OUP user menu

Mitochondrial transcription factors TFAM and TFB2M regulate Serca2 gene transcription

Atai Watanabe, Masashi Arai, Norimichi Koitabashi, Kazuo Niwano, Yoshiaki Ohyama, Yoshihumi Yamada, Noriko Kato, Masahiko Kurabayashi
DOI: http://dx.doi.org/10.1093/cvr/cvq374 57-67 First published online: 27 November 2010

Abstract

Aims Sarco(endo)plasmic reticulum Ca2+-ATPase 2a (SERCA2a) transports Ca2+ by consuming ATP produced by mitochondrial respiratory chain enzymes. Messenger RNA (mRNA) for these enzymes is transcribed by mitochondrial transcription factors A (TFAM) and B2 (TFB2M). This study examined whether TFAM and TFB2M coordinately regulate the transcription of the Serca2 gene and mitochondrial genes.

Methods and results Nuclear localization of TFAM and TFB2M was demonstrated by immunostaining in rat neonatal cardiac myocytes. Chromatin immunoprecipitation assay and fluorescence correlation spectroscopy revealed that TFAM and TFB2M bind to the −122 to −114 nt and −122 to −117 nt regions of the rat Serca2 gene promoter, respectively. Mutation of these sites resulted in decreased Serca2 gene transcription. In a rat myocardial infarction model, Serca2a mRNA levels significantly correlated with those of Tfam (r = 0.54, P < 0.001) and Tfb2m (r = 0.73, P < 0.001). Overexpression of TFAM and TFB2M blocked hydrogen peroxide- and norepinephrine-induced decreases in Serca2a mRNA levels. In addition, overexpression of TFAM and TFB2M increased the mitochondrial DNA (mtDNA) copy number and mRNA levels of mitochondrial enzymes.

Conclusion Although TFAM and TFB2M are recognized as mtDNA-specific transcription factors, they also regulate transcription of nuclear DNA, i.e. the Serca2 gene. Our findings suggest a novel paradigm in which the transcription of genes for mitochondrial enzymes that produce ATP and the gene for SERCA2a that consumes ATP is coordinately regulated by the same transcription factors. This mechanism may contribute to maintaining proper cardiac function.

  • TFAM
  • TFB2M
  • SERCA2a
  • Mitochondria
  • Transcription

1. Introduction

Myocardial force generation depends on the cellular energy supply of ATP, which is mainly controlled by mitochondria. ATP produced by mitochondrial F1/F0-ATPase is delivered to ATP-consuming sites such as myosin-ATPase, sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA2a), and Na+/K+-ATPase. Efficient matching of energy supply and expenditure is essential for beat-by-beat adaptation to changes in cardiac workload and involves key molecules, including ADP and Ca2+.1 In addition to such fine-tuning of cardiac function by these small molecules, energy production and expenditure may also be coordinately regulated by gene transcription to shift the heart to another steady-state level.

Genes for mitochondrial electron transport enzymes and ATP synthase are encoded by both mitochondrial and nuclear genomes. Mitochondrial DNA (mtDNA)-encoded genes are transcribed by mtDNA-specific transcriptional machinery composed of mtDNA-specific transcription factors, mitochondrial transcription factors A (TFAM) and B1 (TFB1M) or B2 (TFB2M), and a mitochondrial RNA polymerase.2,3 Although TFB1M and TFB2M bind to mitochondrial RNA polymerase with a 1:1 stoichiometry, TFB2M has 10 times higher activity in facilitating RNA polymerase to initiate transcription in mammalian mtDNA.2 Moreover, mtDNA replication and transcription are controlled by TFAM. Accordingly, TFAM and TFB2M are key regulators of mitochondrial ATP production. Decreases in the amount of TFAM4 and ATP5 have been reported in the failing heart. Among ATP-consuming proteins, namely ATPases, SERCA2a requires the most free energy to maintain its ATPase function, suggesting that its function can be easily jeopardized by a small decrease in ATP production.6 In fact, decreased Serca2a messenger RNA (mRNA) levels and protein function have been noted in heart failure,7 and SERCA2a gene therapy is currently a promising therapy for advanced heart failure.8

SERCA2a function is highly dependent on the ATP supply. In this study, we tested the hypothesis that two well-known mtDNA-specific transcription factors, TFAM and TFB2M, regulate Serca2 gene transcription to clarify the common control mechanism of gene transcription for energy production and expenditure proteins. To determine the relevance of such mechanism in the stressed heart, we examined the contribution of TFAM and TFB2M to transcriptional activation of the Serca2 gene in the stressed heart.

2. Methods

2.1 Neonatal rat cardiac myocyte culture

Primary neonatal rat cardiac ventricular myocyte cultures were prepared as described previously, with minor modifications.9 Cardiac myocytes were seeded in 35, 60, or 100 mm gelatin-coated culture dishes (1 × 106, 2.5 × 106, and 5 × 106 cells, respectively) in Dulbecco's modified Eagle's medium containing 10% foetal bovine serum. For stress induction in cardiac myocytes, norepinephrine (NE) (40 μmol/L) or hydrogen peroxide (40 μmol/L) was added 4 h after transfection of the expression vector, small interfering RNA (siRNA), or luciferase vector.

2.2 Immunohistochemistry

Immunohistochemical detection was performed using rabbit polyclonal TFAM antibody (kindly provided Dr H. Inagaki, National Institute of Advanced Industrial Science and Technology, Ibaragi, Japan) followed by incubation with Alexa Fluor 488-labelled anti-rabbit immunoglobulin G (IgG) (Invitrogen Co., Carlsbad, CA, USA) for TFAM protein or using goat polyclonal TFB2M antibody (Abcam Plc., Cambridge, UK) followed by Alexa Fluor 647-labelled anti-goat IgG (Invitrogen Co.) for TFB2M protein. Mitochondria were stained by Mitotracker Red CMXRos (Invitrogen Co.) before fixing the cells in Figure 1A. Confocal microscopic images were obtained using LSM 510 META (Carl Zeiss AG, Oberkochen, Germany) (Figure 1A). In Figure 1B, mouse monoclonal cytochrome c oxidase subunit IV (COX IV) and Alexa Fluor 594-labelled anti-chicken IgG (Invitrogen Co.) were used for the detection of mitochondria. A dimeric cyanine nucleic acid dye, BOBO-3 (Invitrogen Co.), was used to stain nuclear DNA. In Figures 1B and 4A and B, goat polyclonal TFB2M antibody (Abcam Plc.) was used followed by incubation with Alexa Fluor 488-labelled anti-goat IgG (Invitrogen Co.).

Figure 1

Localization of TFAM and TFB2M in the nucleus and their effects on the Serca2 gene transcription. (A) Representative confocal microscopic images of TFAM and TFB2M proteins in the neonatal rat cardiac myocytes (×63). Mitotracker indicates location of mitochondria. (B) Confocal microscopic images showing co-localization of TFAM and nuclei, TFAM and mitochondria, TFB2M and nuclei, and TFB2M and mitochondria. Specific antibodies for TFAM, TFB2M, cytochrome c oxidase (mitochondrial marker), and BOBO-3 (cyanine nucleic acid marker) were used. Arrowheads indicate localization of TFAM and TFB2M in the nuclei. Relative levels of luciferase activity after Tfam and Tfb2m expression vectors (C) or scramble siRNA (D) transfection compared with the control empty vector. *P < 0.05 vs. the control group. Values represent the means ± standard deviation of five independent experiments. (E) Transcription levels of serial deletion constructs of the 5′-regulatory region of the rat Serca2 gene were measured in the presence of Tfam and Tfb2m expression vectors or the control vector. The luciferase activity of the control vector-transfected −1829 to +123 nt construct of the Serca2 gene was designated as 1. *P < 0.05 vs. the control in each deletion construct. Values represent the means ± standard deviation of five independent experiments.

2.3 Gene transfer and luciferase assay

Rat Tfam and Tfb2m expression vectors were produced by reverse transcriptase (RT)–polymerase chain reaction (PCR) using pcDNA 3.1 His vector (Invitrogen Co.). Each expression vector (1 μg) was introduced into neonatal rat cardiac myocytes using FuGENE 6 transfection reagent (F. Hoffmann-La Roche, Ltd, Basel, Switzerland), according to the manufacturer's instruction. Pre-designed siRNA duplexes composed of 21 nucleotides were utilized with the Hemaglutinating virus of Japan envelope vector (Ishihara Sangyo Kaisha, Inc., Osaka, Japan) to knockdown designated RNAs. Nucleotide sequences of probes and siRNAs used in this study are summarized in Supplementary material online, Table S1.

For luciferase assay, serial deletion constructs of the 5′-upstream region of the rat Serca2 gene (nucleotides −1829, −1351, −850, −479, −327, −171, and −88 to +123 bp relative to the transcription initiation site) were used to determine the responsive site of TFAM and TFB2M.

2.4 Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assay was performed using rabbit polyclonal anti-TFAM, goat polyclonal anti-TFB2M, anti-acetylated histone H3 in the 9th and 14th lysine residues (Millipore, Billerica, MA, USA), anti-methylated histone H3 in the 4th lysine residue (Abcam Plc.), and anti-histone H3 (Abcam Plc.) antibodies according to the ChIP Assay Kit (Millipore) protocol. Protein/DNA complexes were eluted, and precipitated DNA was purified using a spin column for PCR. PCR primers were designed for the −479 to −1 region of the rat Serca2 5′-regulatory region. Information regarding forward and reverse primer sequences is denoted in Supplementary material online, Table S1.

2.5 Fluorescence correlation spectroscopy

Fluorescence correlation spectroscopy was performed using the MF20 molecular interaction analytical system (Olympus, Tokyo, Japan).10 Recombinant rat TFAM and TFB2M proteins were produced using the pGEX-6P-1 vector (GE Healthcare, Buckinghamshire, UK) or pCold TF DNA vector (Takara Bio, Inc., Shiga, Japan) in Escherichia coli according to manufacturers’ protocols. For the rat Serca2 5′-regulatory region, 30 bp double-stranded probes were produced as follows. 5′-Terminal carboxytetramethyl rhodamine (TAMRA)-labelled sense strand and unlabelled antisense strand oligonucleotides were synthesized by Greiner Bio-One (Frickenhausen, Germany). Double-stranded oligonucleotides were prepared by annealing each oligonucleotide.

Recombinant TFAM or TFB2M protein was incubated with TAMRA-labelled double-stranded DNA, and measurements of translational diffusion time were repeated five times per sample. Elongation of translational diffusion time indicates the direct binding of TFAM or TFB2M protein with the Serca2 gene probe. For the determination of the TFAM and TFB2M binding site in the 5′-regulatory region of the Serca2 gene, mutations were introduced into the TAMRA-labelled probe that encodes from −140 to −111 nt in this region (Serca2 −140 wild), as shown in Figure 4A.

2.6 Rat myocardial infarction model

Experimental myocardial infarction was introduced into Wister rats by ligating the left anterior descending coronary artery, as described previously.11 Five months after coronary occlusion, haemodynamic parameters were measured using a pressure–volume (PV) catheter. Rats were anaesthetized, and a microtip PV catheter (SPR-838, Millar Instruments, Houston, TX, USA) was introduced into the left ventricle through the right carotid artery. Total RNA was isolated using the Total RNA Extraction Kit (Promega Corp., Madison, WI, USA), and protein was obtained by homogenizing tissues with ice-cold radioimmunoprecipitation (RIPA) buffer. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). All experimental protocols were approved by the Animal Care Committee of the Gunma University Graduate School of Medicine.

2.7 Quantitative reverse transcriptase–polymerase chain reaction and quantitative polymerase chain reaction

mRNA levels were also determined by two-step quantitative RT–PCR using Brilliant II QPCR master mix (Agilent Technologies, Santa Clara, CA, USA). Taqman probes for rat genes, Serca2, cytochrome b, NADH dehydrogenase 5 (Nd5), flavoprotein, brain natriuretic peptide (Bnp), and β-actin, were designed and synthesized by Biosearch Technologies, Inc. (Novato, CA, USA). Nucleotide sequences of primers used in this study are summarized in Supplementary material online, Table S1. For the measurement of mtDNA copy number, probes for the D-loop region of mtDNA were synthesized. mtDNA was extracted using the AllPrep DNA/RNA Mini Kit (Qiagen Gmbh, Hilden, Germany), and the copy number of the D-loop region was measured by quantitative PCR.

2.8 Statistical analysis

Results are expressed as means ± standard deviation. Overall differences within groups were determined by one-way analysis of variance, and the difference between the control group and each experimental group was determined by Bonferroni's test. Correlations between the relative Serca2a, cytochrome b or mitochondrial ATPase 6/8 mRNA levels, and Tfam or Tfb2m mRNA levels (Figure 5 and Supplementary material online, Figure S5) were determined by linear regression analysis. A level of P < 0.05 was defined as statistically significant.

3. Results

3.1 TFAM and TFB2M proteins localize to the nucleus and promote Serca2 gene transcription

To determine whether TFAM and TFB2M proteins can bind to the nuclear genome, immunohistochemical analysis was performed using specific antibodies for TFAM and TFB2M. Mitochondria were specifically stained using Mitotracker Red (Figure 1A) and antibodies for cytochrome c oxidase, a mitochondria-specific marker protein (Figure 1B). TFAM and TFB2M proteins localized not only to the mitochondria but also to the nucleus (Figure 1A and B). Western blot analysis also demonstrated localization of TFAM and TFB2M in the nucleus (see Supplementary material online, Figure S1A and B). TFAM and TFB2M protein levels decreased in both the nucleus and the mitochondria, with no preferential decrease in either location in cardiac myocytes exposed to oxidative stress (see Supplementary material online, Figure S1A and B).

To test whether Serca2 gene transcription is also regulated by TFAM and TFB2M, neonatal rat cardiac myocytes were transiently transfected with the luciferase reporter vector harbouring the rat Serca2 gene 5′-regulatory region (−1829 to +123 from the transcription start site) in the presence or absence of rat Tfam and Tfb2m expression vectors or Tfam and Tfb2m siRNAs (Figure 1C and D). The magnitude of the change in the Tfam and Tfb2m levels after overexpression and RNA interference (RNAi) of Tfam and Tfb2m is shown in Supplementary material online, Figure S2. Overexpression of TFAM and TFB2M significantly increased Serca2 gene transcription (1.76- and 2.02-fold higher vs. empty vector, respectively), whereas ablation of TFAM and TFB2M resulted in a significant reduction in its transcription (0.42- and 0.37-fold vs. scramble siRNA, respectively). Although TFB2M greatly enhances the transcription of mitochondrial genes through TFAM,2 TFAM and TFB2M did not have an additive or synergistic effect on Serca2 gene transcription.

3.2 TFAM and TFB2M bind to the 5′-regulatory region of the Serca2 gene

Serial deletion constructs of the 5′-regulatory region of the Serca2 gene with Tfam and Tfb2m expression vectors or the control vector were transfected into the neonatal rat cardiac myocytes. Figure 1E shows that the TFAM and TFB2M responsive sites were located within the −1829 to −1351 and −479 to −88 nt positions of the rat Serca2 gene 5′-regulatory region.

A ChIP assay was performed to determine whether TFAM and TFB2M bind to the Serca2 gene 5′-regulatory region in living cardiac myocytes. A primer pair for amplification of the sequence from −479 to −1 in the rat Serca2 5′-regulatory region was used.1214 As illustrated in Figure 2A and B, TFAM and TFB2M bound to this site. Acetylation and methylation of histone H3 were also present in this region (Figure 2C). Generally, acetylation of the lysine residue and methylation of the arginine and lysine residues in histones H3 and H4 indicate activated transcriptional machinery in the corresponding promoter region.1517 We also confirmed that TFAM and TFB2M did not bind to the −1351 to −172 region of the Serca2 gene in this system (data not shown), suggesting specific binding of TFAM and TFB2M to the −171 to −1 region. Taken together with the luciferase assay showing no responsive site in the −88 to −1 nt position, TFAM and TFB2M appear to bind within the −171 to −89 position of the rat Serca2 gene 5′-regulatory region. A ChIP assay also suggested binding of TFAM and TFB2M to the −1829 to −1352 nt position of the Serca2 gene (data not shown).

Figure 2

Estimation of TFAM and TFB2M responsive sites on the Serca2 gene. (A–C) Chromatin immunoprecipitation assay. Bands at the 500 bp region indicate the binding of TFAM, TFB2M, histone H3, acetylated histone H3 (9th and 14th lysine residues), and methylated histone H3 (4th lysine residue) in the −479 to −1 nt position of the rat Serca2 5′-regulatory region. Polymerase chain reaction products (10 μL) were electrophoresed on 2% gels. Pre-immune immunoglobulin G was used as the control for antibodies of these specific DNA-binding proteins. For input material, 0.01% of soluble chromatin was used. Changes in the translational diffusion time after binding 5 μmol/L recombinant TFAM (D) or TFB2M (E) protein to the terminal carboxytetramethyl rhodamine-labelled 30 bp double-stranded oligonucleotide corresponding to the designated sequence of the 5′-regulatory region of the Serca2 gene. Values represent the means ± standard deviation of five independent measurements. Statistical significance was determined by comparing the TFAM(+) group with the TFAM(−) group (D) and the TFB2M(+) group with the TFB2M(−) group (E) in each sequence.

3.3 Determination of TFAM and TFB2M binding sites in the 5′-regulatory region of the Serca2 gene

Fluorescence correlation spectroscopy was used to determine the TFAM and TFB2M binding site(s) in the 5′-regulatory region of the Serca2 gene. The translational diffusion time of the Serca2 probe in the presence of TFAM protein was significantly prolonged with oligonucleotides corresponding to the −146 to −102 nt region of the Serca2 gene with a peak at −140 to −111 nt [1.58-fold increase in translational diffusion time vs. the TFAM(−) sample, Figure 2D], confirming that the TFAM protein binds to this 5′-regulatory region. Similarly, in the presence of the TFB2M protein, the translational diffusion time was significantly prolonged with oligonucleotides corresponding to the −152 to −102 nt region; the greatest translational diffusion time was achieved with the −140 to −111 nt probe [1.98-fold vs. the TFB2M(−) sample, Figure 2E]. Using fluorescence correlation spectroscopy, we examined whether TFAM and TFB2M bind to the −1829 to −1351 region, which was suggested by the luciferase assay in Figure 1E. Although fluorescence correlation spectroscopy showed several binding sites in this region, the increase in the translational diffusion time in this region (see Supplementary material online, Figure S3A and B) was minimal compared with that in the −140 to −111 region. These data suggest that the binding of TFAM and TFB2M to the −1829 to −1351 region is weak and has a less important role than their binding to the −140 to −111 region for the transcriptional control of the Serca2 gene. Thus, we focused further study on the proximal region including the −140 to −111 nt position.

To further distinguish the location of the TFAM and TFB2M binding sites, mutations were introduced into the TAMRA-labelled probe that encodes the sequence from −140 to −111 nt of the 5′-regulatory region of the Serca2 gene (Serca2 −140 wild), as shown in Figure 3A. Introduction of a mutation in the −122 to −114 nt position abolished the binding activity of TFAM to the Serca2 gene, suggesting that the TFAM binding site exists in this GGGGGCGGG sequence (Figure 3B). Similarly, the mutation assay suggested that the TFB2M binding site is in the GGGGGC sequence, which overlaps with the TFAM binding sequence (Figure 3C).

Figure 3

TFAM (−122 to −114) and TFB2M (−122 to −117) binding sites in the 5′-regulatory region of the Serca2 gene. (A) Series of 3 nt mutations in the terminal carboxytetramethyl rhodamine-labelled 30 bp double-stranded oligonucleotide corresponding to the −140 to −111 nt region of the Serca2 gene (from ‘mutation −134' to ‘mutation −116'). Changes in the translational diffusion time after binding 5 μmol/L recombinant TFAM (B) or TFB2M (C) proteins to these terminal carboxytetramethyl rhodamine-labelled oligonucleotides. Values represent the means ± standard deviation of five independent measurements. (D and E) Effect of 3 nt mutations in the TFAM/TFB2M binding site (the −122 to −114 nt region) on Serca2 gene transcription. Three-nucleotide substitutions of guanine and cytosine residues with alanine residues are introduced into the TFAM/TFB2M binding site in the luciferase reporter constructs as depicted in (D).

To determine the role of the TFAM and TFB2M binding sequences in Serca2 gene transcription, luciferase reporter vectors that harboured 3 nt mutations in the GGGGGCGGG sequence were tested (Figure 3D). As shown in Figure 3E, mutations in this region significantly reduced transcription of the Serca2 gene, suggesting that the TFAM and TFB2M binding sites contribute to basal transcriptional activity of the Serca2 gene.

3.4 Effect of TFAM and TFB2M on mitochondrial DNA replication and transcription

We next examined the effect of overexpression and ablation of TFAM and TFB2M on mtDNA replication and mitochondrial mRNA levels. Figure 4A and B indicates the effects of overexpression and RNAi on the amount of TFAM and TFB2M in the nucleus and mitochondria, respectively. As shown in the respective figures, TFAM and TFB2M accumulated and decreased in the nucleus and mitochondria with no preferential change in either location. The mitochondrial copy number was estimated by the amount of PCR product corresponding to the sequence of the mtDNA D-loop region.18 As shown in Figure 4C, the mtDNA copy number was under the control of TFAM and TFB2M. Overexpression of TFAM and TFB2M increased the mtDNA copy number, whereas ablation of these transcription factors decreased its number. Similarly, overexpression of TFAM and TFB2M increased mRNA levels for cytochrome b (complex III), NADH dehydrogenase 5 (complex I), and flavoprotein (complex II), whereas ablation by their siRNAs significantly reduced mRNA levels of these mitochondrial proteins (Figure 4D–F). In addition, overexpression and ablation of TFAM, respectively, increased and significantly decreased the ATP content in cardiac myocytes (see Supplementary material online, Figure S4).

Figure 4

Effect of overexpression and ablation of TFAM and TFB2M on mtDNA copy number and mitochondrial mRNA levels. (A and B) Tfam and Tfb2m expression vectors and siRNAs at designated concentrations were transfected into neonatal rat cardiac myocytes. After 24 h transfection, TFAM (A) or TFB2M (B) proteins were immunohistochemically detected using rabbit polyclonal TFAM antibody followed by incubation with Alexa Fluor 488-labelled anti-rabbit or using goat polyclonal TFB2M antibody followed by anti-goat immunoglobulin G, respectively. (CF) All measurements were performed 24 h after transfection of plasmids or 48 h after treatment with siRNAs. Values represent the means ± standard deviation of five independent experiments. The mtDNA copy number and mitochondrial mRNA levels in the empty vector-transfected and scramble siRNA-transfected groups were designated as 1. The Tfam- and Tfb2m-transfected groups were compared with the empty vector-transfected group. The Tfam and Tfb2m siRNA-transfected groups were compared with the scramble siRNA-transfected group.

3.5 Tfam and Tfb2m mRNA levels are significantly decreased relative to the Serca2a mRNA levels in the rat myocardial infarction model

Changes in Tfam and Tfb2m mRNA levels were examined in the failing rat heart at 5 months after myocardial infarction (Figure 5A). Tfam and Tfb2m mRNA levels were inversely correlated with tau, an index of early diastolic dysfunction (Tfam, r = −0.49, P < 0.01, Figure 5B; Tfb2m, r = −0.51, P < 0.01, Figure 5C), and positively correlated with Emax, an index of systolic function (Tfam, r = 0.58, P < 0.001; Tfb2m, r = 0.46, P < 0.01; data not shown). These data suggest that mRNA levels of both factors significantly decreased as the severity of systolic and diastolic heart failure increased.

Figure 5

Serca2a mRNA level correlates with Tfam and Tfb2m mRNA levels in the failing rat heart at 5 months after myocardial infarction. Myocardial infarction and sham operation were introduced into 21 and 8 rats, respectively. (A) Representative northern blot of Tfam, Tfb2m, Serca2a, Bnp, cytochrome b, and mitochondrial ATPase 6/8 mRNAs and 28S rRNA in the sham-operated and heart failure after myocardial infarction groups. Correlation between tau and the relative Tfam (B) and Tfb2m (C) mRNA levels. Linear regression analysis was performed to detect possible correlations. Correlation between the mRNA levels of Tfam (D) or Tfb2m (E) and Serca2a.

mRNA levels of Tfam, Tfb2m, Serca2a, Bnp, cytochrome b, and mitochondrial ATPase 6/8 (complex V) were measured to determine whether Tfam and Tfb2m mRNA levels correlate with Serca2a and mitochondrial mRNA levels. As shown in Figure 5D and E, mRNA levels of Tfam and Tfb2m significantly correlated with Serca2a mRNA levels (Tfam, r = 0.54, P < 0.001; Tfb2m, r = 0.73, P < 0.001). Additionally, Tfam and Tfb2m mRNA levels positively correlated with those of cytochrome b and mitochondrial ATPase 6/8 (Tfam vs. cytochrome b, r = 0.48, P < 0.01; Tfam vs. mitochondrial ATPase 6/8, r = 0.49, P < 0.01; Tfb2m vs. cytochrome b, r = 0.58, P < 0.001; Tfb2m vs. mitochondrial ATPase 6/8, r = 0.57, P < 0.001; Supplementary material online, Figure S5).

3.6 TFAM and TFB2M overexpression blocks stress-induced reductions in Serca2a mRNA levels

We further examined whether TFAM and TFB2M have beneficial effects on the expression of the Serca2 gene in cardiac myocytes under stress. We found that 40 μmol/L NE reduced Serca2 gene transcription to 0.45-fold that of control levels, and mutation of the TFAM/TFB2M binding site (−122 to −114 nt from the transcription initiation site, Figure 3D) further decreased its transcription (Figure 6A). Hydrogen peroxide, a representative reactive oxygen species produced in the failing heart, showed similar results for Serca2 gene transcription. Overexpression of TFAM and TFB2M partially blocked the NE- and hydrogen peroxide-induced decreases in Serca2a mRNA levels in cardiac myocytes (Figure 6B). In contrast, TFAM and TFB2M ablation by siRNA resulted in further reduction in Serca2a mRNA levels in cardiac myocytes treated with NE or hydrogen peroxide (Figure 6C).

Figure 6

Effect of overexpression and ablation of TFAM/TFB2M on the transcription of Serca2 and Bnp genes in cardiac myocytes under stress. (A) Effect of 3 nt mutations in the TFAM/TFB2M binding site (−122 to −114 nt region) on Serca2 gene transcription after exposure to norepinephrine or hydrogen peroxide. The same reporter gene constructs as in Figure 4E were used. Serca2a mRNA levels after overexpression (B) and ablation (C) of TFAM and TFB2M following norepinephrine and hydrogen peroxide exposure. Bnp mRNA levels after overexpression (D) and ablation (E) of TFAM and TFB2M following norepinephrine and hydrogen peroxide exposure. All values represent the mean ± standard deviation of five independent measurements.

In accordance with these changes in Serca2a mRNA levels, TFAM and TFB2M overexpression partially blocked NE- and hydrogen peroxide-induced increases in Bnp levels in cardiac myocytes (Figure 6D), whereas TFAM and TFB2M ablation resulted in further increases in Bnp levels (Figure 6E).

4. Discussion

ATP production is a critical determinant of cardiac contractile and relaxation performance. The function of ATPases, especially SERCA2a, is highly dependent on the supply of ATP. In turn, the activity level of key enzymes for ATP production, such as pyruvate dehydrogenase, isocitrate dehydrogenase, α-ketoglutarate dehydrogenase, and ATP synthase, is dependent on the intracellular Ca2+ concentration.19 Thus, ATP production and Ca2+ handling are tightly coupled in cardiac performance. In this study, we provide the first evidence that transcription of genes involved in ATP production and Ca2+ handling is commonly regulated by TFAM and TFB2M, which were thought to be transcription factors specific for the mitochondrial genome.

Using several independent methods, including immunostaining, luciferase assay, ChIP assay, and fluorescence correlation spectroscopy, we confirmed that TFAM and TFB2M bind to the −122 to −114 and −122 to −117 nt regions of the 5′-regulatory region of the Serca2 gene, respectively. Although both TFAM and TFB2M have a mitochondrial-targeted signal at their N-terminal end, they do not possess a classical nuclear localization signal. Recently, Pastukh et al.20 have demonstrated that several amino acids in two high-mobility group domains and the C-terminal tail in TFAM play a significant role in the nuclear localization of TFAM. However, the high-mobility group domains found in TFAM do not exist in TFB2M. Regardless, we demonstrated that TFB2M has an immunofluorescence pattern similar to that of nuclear-localized TFAM and that TFB2M binds to the Serca2 gene in the nucleus of living cardiac myocytes, as shown by the ChIP assay. These data suggest that TFB2M possesses non-classical nuclear localization signals.

TFAM binds to mtDNA both non-specifically and in a sequence-specific manner. Kanki et al.21 demonstrated that the TFAM molecule is 1000-fold more abundant than mtDNA, abundant enough to completely enwrap mtDNA, confirming the architectural role of TFAM in the maintenance of mtDNA. Indeed, like histone for nuclear DNA, TFAM can non-specifically bind to mtDNA, most likely to protect mtDNA from various insults. Moreover, TFAM promotes bi-directional transcription by binding upstream of the light- and heavy-strand promoters of mtDNA in a sequence-specific manner. Therefore, TFAM has a dual effect on mtDNA, namely, mtDNA protection and initiation of transcription from mtDNA. In this study, we demonstrated that TFAM binds to the 5′-regulatory region of the Serca2 gene in a sequence-specific manner and increases the transcription and mRNA expression of this gene to approximately two-fold than that of the control. Thus, unlike the dual mode found for mtDNA, the mode of TFAM binding to nuclear DNA is sequence-specific rather than non-specific.

The 5′-regulatory region of the Serca2 gene possesses a large guanine cytosine (GC) box. We previously demonstrated that the SP1 transcription factor, a GC box-binding transcription factor, plays a significant role in the transcription of the Serca2 gene.12,14 The present study also demonstrated that the TFAM binding site is in the GC box. Therefore, TFAM and SP1 may compete for binding to the GC box in the 5′-regulatory region of the Serca2 gene. The TFAM binding site determined in this study overlaps with the reported SP1 binding sites,13 which was confirmed by fluorescence correlation spectroscopy (Figure 2B and Supplementary material online, Figure S6A). However, overexpression of SP1 increased Serca2 gene transcription in Serca2 constructs with mutations in the −122 to −114 region (TFAM binding site), suggesting that SP1 and TFAM bind separately (see Supplementary material online, Figure S6B). Double transfection of TFAM and SP1 increased Serca2 gene transcription to a greater degree compared with single transfection of either transcription factor alone (see Supplementary material online, Figure S6C). Moreover, fluorescence correlation spectroscopy revealed that TFAM and SP1 do not bind with each other (see Supplementary material online, Figure S6D). Taken together, TFAM and SP1 act in an independent fashion to regulate Serca2 gene transcription, although their binding sites are likely in close proximity to one another.

Serca2a mRNA levels are decreased in cardiac tissues in the context of heart failure,7,22 which we confirmed here using an experimental rat heart failure model following myocardial infarction (Figure 5A). Importantly, Tfam and Tfb2m mRNA levels correlated with the Serca2a mRNA level and tau (left ventricular diastolic parameter determined by the SERCA2a function) in this heart failure model. Correlation of Tfam and SERCA2a mRNA levels was also demonstrated in the models of adaptive cardiac hypertrophy induced by exercise training23,24 and pathological hypertrophy induced by pressure overload.25,26 Moreover, overexpression of TFAM and TFB2M increased Serca2 gene transcription, whereas their ablation or mutation of their binding sites in the Serca2 gene decreased Serca2 gene transcription. These data suggest that TFAM and TFB2M play roles in the maintenance of Serca2a mRNA levels under physiological conditions and that lack of support from TFAM and TFB2M leads to the decrease in Serca2a mRNA levels. Since SERCA2a plays a central role in the intracellular Ca2+ handling, TFAM and TFB2M may also play a role in the excitation–contraction coupling. The precise mechanism by which Tfam and Tfb2m mRNA levels decrease in the failing heart has not been fully elucidated. Tfam gene transcription is controlled by nuclear respiratory factors 1 and 2, upstream transcription factors, and its co-factor, peroxisome proliferator-activated receptor γ coactivator-1α. This coactivator is further regulated by its upstream signal molecules, myocyte-enhancing factor 2, histone deacetylase, and Ca2+-calmodulin-dependent protein kinase II.27 Wu et al.28 demonstrated that Ca2+-calmodulin-dependent protein kinase II is activated by a change in perinuclear Ca2+ concentration, which is controlled by inositol 1,4,5-triphosphate and cell membrane G-protein-coupled receptor. Heart failure is induced by an excess of various neurohumoral factors, including those that activate G-protein-coupled receptor signalling, and the present study demonstrated that high doses of NE and endothelin-1 decrease Tfam and Tfb2m mRNA expression (see Supplementary material online, Figure S7).

We also noted a preventive effect of TFAM and TFB2M on the decrease in Serca2a mRNA expression (Figure 6). Overexpression of TFAM and TFB2M restored the decreased Serca2a mRNA levels following NE and hydrogen peroxide exposure, whereas their ablation further decreased the Serca2a mRNA expression. This phenomenon is likely explained by a direct effect of TFAM and TFB2M on Serca2 gene transcription. However, the effect of TFAM and TFB2M overexpression is not limited to Serca2 and mitochondrial genes. Our microarray experiments showed increased expression of various other genes due to TFAM and TFB2M overexpression in neonatal rat cardiac myocytes, which together may help preserve the function of cardiac myocytes (see Supplementary material online, Table S2). In fact, we confirmed that TFAM and TFB2M also bind to the 5′-regulatory region of the gene of Na/K-ATPase, another critical ion transport protein in the plasma membrane (data not shown). The significance of TFAM has also been demonstrated by genetically engineered heart-specific Tfam knockout mice, which died in the neonatal period or showed dilated cardiomyopathy with atrioventricular conduction blocks.29 Conversely, TFAM transgenic mice inhibited left ventricular remodelling and improved survival following myocardial infarction.30 Our data thus further verify the molecular basis for the favourable effect of TFAM overexpression in the failing heart.

TFAM and TFB2M overexpression also increased the mtDNA copy number and mRNA levels for mitochondrial respiratory chain enzymes (Figure 4). Maniura-Weber et al.31 reported that transient overexpression of TFAM stimulated mtDNA transcription but not sufficiently enough to increase the mtDNA copy number in HeLa and HEK293 cells. In contrast, Ekstrand et al.32 reported that overexpression of TFAM in the mouse embryo resulted in the up-regulation of the mtDNA copy number. Thus, the effect of TFAM overexpression on the mtDNA copy number may depend on the cell type and the duration of overexpression. Together with the increase in mitochondrial respiratory chain enzyme mRNA levels, these data suggest that overexpression of TFAM and TFB2M may produce a therapeutic benefit in the failing heart by restoring mitochondrial function.

In conclusion, TFAM and TFB2M appear to regulate transcription of the nuclear genome-encoded Serca2 gene, in addition to mitochondrial genome-encoded genes (see Supplementary material online, Figure S8). These findings afford a molecular basis for the coordinate regulation of genes for mitochondrial ATP production and ATP consumption in the sarcoplasmic reticulum, which is necessary for maintaining dynamic cardiac function. TFAM and TFB2M may serve as a therapeutic target in the context of heart failure by protecting mitochondrial genome-encoded genes, as well as the Serca2 gene and other nuclear genome-encoded genes.

Funding

This work was supported in part by a Grant-in-Aid for Scientific Research (KAKENHI B-17390224) from the Japan Society for the Promotion of Science (JSPS).

Acknowledgements

We thank Dr Yasuhiro Ikeda (Yamaguchi University Graduate School of Medicine) for his technical advice and helpful discussion.

Conflict of interest: none declared.

References

View Abstract