Functional interaction of regulatory factors with the Pgc-1α promoter in response to exercise by in vivo imaging

Abstract

Real-time optical bioluminescence imaging is a powerful tool for studies of gene regulation in living animals. To elucidate exercise-induced signaling/transcriptional control of the peroxisome proliferator-activated receptor-γ coactivator-1α (Pgc-1α) gene in skeletal muscle, we combined this technology with electric pulse-mediated gene transfer to cotransfect the Pgc-1α reporter gene with plasmid DNA encoding mutant/deletion forms of putative regulatory factors and, thereby, assess the responsiveness of the promoter to skeletal muscle contraction. We show that each of the myocyte enhancer factor 2 sites on the Pgc-1α promoter is required for contractile activity-induced Pgc-1α transcription. The responsiveness of the Pgc-1α promoter to contractile activity could be completely blocked by overexpression of the dominant-negative form of activating transcription factor 2 (ATF2), the signaling-resistant form of histone deacetylase (HDAC) 5 (HDAC5), or protein kinase D (PKD), but not by HDAC4. These findings provide in vivo evidence for functional interactions between PKD/HDAC5 and ATF2 regulatory factors and the Pgc-1α gene in adult skeletal muscle.

extensive research in signaling and molecular mechanisms of skeletal muscle plasticity suggests that peroxisome proliferator-activated receptor-γ coactivator-1α (Pgc-1α), a versatile transcriptional coactivator (15), plays a pivotal role in exercise-induced genetic reprogramming in skeletal muscle (1, 3, 10, 17, 27). Although Pgc-1α activity could be controlled by posttranslational modification (8, 16), there is clear evidence that the Pgc-1α gene is transcriptionally activated in response to exercise, which could play an important role in skeletal muscle adaptation. Elucidating the signal transduction that leads to transcriptional regulation of the Pgc-1α gene will, therefore, likely provide valuable insights into the fundamental mechanisms of skeletal muscle plasticity and help in development of effective therapeutics for chronic diseases that are related to skeletal muscle contractile and metabolic functions. A great challenge to exercise scientists is that exercise training cannot be faithfully recapitulated in experimental models in vitro. An imaging system with real-time imaging of gene regulation in living animals will significantly facilitate this important research area.

The mouse Pgc-1α gene, similar to its mammalian homologs, contains conserved myocyte enhancer factor 2 (MEF2; at −2901 and −1539) and cAMP response element (CRE; at −222) sequences in the promoter that converge biological signals for the transcriptional control of the gene (6, 12). We previously established a novel optical bioluminescence imaging system in living mice and showed that double mutation of the two MEF2 sites or single mutation of the CRE site resulted in loss of Pgc-1α promoter responsiveness to motor nerve stimulation in skeletal muscle (2). These findings support the notion that exercise-induced Pgc-1α transcription is mediated by regulatory factors that interact with these cis elements on the Pgc-1α promoter. The functional interaction between regulatory factors and Pgc-1α promoter in response to increased neuromuscular activities remains to be elucidated in vivo.

MATERIALS AND METHODS

Animals.

Male C57BL/6J mice (7–8 wk old) were obtained commercially (Jackson Laboratory) and housed in temperature-controlled (21°C) quarters with a 12:12-h light-dark cycle and water and chow (Purina) provided ad libitum. All experimental protocols were approved by the Duke University Institutional Animal Care and Use Committee.

Plasmid DNA constructs.

Pgc-1αL is a construct containing a 3.1-kb 5′-flanking region of the mouse Pgc-1α gene upstream of the firefly luciferase coding sequence (6), and mutations of each of the MEF2 binding sites, ΔMEF2(−2901) and ΔMEF2(−1539), were generated by site-directed mutagenesis based on the sequence information described previously (6). Pgc-1αL with CRE site mutation, ΔCRE(−222), has been used previously (2). Plasmid HDAC5(S2A) encodes FLAG-tagged histone deacetylase 5 (HDAC5) with mutations of serine sites at 259 and 498 to nonphosphorylatable alanine sites (31). Similarly, HDAC4(S3A) encodes FLAG-tagged HDAC4 with serine-to-alanine mutations at residues 246, 467, and 632 (38). These mutations render the mutant HDAC proteins resistant to upstream signals (31, 38). Plasmid PKD(K/W) encodes a catalytically inactive mutant of protein kinase D (PKD) with a hemagglutinin (HA) tag (31). The dominant-negative ATF2ΔN encodes a FLAG-tagged, truncated activating transcription factor 2 (ATF2) with NH₂-terminal deletion from amino acid 1 to 137 (30). pCI-neo is a mammalian expression vector without coding region (Promega) used as a negative control for the mutant/deletion expression vectors described above. pEGFP-N1 was purchased from Invitrogen. All plasmid DNAs were purified using the Endo-Free Plasmid Mega kit (Qiagene) and dissolved at a concentration of 2.5 mg/ml in 0.9% NaCl solution.

Electric pulse-mediated gene transfer.

Electric pulse-mediated gene transfer was modified from previous studies (2, 9, 25). Briefly, under pentobarbital sodium anesthesia (50 mg/kg ip), both tibialis anterior (TA) muscles were injected with a mixture of DNA (15 μg of Pgc-1αL and 50 μg of plasmid DNA encoding a mutant/deletion form of regulatory factor) by use of a 0.5-ml syringe with a 28-gauge needle at a rate of <0.015 ml/min. Eight electric pulses (100 ms, 1 Hz, 100 V) were delivered immediately to the injected TA muscle using a square-pulse stimulator (model S88K, Grass Telefactor) through a two-needle array (model 533, BTX) placed on the medial and lateral sides of the TA muscle, so that the electrical field was perpendicular to the long axis of the myofibers. Mice were allowed to recover for 10 days before electrode implantation, nerve stimulation, and imaging analysis.

Motor nerve stimulation and optical bioluminescence imaging.

Under anesthesia, electrodes were implanted, and motor nerve stimulation was initiated within 30 min and continued for 2 h. In vivo bioluminescence images were acquired and analyzed as described previously (2). The total flux (in photons·s⁻¹·cm⁻²·steradian) within the region of interest was measured. The ratio of the total flux from the stimulated muscle to that from the contralateral control muscle was determined before and after motor nerve stimulation.

Western immunoblot analysis.

TA muscles were homogenized and analyzed by immunoblot as described elsewhere (32). Antibodies used were ANTI-FLAG polyclonal antibody (F-7425, Sigma), anti-HA antibody (Roche), and anti-α-tubulin antibody (Sigma).

Statistics.

Luciferase activity in the stimulated TA muscle was divided by that in the contralateral control muscle. Values before and after motor nerve stimulation were compared by paired Student's t-test, with P < 0.05 considered statistically significant. Values are means ± SE.

RESULTS AND DISCUSSION

To further define the functional role of the cis elements on the Pgc-1α promoter, we performed site-directed mutagenesis for each of the MEF2 sites individually and confirmed that they are equally important for the transcriptional activation of the Pgc-1α gene in skeletal muscle in vivo (Fig. 1). These findings raised the possibility that exercise-induced Pgc-1α transcriptional activation is fully dependent on functional interactions of regulatory factors with the MEF2 and the CRE cis elements on the Pgc-1α promoter.

Fig. 1.

Peroxisome proliferator-activated receptor-γ coactivator-1α (Pgc-1α) promoter activity is dependent on each of the myocytic enhancer factor 2 (MEF2) and cAMP response element (CRE) sites. After transfection of both tibialis anterior (TA) muscles by Pgc-1αL or its derivative mutant constructs, the motor nerve of one of the TA muscles was stimulated for bioluminescence imaging analysis. A: schematic presentation of plasmid DNAs. Pgc-1α promoter (3.1 kb) is presented as a solid line with two MEF2 sites and one CRE site. Crosses denote site-directed mutagenesis as described elsewhere (6, 12). B: in vivo imaging of luciferase activity in mouse TA muscles from Pgc-1αL and its mutant derivatives ΔMEF2(−2901), ΔMEF2(−1539), and ΔCRE(−222). Imaging analysis was performed 10 days after gene transfer. Pseudocolors overlaid on the image indicate intensity of luminescent signals from luciferase reporter gene activity. Animals' right TA muscles were stimulated (S), and their left TA muscles were sham-operated without stimulation and used as reference control (C). C: quantification of luciferase activity in TA muscles. Dashed horizontal line denotes basal level before stimulation. Values are means ± SE; n = 6–8. **P < 0.01 vs. poststimulation values.

To address this issue, we chose to employ a loss-of-function approach by cotransfecting mutant/deletion forms of the putative regulatory factors in adult skeletal muscle with the Pgc-1αL reporter gene. This approach, analogous to cotransfection approaches used extensively in cultured cells, had not been possible in a living animal until recent improvement of the gene transfer technique in adult skeletal muscle (9, 25). We first checked the efficiency of in vivo cotransfection to ensure possible dominant-negative effect of the transgenes. Gene transfer of a plasmid DNA containing the enhanced green fluorescent protein transfected ≥60% of the myofibers in mouse TA muscles (Fig. 2A), which maintained normal morphology following the recovery (10 days). We then cotransfected TA muscles with plasmid DNAs containing Pgc-1αL reporter gene (3.1-kb mouse Pgc-1α promoter driving firefly luciferase) and an empty control vector (pCI-neo) or plasmid DNA encoding epitope-tagged mutant/deletion forms of putative regulatory proteins. Consistent with efficient gene transfer, transgene products, but not the empty control vector, could be readily detected by immunoblot analysis (Fig. 2B). These results demonstrated high feasibility for the in vivo approach and prompted us to proceed with the investigation of functional interactions of the putative regulatory factors with the MEF2 and CRE cis elements on the Pgc-1α promoter.

Fig. 2.

Efficient gene expression following electric pulse-mediated gene transfer. Plasmid DNA [65 μg of pEGFP-N1 or 15 μg of Pgc-1αL and 50 μg of plasmid DNA encoding an empty control vector (pCI-neo) or a mutant/deletion form of regulatory factor] was transfected into TA muscles, which were visualized by epifluorescent microscopy (A) and subjected to homogenization and immunobloting analysis (B). A: TA muscle transfected with pEGFP-N1 was harvested and sectioned for phase contrast (Phase) and fluorescent microscopic examination for expression of green fluorescent protein (GFP). There were ∼60% GFP-positive myofibers. Red lines outline boundary of muscle sections. *, Myofibers that are negative for GFP signals. Scale bars, 250 and 50 μm for ×4 and ×40 images, respectively. B: expression of ATF2ΔN, HDAC5(S/A), HDAC4(S/A), and PKD(K/W) could be easily detected in transfected TA muscles by immunoblot using anti-FLAG or anti-hemagglutinin (HA) antibodies for tagged proteins and anti-α-tubulin antibodies for protein loading. ATF2, activating transcription factor 2; HDAC, histone deacetylase; PKD, protein kinase D.

MEF2 activity has long been implicated in muscle plasticity (35, 36), and association of MEF2 with class II HDACs, such as HDAC4 and HDAC5, results in repression of the MEF2 function (19). It is believed that intracellular signals elicited by the neuromuscular activities result in release of the MEF2 from inhibition by HDACs, and, as a consequence, the MEF2 becomes transcriptionally active and transactivates the target genes, such as Pgc-1α, directing phenotypic adaptation in skeletal muscle. The interaction between class II HDACs and MEF2 serves as an ideal “on-and-off” control for skeletal muscle adaptation. A recent study in which skeletal muscle-specific overexpression of a constitutively active form of MEF2C enhanced expression of proteins in the oxidative metabolism pathway and resulted in formation of more oxidative fibers in otherwise glycolytic muscles provides direct proof of this concept (28). Indeed, HDAC-mediated control of metabolic function has also been shown in Pgc-1α gene expression in oxidative cardiac myocytes; loss of such control leads to the detrimental consequence of heart failure due to reduced mitochondrial biogenesis and function (6). However, it has yet to be demonstrated in a loss-of-function manner that MEF2 controls the Pgc-1α gene transcription in adult skeletal muscle in vivo.

We chose to focus on two class II HDACs, HDAC4 and HDAC5, because they bind to and repress MEF2 activities (19, 22) and their activities are highly regulated by neuromuscular activity (18, 20, 28). Cotransfection of Pgc-1α reporter gene with an empty control vector (pCI-neo) in TA muscle followed by motor nerve stimulation resulted in a twofold increase in luciferase activity (Fig. 3), and cotransfection of a nonphosphorylatable, signal-resistant mutant of HDAC4 [HDAC4(S3A)] or HDAC5 [HDAC(S2A)] did not affect the basal activity of the Pgc-1α reporter gene (not shown). However, motor nerve stimulation-induced Pgc-1α transcription was completely blocked by overexpression of HDAC5(S2A), but not HDAC4(S3A) (P < 0.05 vs. contralateral control TA muscle; P = 0.43 vs. pCI-neo empty vector control). These findings suggest that the Pgc-1α gene is preferentially controlled by the HDAC5-MEF2 interaction.

Fig. 3.

Contractile activity-dependent functional interaction between Pgc-1α promoter and upstream regulatory factors. A: in vivo imaging of luciferase activity in mouse TA muscles from Pgc-1αL cotransfected with an empty control vector (pCI-neo), ATF2ΔN, HDAC5(S/A), HDAC4(S/A), or PKD(K/W) before (Pre) and 2 h after (Post) low-frequency (10-Hz) nerve stimulation (2 h). Animals' right TA muscles were stimulated (S), and their left TA muscles were sham-operated without stimulation and used as reference control (C). B: quantitative analysis of luciferase activity from Pgc-1αL. Dashed horizontal line denotes basal level before stimulation. Values are means ± SE. *P < 0.05; **P < 0.01 vs. poststimulation values.

The protein kinase C (PKC) downstream effector PKCμ/PKD is a confirmed kinase for HDAC5 (13, 31). We hypothesized that PKD transduces the signals from neuromuscular activity to the Pgc-1α gene by phosphorylating HDAC5 and releasing MEF2 from inhibition. Consistent with this hypothesis, cotransfection of a signal-resistant PKD [PKD(K/W)] completely blocked motor nerve stimulation-induced transcriptional activation of the Pgc-1α gene in skeletal muscle. More recently, it has been shown that HDAC5 can be phosphorylated by AMP-activated protein kinase, leading to reduced association of HDAC5 with the promoter of glucose transporter 4 and enhanced the gene expression (21). Taken together, these findings indicate the importance of HDAC5 in the control of the skeletal muscle metabolic phenotype through the Pgc-1α gene.

The p38 mitogen-activated protein kinase (MAPK) pathway in skeletal muscle has been a focus of research (26, 29, 37), inasmuch as various forms of contractile activity activate the p38 MAPK pathway in skeletal muscle (4, 5, 14, 33), suggesting its importance in muscle adaptation. We and others recently obtained evidence that activation of the p38 MAPK and its downstream effector ATF2 participates in Pgc-1α gene expression in skeletal muscle (1, 34). ATF2, as a member of the leucine zipper family of transcription factors, binds to the CRE for its transactivation function. Pharmacological inhibition of the p38 MAPK pathway or overexpression of a dominant-negative form of ATF2 blocks the transcriptional activation of the Pgc-1α gene in skeletal muscles in culture (1, 34). However, the functional role of ATF2 in transcriptional control of the Pgc-1α gene in vivo has not been confirmed in skeletal muscle in a physiological model of exercise. Here, we showed that overexpression of a truncated, dominant-negative form of ATF2 (30) completely blocked contractile activity-induced Pgc-1α reporter gene activity, providing direct evidence that ATF2 plays an essential role in the transcriptional control of the Pgc-1α gene in skeletal muscle (Fig. 3).

Interestingly, the PKD/HDAC5/MEF2 and p38/ATF2 signal transductions have been linked to the stress and the calcium signals in striated muscles (7, 23, 31, 34). The functional interactions of these two regulatory modules with the Pgc-1α promoter observed in the present study are consistent with the notion that multiple cellular signals converge to control a powerful regulatory factor, such as Pgc-1α, in defining skeletal muscle phenotype (Fig. 4). The multiple signal/transcription control system would not only allow for more sophisticated fine tuning of the adaptive functions, but it would also ensure the validity of the signals to avoid unnecessary dysregulation of the system. Future research should focus on dissection of each one of the physiological signals in skeletal muscle adaptation. The analytic system employed in the present study by combining efficient gene transfer of regulatory factors with biochemiluminescence imaging of reporter gene is an exciting step toward a complete elucidation of all the signaling and transcription mechanisms of skeletal muscle adaptation and could be applicable to other organ systems and disease models.

Fig. 4.

Model for exercise-induced transcription of the Pgc-1α gene in skeletal muscle. Increased neuromuscular activities elicit signals leading to activation of the p38 MAPK pathway and transcriptional upregulation of the Pgc-1α gene through ATF2 (1, 34) and MEF2 (11, 37). Activation of the p38 MAPK pathway can also promote Pgc-1α transcription by inhibiting p160 and derepressing PGC-1α protein function (8), exerting a positive regulation of Pgc-1α gene transcription through MEF2 (12, 24). In parallel, exercise also promotes Pgc-1α transcription through PKD/HDAC5-mediated derepression of MEF2 function (6). These signaling-transcription coupling cascades integrate contractile and metabolic cues from neuromuscular activity to control expression of the Pgc-1α gene in skeletal muscle adaptation. Regulatory factors/cis elements, the functional role of which has been tested and confirmed in the present study, are shown as gray boxes with solid lines. Other relevant regulatory factors that were not investigated in the present study are shown as open boxes with dashed lines.

GRANTS

This study was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant AR-050429 to Z. Yan.