sites led to a drop of the activation factor from 10- to 3-fold. The remaining promoter sequence still contains a pair of putative E boxes, the new sequence element (22), and two NRF-2 binding sites.

The proximal Cyt c promoter region (from bp −111 to −68) necessary and sufficient for activation during myogenesis contains a CRE site, previously shown to be functional in fibroblasts upon cAMP stimulation (24). It also contains the new sequence element involved in Cyt c promoter regulation, as the activation factor decreased twofold when it was mutagenized (22). In order to determine the function of the CRE site during differentiation, a construct bearing a 3-nucleotide mutation (Cyt c-111mut) destroying the binding motif for CREB-1 family proteins (Fig. 3B) was transfected. Mutation of the core binding sequence strongly reduced promoter activity levels in myoblasts and completely abolished its activation in myotubes.

CREB binds to the downstream CRE sequence of the Cyt c promoter in vitro.

To further elucidate the putative regulatory function of CRE sites for NEM gene expression, their interaction with DNA-binding proteins was studied first by EMSA (Fig. 4A). Nuclear extracts from myoblasts and myotubes were used to visualize protein binding to 25 bp containing the downstream CRE of the Cyt c promoter. Specific protein-DNA complexes of similar sizes were found, showing that the proteins interacting with this sequence are most likely identical. In order to verify that these protein-DNA complexes indeed contain CREB proteins, further retardation was confirmed either with anti-ATF-1/CREB or with anti-phospho-CREB antibodies (Fig. 4B and C). When a probe containing the same mutation of the CRE element as that used in the luciferase assay (Fig. 3B) was employed, neither a specific shift nor a supershift was observed (Fig. 4B and C). In order to quantify the DNA-binding capacities of CREB proteins present in nuclear extracts of myoblasts versus those for myotubes, a quantitative assay was performed. While the amounts of total CREB protein able to bind to a consensus CRE motif were similar in myoblasts and myotubes, the amount of phosphorylated CREB able to bind was twofold larger in myotubes than in myoblasts (see Fig. S3 in the supplemental material).

FIG. 4.

CREB binds the Cyt c promoter in vitro. (A) Protein-DNA complex formation (arrow) using nuclear extracts of myoblasts or myotubes and the Cyt c promoter sequence from bp −123 to −90 containing the CRE site. Sequence specificity of the complexes was shown by competition assays with increasing amounts (n-fold competition) of unlabeled DNA probe. A strong double band was considered unspecific, since very large amounts of unlabeled probe were needed for successful competition. (B and C) The nuclear protein-Cyt c promoter complex (shift, shown by white-head arrows) was incubated alone (−) or in combination with anti-ATF-1/CREB antibody (B) or anti-PCREB antibody (C) (+). The slower-migrating antibody-protein-DNA complex (supershift) is shown by black-head arrows.

Altered expression and phosphorylation status of ATF-1 and CREB-1 isoforms.

Western blot analysis of nuclear extracts detected three CREB-1 family proteins, which were identified as two CREB-1 isoforms and ATF-1, according to the sizes and immunoreactivities of the antibodies as indicated by the manufacturers (Fig. 5A, upper panels). All three proteins are phosphorylated, as shown by an antibody directed against the P-Ser133 site (Fig. 5A, lower panels) as well as by treatment of the membrane with phosphatase (Fig. 5B). During differentiation, total levels of CREB-1 family proteins decreased, and the ratio between the two isoforms as well as the ratio between the phosphorylated isoforms changed (Fig. 5A). A panel of antibodies (Fig. 5C) was used to identify the two CREB-1 isoforms as the splice variants CREB-1α (slower migrating) and CREB-1Δ (faster-migrating band) (Fig. 5D).

FIG. 5.

CREB-1α, CREB-1Δ, and ATF-1 protein expression during differentiation. Nuclear proteins were analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting. (A) Three independent samples for each group were loaded, and membranes were probed with an antibody recognizing all CREB-1 family members (in the first two panels, much longer exposure times were necessary to detect ATF-1) or with an antibody reacting only with phospho133-CREB proteins (fourth panel). Two different CREB-1-reactive bands with similar sizes (approximately 42 kDa) were detected by both antibodies. The membranes were reprobed with an anti-histone H1 antibody (third and fifth panels) as a loading control. (B) Nuclear proteins from confluent myoblasts were subjected to Western blotting and treated with λ-phosphatase where indicated (+). Anti-CREB and anti-phospho133-CREB antibodies (Cell Signaling) were used to detect CREB-1 family proteins. (C) CREB-1-specific antibodies were used in nuclear extracts prepared from two independent samples of confluent cells to identify the two CREB-1-reactive bands at 42 kDa. The antibody from Cell Signaling recognizes all CREB-1 family members, while the Upstate antibody specific for CREB-1 isoforms was used to exclude CREM. An antibody from Santa Cruz specific for the exon 5 domain recognized the slow-migrating band only, indicating that the two observed isoforms are CREB-1α and CREB-1Δ. (D) Exon organization of the CREB-1α and -Δ isoforms.

In order to further confirm the presence of CREB-1α and CREB-1Δ, RNA was analyzed by RT-PCR using a primer pair which amplifies both isoforms. PCR products of the expected sizes were detected (Fig. 6A and B). CREB-1Δ mRNA was the predominant form in all cells; however, no changes in the CREB-1α versus CREB-1Δ mRNA ratio occurred during differentiation (Fig. 6B). As it was not possible to detect the level of CREB-1Δ mRNA alone, we quantitated both isoforms together plus CREB-1α mRNA alone by quantitative RT-PCR. This showed that the levels of CREB-1α mRNA as well as CREB-1α and -Δ together remained unchanged (Fig. 6C), so CREB-1Δ mRNA remained also constant.

FIG. 6.

Levels of CREB-1α and CREB-1Δ transcripts analyzed by RT-PCR and real-time RT-PCR. (A) RNA was prepared from myoblasts (B), confluent cells (C), and myotubes (T), and cDNA was synthesized and used as a template for RT-PCR. PCRs were performed from 22 to 36 cycles to determine the optimal cycle number for densitometric evaluation. (B) PCRs from three independent samples for each group were performed for 27 cycles to compare the ratios between CREB-1α and -Δ products. (C) Real-time RT-PCR was performed for the above-mentioned samples, and the levels of the CREB-1α and -Δ isoforms together (black bars) or the CREB-1α isoform alone (white bars) were determined. Data were normalized to endogenous control GAPDH levels and shown as ΔC_T values. Results are expressed as means ± standard deviations (n = 3). * indicates a significant difference (P < 0.05) between myoblasts and confluent cells as determined by Student's t test.

In conclusion, we found an isoform switch from CREB-1α < CREB-1Δ in myoblasts toward CREB-1α > CREB-1Δ in myotubes during differentiation, while ATF-1 levels strongly decreased. Similar changes were observed with the phosphorylated forms of these proteins, with a pronounced shift toward phosphorylated CREB-1α, indicating that this transcription factor might be important for upregulation of the Cyt c promoter.

Modulation of endogenous CREB levels and function.

In order to further test this hypothesis, CREB-1 protein was reduced by a specific small interfering RNA (siRNA) but was also reduced by control siRNA (see Fig. S4A in the supplemental material). Even under optimized conditions of siRNA concentration, both specific and control siRNAs decreased CREB-1 protein levels to 25% (data not shown). Since at the same time luciferase activity driven by the Cyt c promoter was reduced (see Fig. S4B in the supplemental material), a clear correlation between CREB-1 levels and Cyt c promoter activity was found. However, since CREB-1 expression seems to be extremely sensitive to siRNA transfection, the observed effect does not convincingly demonstrate a causal relation.

In order to show that enhanced activation of CREB is sufficient to stimulate the Cyt c promoter, plasmids encoding a constitutively active (C2/CREB) as well as a dominant-negative (A-CREB) CREB construct (62) were cotransfected with reporter plasmids (Fig. 7). C2/CREB, a fusion protein of the CREB-2 activation domain and the CREB-1 leucine zipper DNA-binding domain, active independently of phosphorylation, activated the Cyt c promoter in myoblasts and, although less, in confluent cells (Fig. 7A). It also strongly activated the α-inhibin promoter, used as a positive control, which has been well-described to contain functional CRE elements (Fig. 7B). The activation of both promoters was diminished, both in myoblasts and in confluent cells, when the dominant-negative A-CREB construct was simultaneously expressed (Fig. 7A and B). In conclusion, overexpression of a constitutively active CREB protein is sufficient to activate the Cyt c promoter in these cells.

FIG. 7.

Cyt c and α-inhibin promoters are regulated by constitutively active and dominant-negative CREB constructs. Luciferase reporter gene assays were performed to study Cyt c (A) and α-inhibin (B) promoter activities in myoblasts or in confluent cells. The C2/CREB constitutively active CREB construct, pCMV empty vector, and CREB dominant-negative construct (A-CREB) plasmids were cotransfected with the reporter plasmids. Results are expressed as means ± standard deviations (n = 6). # indicates a significant difference at P values of <0.05, and ### indicates a significant difference at P values of <0.001, between promoter values in the presence or absence of C2/CREB obtained by Student's t test. *** indicates a significant difference (P < 0.001) between promoter values for cotransfection with C2/CREB in the absence or in the presence of the A-CREB plasmid as determined by Student's t test.

CREB-1α binds to the Cyt c promoter in situ.

Finally, ChIP showed that a member of the CREB-1 family binds to the Cyt c promoter sequence between bp −262 and −22 in situ (Fig. 8A) and that this protein is recognized by a P-CREB antibody (Fig. 8B). The amount of total CREB as well as that of P-CREB bound to this region remains unchanged during differentiation (Fig. 8D). Using an antibody specific for CREB-1α, ChIP shows that indeed in situ, this region is occupied by CREB-1α in confluent cells and myotubes but not in myoblasts (Fig. 8C and D). In conclusion, recruitment of CREB-1α to this CRE element is probably responsible for the upregulation of the Cyt c promoter during muscle differentiation.

FIG. 8.

CREB, P133-CREB, and CREB-1α interact in vivo with the Cyt c promoter. ChIP assays were carried out from myoblasts, confluent cells, and myotubes. DNA-bound protein complexes were immunoprecipitated by anti-CREB (A), anti-phospho133-CREB (B), or anti-CREB-1α (C) antibodies (Ab). A shortened sequence (bp −262 to −22) of the mouse Cyt c promoter was amplified by PCR. (D) Densitometric evaluation of PCR products as shown in panels A to C after ChIP. Results were obtained from two ChIP experiments and three PCRs and are expressed as means ± standard errors.

DISCUSSION

Muscle differentiation is accompanied by a pronounced rise in mitochondrial content during embryonic development in vivo (43). Whether this is an adaptation to the enhanced energy consumption or rather an “anticipation” of the higher energy demand of the contracting muscle, and how it is coordinated with muscle-specific gene expression, is unclear. Differentiation of C2F3 cells to myotubes is accompanied by a dramatic increase in total energy demand when ATP turnover is calculated per nucleus as the most useful denominator (Table 2). We postulate that this cannot be covered sufficiently by anaerobic glycolysis alone and consequently requires an increase in mitochondrial biogenesis.

The increase of mitochondrial mass is mainly achieved by the upregulation of NEM gene expression; however, levels of these mRNAs rise together with total RNA (Fig. 2B), i.e., cytosolic rRNAs and tRNAs. Such a coordinated upregulation of NEM gene expression with total tissue RNA is a consistent observation also in vivo under conditions of induced mitochondrial biogenesis, such as thyroid hormone treatment (71), cold adaptation in brown fat (32), hypertrophy of the heart (70), and endurance training of muscle (57). On the other hand, mRNAs derived from mtDNA are almost consistently elevated beyond the total RNA pool in these situations, while mtDNA levels do not necessarily change; both observations were confirmed by our findings here (Fig. 2C and D). Thus, an increase of mtDNA copy number is not necessary to increase mitochondrial mass, since mtDNA is present in excess in most cells (68, 69). Taking into account the 8-fold rise in total RNA, COXIII mRNA levels are found to increase 25-fold when normalized to numbers of nuclei and about 40-fold when normalized to mtDNA levels. Thus, we can also conclude that mtDNA-encoded proteins are probably produced in excess and that assembly of functioning organelles is instead controlled by nuclear gene products.

As TFAM is increased fourfold in myotubes (Fig. 1B and C), while mtDNA levels decrease (Fig. 2C), it may be the resulting higher ratio of TFAM to mtDNA that triggers this massive stimulation of mtDNA transcription. There is considerable debate as to whether TFAM is indeed a transcription factor in situ or rather an mtDNA packaging protein, because three groups have reported a stoichiometry between TFAM and mtDNA of about 50:1 (14, 18, 41), while two other groups found enough molecules to cover the whole mitochondrial genome (15, 60). This question is not yet solved; however, import into mitochondria (20) or overexpression in cells of TFAM increases mtDNA transcription but not mtDNA copy number (41), indicating an active role for TFAM in regulation of transcription but not necessarily in mtDNA replication.

The three genes studied here were chosen as representative members encoding different functional groups of mitochondrial proteins: Cyt c was shown to be a good indicator of mitochondrial content in various tissues (72), and extensive data on promoter regulation are available for several cell culture models (27, 53, 73). COXIV is thought to have an important function as a nucleus for COX assembly, and it is the only nucleus-encoded subunit present in all mitochondria from organisms as diverse as yeast and animals. Furthermore, the COXIV gene was shown to be regulated differently from Cyt c in several cell culture models; thus, they seem to be targets for alternative pathways (24). Finally, TFAM is an important component involved in the coordination of nuclear and mitochondrial gene expression. The main goal of our study was to explore and identify common regulatory mechanisms controlling NEM gene expression in our model of mitochondrial biogenesis. Therefore, we first analyzed similarities and common motifs shared by the promoters of Cyt c, COXIV, and TFAM. The three promoter sequences contain binding sites for either NRF-1 (Cyt c), NRF-2 (COXIV), or both (TFAM). Nuclear respiratory factors regulate the expression of many genes encoding respiratory chain subunits and some proteins involved in mtDNA maintenance (TFAM, TFBMs, and MRP-RNA) or heme synthesis (ALAs). However, they do not control genes encoding proteins involved in fatty acid metabolism, which are instead regulated by peroxisome proliferator-activated receptors (30). Another important transcriptional regulator of many NEM genes is CREB-1 and its related protein family members. Two functional CRE sites were already described to occur in the Cyt c promoter (24), while CRE consensus sequences in the promoters of TFAM and COXIV have not been characterized in functional studies so far.

When analyzing the potential role of transcriptional modulators in the regulation of NEM gene expression during myoblast differentiation, we distinguished between two functions: first, some of these factors may be necessary for “basal” expression of these genes in myoblasts, and second, the same or other regulators may be responsible for induction of gene expression during differentiation. Our results show that the Cyt c promoter sequence lacking the NRF-1 binding site (CytP-145) has severely reduced basal promoter activity but is still sufficient for full activation (Fig. 3A). On the other hand, activation of the shortest TFAM promoter sequence is severely impaired, although it still contains NRF-1 and NRF-2 sites. These findings indicate that NRF-1 and NRF-2 clearly play a significant role in basal transcription of NEM genes in C2F3 myoblasts but are obviously not responsible for the induction of mitochondrial biogenesis during muscle differentiation.

Expression of genes specific for muscle is mainly achieved by transcription factors belonging to the myoD family, which bind to DNA motifs called E boxes (47). In skeletal muscle cells, CREB-1 is associated with MyoD and targeted to the retinoblastoma gene promoter to enhance its transcription, and the elevated level of retinoblastoma protein is essential for myoblast cell cycle arrest and therefore for terminal differentiation and for survival of myocytes (40). Interestingly, overexpression of myogenin in the muscles of transgenic mice was shown to stimulate mitochondrial proliferation (29). However, although both Cyt c and COXIV promoters contain putative E boxes, no influence of these motifs on gene activation during myogenesis could be detected, arguing against a direct interaction of myogenic factors with these promoters.

A common regulatory element found in the promoters of several NEM genes is the CRE, which binds CREB-1 and its closely related family members ATF-1 and CREM. CREB-1 was previously shown to be a crucial transcription factor in a retrograde communication pathway signaling mitochondrial malfunction to the nucleus (4, 42). CREB-1 family proteins are activated by phosphorylation upon various stimuli, such as an elevation of intracellular cAMP, an increase in cytosolic calcium concentration (25), or a stimulation with growth factors (48). Phosphorylation (7, 21) enhances its interactions with CREB binding protein (CBP/p300) and proteins belonging to the basal transcription machinery, thus stimulating transcription (for an overview, see reference 59). Various splice variants of CREB-1 and CREM with different properties regarding promoter activation make the regulation of genes by CREB-1 proteins even more complex. The three different promoters analyzed here contain several putative CRE sites. In fibroblasts, the activation of the Cyt c promoter by either serum stimulation or an increase in cAMP levels requires both CRE sites (24). However, the COXIV promoter, although containing two putative CRE sites, is not activated under these conditions (27). Furthermore, electrical stimulation of neonatal cardiomyocytes led to Cyt c promoter activation, which was partially mediated by these CRE sites (73). Here, we show that the basal expression of Cyt c in myoblasts, and moreover, during myogenesis, is strongly dependent on regulation by CREB-1, but the context of the CRE site and potential interactions with other regulatory factors are crucial. While the importance of the CRE site at bp −108 was clearly demonstrated, the deletion of a second upstream CRE site has no effect on Cyt c promoter activation (Fig. 3A). For the COXIV promoter, a clear influence of the upstream CRE sites on myogenic gene activation was also found; however, deletion of these sites did not completely abolish the induction of the promoter, while the upstream CRE sites in the TFAM promoter seem to be not important (Fig. 3A).

Proteins binding to CRE elements were detected by EMSA and supershift experiments, and binding assays confirmed that CREB and P-CREB are present in these DNA-protein complexes (Fig. 4; also see Fig. S3 in the supplemental material). While total CREB-1 protein levels decrease in nuclear extracts during differentiation, the ratio of CREB-1α to CREB-1Δ as well as that of P-CREB-1α to P-CREB-1Δ increases, while total ATF-1 and P-ATF-1 decline (Fig. 5A). Since the major mRNA species in all cells was the mRNA for CREB-1Δ (Fig. 6), posttranscriptional processes have to be responsible for recruitment of CREB-1α to the Cyt c promoter in myotubes. Although the presence of the CREB-1α and -Δ isoforms had been discovered already in 1990, their individual function remains obscure. Yamamoto et al. showed that in rat brain the predominant isoform on mRNA as well as protein level is CREB-1Δ. They found that in vitro, the capacity for transactivation of the rat CREB-1α protein is approximately 10 times higher than that for CREB-1Δ (74). However, other in vitro studies indicated that, in mouse and human, CREB-1α and -Δ transcriptional activities are equal, and the different activities in the three species were explained by minor amino acid variations in the α domain (50). In all studies, CREB-1Δ mRNA was found to be the most abundant isoform (6, 50), in accordance with our results in C2F3 muscle cells (Fig. 6). However, we found a marked shift not only in the CREB-1α/CREB-1Δ protein ratio but even more pronounced in the ratio between the phosphorylated α and Δ isoforms: While the predominant phosphorylated isoform in myoblasts was P-CREB-1Δ, myotubes switched to the P-CREB-1α isoform (Fig. 5). These results were supplemented by ChIP experiments, which showed that indeed it is mainly the phosphorylated form of CREB-1α binding to the Cyt c promoter in situ in myotubes (Fig. 8). The presence of distinct CREB-1 isoforms and the differential recruitment to the CRE-element suggest that this isoform switch has an important functional role during muscle differentiation. We also show here that the regulation of CREB-1 function takes place predominantly at the posttranslational level, by phosphorylation (Fig. 5) and recruitment into transcription complexes (Fig. 8), as others have suggested (28, 52).

CREB-1 was shown to be phosphorylated at Ser-133 by many kinases in addition to protein kinase A (PKA), such as PKC, Akt, Msk, Rsk, p38, calmodulin-dependent protein kinase II, and calmodulin-dependent protein kinase IV, and dephosphorylated by multiple phosphatases, for example calcineurin (PP2B), PP2A, and PP1. Several of them are known to be activated by Ca²⁺ (59). In muscle cells, elevation of intracellular Ca²⁺ triggers contraction and might be a signal for these adaptational changes. While a long-term increase in cytosolic Ca²⁺ moderately induces the Cyt c promoter via PKC in differentiated muscle cell lines (19), forced contraction does so too. However, the invoked Ca²⁺ transients alone were not sufficient and promoter induction by SP-1 required contractile activity and probably increased ATP turnover (13). Whether an increased Ca²⁺ level plays a major role in activation of NEM genes via CREB-1 during myogenesis and which kinases or phosphatases are responsible for activating CREB-1 directly remain to be determined.

In conclusion, myotube formation is accompanied by a massive stimulation of ATP turnover, which is covered by enhanced mitochondrial biogenesis. This is achieved by a coordinated increase of transcripts encoded by NEM genes together with total cellular RNA. The activation of NEM gene expression, especially of the Cyt c promoter, depends on CRE elements, and the recruitment of P-CREB-1α protein seems to play the key role in this model of mitochondrial biogenesis.