Mitochondrial biogenesis and dynamics in the developing and diseased heart

Mitochondrial genomic and energy transduction machinery

The mitochondrion is a double-membrane organelle with an ion-permeable inner membrane and an outer membrane permeable to factors <5 kDa (Balaban 1990). This complex membrane structure enables ATP generation via OXPHOS. An electrochemical gradient established across the inner membrane drives OXPHOS and ATP synthesis. This gradient involves electrons donated from reduced forms of nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FADH₂) generated by oxidation of acetyl-CoA via the tricarboxylic acid (TCA) cycle. The acetyl-CoA is generated by oxidation of energy substrates, chiefly fatty acids and glucose–pyruvate in normal adult hearts (Stanley et al. 2005). The electrons are accepted and transferred across components of the electron transport chain (ETC) in the inner mitochondrial membrane. The transfer of electrons along the ETC is coupled to the transport of protons across the inner membrane, establishing the electrochemical gradient that drives ATP synthesis.

The mitochondrion is a unique organelle in that it contains its own self-replicating genome. The mitochondrial DNA (mtDNA) encodes 13 essential components of the ETC as well as all rRNAs and tRNAs necessary for translation of the mtDNA-encoded proteins (Scarpulla et al. 2012). However, the vast majority of mitochondrial proteins (>1000) (Pagliarini et al. 2008; Lotz et al. 2014) are encoded by the nuclear genome. Therefore, the biogenesis of mitochondria requires exquisite coordination of both mitochondrial and nuclear genomes. The control mechanisms for the orchestrated control of mitochondrial biogenesis are achieved largely through factors encoded by the nuclear genome, as described below. Indeed, all of the machinery needed for replication and transcription of the mitochondrial genome is contained within the >1000 nuclear-encoded mitochondrial proteins.

Biogenic machinery: transcriptional networks

A complex transcriptional network orchestrates nuclear and mitochondrial genome transcription and replication to conduct robust and dynamic mitochondrial biogenic responses in the heart. This system must coordinate both genomes during development and in response to physiological cues when there are changes in fuel substrate availability or energetic demands. A key breakthrough in our understanding of the transcriptional control of mitochondrial biogenesis and maturation came with the identification of the transcriptional coregulator peroxisome proliferator-activated receptor γ (PPARγ) coactivator 1α (PGC-1α). PGC-1α was first identified in brown adipose as a direct coactivator of PPARγ, a master regulator of adipogenesis (Puigserver et al. 1998). PGC-1α is a member of a family of transcriptional coregulators, including the closely related PGC-1β and a more distant member, PGC-1-related coactivator (PRC) (Andersson and Scarpulla 2001; Kressler et al. 2002; Lin et al. 2002). Gain- and loss-of-function studies have revealed a critical role for the PGC-1 coactivators in driving mitochondrial biogenesis and function in the heart and other mitochondrial-rich organs such as brown adipose tissue and skeletal muscle. Cardiac-specific transgenic overexpression of PGC-1α in mice results in an exuberant mitochondrial biogenic response during the postnatal period together with increased expression of nuclear-encoded mitochondrial genes (Lehman et al. 2000). Genetic deletion of either PGC-1α or PGC-1β in mice does not lead to overt abnormalities in mitochondrial function or biogenesis under basal conditions, indicating functional redundancy. However, loss of PGC-1α or PGC-1β accelerates cardiac dysfunction following the stress of pressure overload (Arany et al. 2006). Germline deletion of both PGC-1α and PGC-1β evokes perinatal lethal heart failure caused by a complete lack of cardiac mitochondrial biogenesis (Fig. 1; Lai et al. 2008). Mice lacking PRC exhibit an embryonic-lethal phenotype, dying shortly after implantation. These latter results support a critical and novel role for PRC in embryonic development (He et al. 2012). Taken together, these results indicate that PGC-1 signaling is both sufficient and necessary for cardiac mitochondrial biogenesis at birth and that PGC-1α and PGC-1β have overlapping functions.

The mouse phenotypes resulting from developmental stage-specific targeting of PGC-1β on a generalized PGC-1α-deficient background in mice has provided important insight into the roles of the coregulators in postnatal and adult mammalian hearts. Targeting of the PGC-1 coactivators after the perinatal biogenic response results in a progressive postnatal cardiomyopathy associated with dramatic mitochondrial morphological derangements indicative of derangements in mitochondrial fusion and fission (Fig. 1; Martin et al. 2014). These latter results demonstrate the importance of PGC-1 signaling for postnatal mitochondrial maturation, including a critical role for mitochondrial dynamics during growth of postnatal hearts. As described below, mitophagy is also required for this maturation process and is therefore included as a component of the process of mitochondrial dynamics.

Surprisingly, and in contrast to the germline and postnatal knockout phenotypes, inducible deletion of PGC-1 in adult hearts in mice does not compromise normal cardiac function or result in overt abnormalities of mitochondrial structure. However, adult PGC-1α/β-deficient mice exhibit global down-regulation of transcripts encoding components of the fatty acid oxidation (FAO), TCA, and ETC/OXPHOS pathways together with decreased respiratory capacity (Martin et al. 2014). These findings as well as observations by others (Chen et al. 2011) indicate that rates of mitochondrial fission and fusion (and mitochondrial turnover) are much lower in normal adult hearts compared with during postnatal developmental stages. Interestingly, a subset of mitochondria in adult PGC-1 double-knockout mice displayed ultrastructural mitochondrial cristae abnormalities similar to what is observed in Barth syndrome, a congenital disease caused by altered cardiolipin biosynthesis (Lai et al. 2014b). These latter results implicate PGC-1 in mitochondrial membrane lipid biosynthesis. Taken together with the results of the perinatal and postnatal PGC-1-deficient phenotypes, these studies have identified developmental stage-specific roles for the PGC-1 coactivators in the heart (Fig. 1).

The PGC-1 coactivators exert their function through direct interactions with target transcription factor effectors, allowing coordinate control of the various pathways involved in building new mitochondria (Fig. 2; Vega et al. 2015). Most of the work defining such interactions has focused on PGC-1α. PGC-1α enhances transcription by interacting directly with members of the nuclear receptor superfamily via specific LXXLL recognition domains, recruiting molecules that mediate chromatin remodeling via histone acetylation, and interacting with the TRAP/DRIP complex to recruit RNA polymerase II (Ge et al. 2002). Effector transcription factors within this cascade include members of the PPAR, estrogen-related receptor (ERR), and nuclear respiratory factor-1 (NRF-1) transcription factor families. As shown in Figure 2, these transcription factors control and regulate distinct, but at times overlapping, aspects of mitochondrial energy metabolism, respiration, and biogenesis.

NRF-1 and NRF-2 (GABP) regulate expression of virtually every complex in the ETC (Scarpulla 2008; Satoh et al. 2013). In addition, the PGC-1/NRF-1 interaction activates downstream factors involved in mtDNA replication. Specifically, NRF-1 activates transcription of genes encoding factors that mediate replication and transcription of the mitochondrial genome, including TFAM and TFB2M (Gleyzer et al. 2005). Consistent with these roles, genetic loss of either NRF-1 or NRF-2 results in embryonic lethality with reduced mtDNA content and ETC activity, underscoring the necessary role for each of these factors (Huo and Scarpulla 2001; Ristevski et al. 2004).

The ERR family of nuclear receptors serves a central function in the PGC-1 regulatory circuitry (Fig. 2). ERRα, ERRβ, and ERRγ have significant homology with classic estrogen receptors in the ligand-binding domain but do not bind estrogen or any known endogenous ligand (“orphan” nuclear receptors). The first evidence supporting a role for ERRα in the regulation of mitochondrial metabolism was the discovery that it regulates transcription of the gene encoding medium chain acyl-CoA dehydrogenase (MCAD), an enzyme that catalyzes the first step in mitochondrial FAO (Sladek et al. 1997; Vega and Kelly 1997). Broader links to PGC-1 and mitochondrial biogenesis were made when it was shown that PGC-1α interacts with ERRα via the results of a yeast two-hybrid screen (Huss et al. 2002; Schreiber et al. 2004). In cooperation with PGC-1α, ERRα has since been shown to regulate genes in virtually every pathway of mitochondrial energy transduction and ATP synthesis, including FAO, the TCA cycle, and ETC/OXPHOS (Huss et al. 2004). Genome-wide chromatin immunoprecipitation (ChIP) surveys confirmed direct binding of ERRα and ERRγ to the promoter regions of many nuclear-encoded mitochondrial genes (Dufour et al. 2007). Generalized ERRα knockout mice are fertile and viable but are susceptible to cardiac dysfunction in response to pressure overload and have reduced ATP synthesis rates following ischemic insult (Huss et al. 2007). ERRγ knockout mice die shortly after birth of heart failure with an inability to shift to oxidative metabolism (Alaynick et al. 2007). These latter results point to a critical role by ERRγ in the postnatal energy metabolic maturation of developing hearts. Recently, it has been shown that combined cardiac-specific deletion of ERRα/ERRγ displays premature lethality with reduced expression of genes critical for mitochondrial energy production (Wang et al. 2015). The role of the third member of the family, ERRβ, in cardiac mitochondrial function and energy production is not known. Notably, ERRα and ERRγ have also been shown to regulate contractile and calcium-handling gene expression in the heart and skeletal muscle, suggesting functions in coordinate control of mitochondrial energy production and contractile function in striated muscle (Dufour et al. 2007; Gan et al. 2013).

The PPAR family of nuclear receptor transcription factors can also serve as effectors of the PGC-1 coactivators. The PPARs (PPARα, PPARδ [also known as PPARβ], and PPARγ) were originally identified as regulators of peroxisomal β-oxidation (Issemann and Green 1990). PPARs are now known to regulate genes involved in mitochondrial FAO and many other cellular lipid metabolic pathways (van der Meer et al. 2010; Scarpulla et al. 2012; Kersten 2014; McMullen et al. 2014). Given that fatty acids are the preferred fuel substrate in normal adult hearts, accounting for 60%–90% of ATP production, the PPARs serve a critical role in the biogenesis of mitochondria specialized for high-capacity FAO in the heart. Target genes for PPARα in the heart include enzymes and proteins involved in fatty acid transport, activation, and mitochondrial FAO (Gulick et al. 1994; Brandt et al. 1998; Mascaro et al. 1998). The function of PPARs as regulators of cardiac myocyte fatty acid metabolism has been validated in vivo by PPARα loss of function and transgenic overexpression in mice (Aoyama et al. 1998; Djouadi et al. 1999; Leone et al. 1999; Watanabe et al. 2000; Finck et al. 2002). PPARs have large, hydrophobic ligand-binding domains that accommodate a variety of endogenous fatty acids and eicosanoids as activating ligands. In this way, PPARs serve as metabolic sensors to match fuel substrate availability and delivery with high FAO capacity. Although all PPARs bind as heterodimers to their cognate DNA elements with the retinoid X receptor (RXR), they may also interact with other transcription factors. For instance, recent work has shown that the Kruppel-like factor 15 (KLF15) interacts directly with PPARα in the heart to regulate genes involved in FAO (Prosdocimo et al. 2014, 2015).

In addition to the PGC-1 regulatory circuit, several other transcription factors have been shown to contribute to mitochondrial biogenesis. Such factors likely function independently of the PGC-1 coactivators in distinct physiologic or pathophysiologic contexts. The c-Myc transcription factor has been shown to induce mitochondrial function and activate NRF-1 and TFAM (Li et al. 2005; Kim et al. 2008). In the heart, interestingly, c-Myc triggers increased reliance on glucose and decreased FAO during times of growth or ischemia, a possible adaptive response to stress (Ahuja et al. 2010). In support of this latter role, c-Myc expression increases with pressure overload and improves recovery following an ischemic insult (Ahuja et al. 2010). These results suggest that c-Myc plays a unique role in the control of mitochondrial biogenesis and function when canonical pathways (PGC-1/ERR and PGC-1/PPAR) are not active, such as occurs during fetal periods or hypertrophic growth when the heart relies to a greater extent on glucose as a fuel substrate.

Regulatory inputs into mitochondrial biogenesis via regulation of PGC-1 coactivators

PGC-1α receives inputs from multiple pathways to control its expression and activity in diverse developmental and physiological contexts. As shown in Figure 2, PGC-1α expression is highly induced by exposure to cold and exercise, mediated largely through β-adrenergic stimulation and cAMP/CREB (cAMP response element-binding protein) signaling (Wu et al. 1999; Baar et al. 2002; Terada et al. 2002). In skeletal muscle, p38 MAPK also stimulates PGC-1α gene expression following a single bout of exercise mediated through the ATF2 transcription factor (Akimoto et al. 2005). Activation of AMPK (AMP-activated kinase) following acute exercise may also increase PGC-1α expression in skeletal muscle (Little et al. 2010). Activation of the calmodulin-dependent kinase and calcineurin by calcium signaling provides an additional mechanism to increase PGC-1α expression (Czubryt et al. 2003; Schaeffer et al. 2004). Finally, in the heart, the proviral integration site for Moloney murine leukemia virus (PIM) kinases has been shown to regulate PGC-1 expression and capacity for mitochondrial energy production. Interestingly, genetic deletion of all three Pim kinases (Pim-1, Pim-2, and Pim-3) results in a cardiac senescence phenotype with marked reduction of PGC-1α and PGC-1β concomitant with mitochondrial defects and decreased ATP levels (Din et al. 2014).

PGC-1α activity is also regulated by post-translational modifications, including phosphorylation and acetylation (Fig. 2). There is some evidence that phosphorylation by AMP kinase activates PGC-1α, providing a link between cellular energy status and mitochondrial biogenesis (Jager et al. 2007). PGC-1α has been shown to be acetylated at multiple lysine residues and is a substrate for the NAD⁺-dependent deacetylase sirtuin 1 (SIRT1) (Gerhart-Hines et al. 2007; Coste et al. 2008; Canto et al. 2009). As NAD⁺ levels are also regulated by AMPK, activation of PGC-1α through deacetylation by SIRT1 potentially creates a link between energy status, redox state, and mitochondrial function (Canto et al. 2009).

PGC-1 coactivators have been shown to cooperate with other factors to promote mitochondrial biogenesis and function (Fig. 2). The PGC-1 and ERR regulator in muscle 1 (Perm1) was recently shown to be required for PGC-1-induced mitochondrial biogenesis (Cho et al. 2013). Perm1 expression is enriched in skeletal muscle, heart, and brown adipose and regulates the expression of certain PGC-1 and ERR target genes. Perm1 itself is activated by ERR/PGC-1, providing a feed-forward mechanism to promote mitochondrial biogenesis. A similar type of regulatory loop is active in melanocytes, where PGC-1α activates expression of the microphthalmia-associated transcription factor (MITF) to promote the tanning response (Shoag et al. 2013). MITF also directly activates PGC-1α expression to increase oxidative capacity to complete this feed-forward regulatory loop (Haq et al. 2013). This regulatory loop has been described in a subset of melanomas (Vazquez et al. 2013). Finally, interactions between PGC-1 and the phosphatidate phosphatase lipin 1 have been shown to modulate PGC-1 activity. Lipins were first discovered as enzymes that catalyze the formation of diacylglycerol in the glycerolipid biosynthesis pathway (Csaki and Reue 2010). However, lipin 1 also has a nuclear function by which it physically interacts with PGC-1α and PPARα to activate FAO gene expression (Finck et al. 2006). In the heart, lipin 1 expression is activated by ERR/PGC-1α forming a feed-forward loop to modulate cardiac metabolism (Mitra et al. 2011).

Fuel shifts and energetic remodeling in hypertrophied and failing hearts

Pathological cardiac hypertrophy and heart failure are characterized by an increase in the expression of many contractile and structural genes that are normally expressed in fetal hearts along with decreased expression of normal adult genes. This reactivation of the so-called “fetal gene program” also applies to energy metabolism in heart failure. For instance, expression of PGC-1/PPAR-driven FAO genes is down-regulated during cardiac hypertrophy even prior to overt cardiac dysfunction (Sack et al. 1996; Lai et al. 2014a). Consistent with this, FAO rates and the contribution of β-oxidation-generated acetyl-CoA into the TCA cycle are reduced in pressure overload-induced cardiac hypertrophy and in failing hearts (Akki et al. 2008; Kolwicz et al. 2012; Zhang et al. 2013). In humans, cardiac hypertrophy has also been shown to be associated with reduced myocardial fatty acid utilization (de las Fuentes et al. 2006). These observations indicate that mitochondria are remodeled to a phenotype with reduced capacity for FAO during the development of cardiac hypertrophy and heart failure.

PGC-1/ERR signaling is also decreased in failing hearts in preclinical animal models and humans. Transcriptomic profiling has demonstrated down-regulation of these factors and many of the downstream targets in end-stage failing human heart samples (Gupte et al. 2014). Metabolomic signatures of failing human hearts provide evidence of metabolic dysfunction and a decline in oxidative metabolism with decreased short and medium chain acylcarnitines as well as TCA cycle intermediates (Gupte et al. 2014). Interestingly, similar profiling in earlier stages of heart failure in mice, while demonstrating the expected down-regulation of FAO enzyme gene expression (altered PPAR activity), revealed only minimal evidence for down-regulation of ETC and OXPHOS gene expression (Lai et al. 2014a). Taken together, these results imply progressive, pathologic remodeling and dysregulation of mitochondrial energy production during the development of heart failure. It is likely that deactivation of PPARα and subsequent decreased FAO are early events that contribute to a switch to other fuel substrates, such as glucose and ketones. However, as the disease progresses, expression of ERR and PGC-1 is also down-regulated, resulting in a greater reduction of mitochondrial oxidative capacity. Additionally, the results suggest that post-translational modifications likely contribute to the metabolic derangements during the development of heart failure (Lai et al. 2014a).

Pathologic cardiac remodeling, biogenesis, and mitophagy

Given the known links between mitophagy and mitochondrial biogenesis, what role does mitophagy play during pathologic cardiac remodeling relevant to heart failure? Much less is known about this area, but interest is emerging. As stated above, although unstressed Parkin-deficient mice have normal cardiac function, they accumulate abnormal mitochondria with age (Kubli et al. 2013a). In addition, deletion of Parkin and impaired mitophagy result in sensitization to ischemic injury, as Parkin knockout mice have larger infarcts and pathologic remodeling following myocardial infarction (Kubli et al. 2013a). Parkin levels and mitophagy are also rapidly increased in the border zone following infarct. There is also evidence that increased mitophagy provides a mechanism for protective ischemic preconditioning (Huang et al. 2011). Ischemic preconditioning stimulates Parkin translocation to the mitochondria and is attenuated in Parkin knockout mice (Huang et al. 2011). Given these results, it is tempting to speculate that pathologic remodeling in the heart involves waves of mitochondrial biogenesis and mitophagy. It is also possible that physiological stressors such as exercise or pregnancy could activate rounds of adaptive mitophagy and biogenesis. Answers to these questions may lead to insights into targeting these pathways for therapeutic benefit in the setting of heart failure.