Neurohormonal connections with mitochondria in cardiomyopathy and other diseases

Abstract

Neurohormonal signaling and mitochondrial dynamism are seemingly distinct processes that are almost ubiquitous among multicellular organisms. Both of these processes are regulated by GTPases, and disturbances in either can provoke disease. Here, inconspicuous pathophysiological connectivity between neurohormonal signaling and mitochondrial dynamism is reviewed in the context of cardiac and neurological syndromes. For both processes, greater understanding of basic mechanisms has evoked a reversal of conventional pathophysiological concepts. Thus, neurohormonal systems induced in, and previously thought to be critical for, cardiac functioning in heart failure are now pharmaceutically interrupted as modern standard of care. And, mitochondrial abnormalities in neuropathies that were originally attributed to an imbalance between mitochondrial fusion and fission are increasingly recognized as an interruption of axonal mitochondrial transport. The data are presented in a historical context to provide insight into how scientific thought has evolved and to foster an appreciation for how seemingly different areas of investigation can converge. Finally, some theoretical notions are presented to explain how different molecular and functional defects can evoke tissue-specific disease.

INTRODUCTION

The physiological effects of a neurohormone were first recognized at University College of London by the English physician George Oliver and physiologist Edward Schafer, who described increased heart rate and blood pressure with arteriole constriction in animals receiving extracts of adrenal medulla (1). The active principal, epinephrine, was purified in 1899 by John Jacob Abel at Johns Hopkins University, and independently in 1900 by Japanese chemist Jokichi Takamine (2). The chemical form, derived by Thomas Aldrich of Parke-Davis in 1901, was patented and ushered in a century of research and therapeutic development based on modulating neurohormone signaling.

Central to understanding neurohormonal effects was the notion, advanced by William Bayliss and Ernest Starling from their studies of secretin, that epinephrine acted as a chemical messenger secreted by one organ to modulate the function of others. The concept that adrenaline effects were specifically mediated by “adrenotropic” receptors emerged from multiple studies linking sympathetic nerve activity to epinephrine-like responses in uterine, bronchial, and arteriolar smooth muscle. However, the observation that sympathetic stimulation evoked distinct end-organ effects under different experimental conditions, such as in the presence of sympatholytic ergot, engendered a controversy over whether there were multiple factors released from sympathetic nerve endings versus multiple different receptor/detectors for a single factor. The controversy was resolved in favor of multiple receptors by H.H. Dale who described both a “receptive mechanism for adrenaline” at the neuromuscular junction, and a “mixed response” consisting of either smooth muscle contraction or relaxation in response to the single hormone (3, 4). Thirty years later, the hypothetical adrenotropic receptors were pharmacologically defined in groundbreaking studies by Raymond P. Ahlquist who discovered differences in the rank-order potency of a series of structurally distinct sympathomimetic amines to provoke vasoconstriction, pupillary dilation and uterine contraction. Ahlquist designated the two types of receptors “α-” and “β-adrenotrophic receptors” (5). This classic study laid the foundation for subsequent identification of hundreds of individual members of the superfamily of 7-transmembrane (7-TM) spanning receptors (6) and the rational therapeutic application of receptor subtype-specific agonists (as in β-AR activating isoprenaline and its successors for asthma) and antagonists (as in β1-AR blocking metoprolol for cardiac insufficiency).

Different functional effects of the nine modern adrenergic receptor (AR) subtypes (α_1A, α_1B, α_1D, α_2A, α_2B, α_2C, β₁, β₂, and β₃) and indeed for all 7-TM receptors, derive from their coupling to specific heterotrimeric (i.e., having 3 different protein subunits, α, β, and γ) G-protein effectors. This mix-and-match receptor-effector design confers rich functional diversity through tissue-specific expression of 7-TM receptors (providing organ-selective responses to hormonal stimulation) and coupling of distinct receptor subtypes to different G-protein signal effectors. Whereas many hundreds of 7-TM receptors have been identified, they are coupled to only a few heterotrimeric G-proteins, encoded by 16 α subunit genes, 6 β subunit genes, and 12 γ subunit genes. Using prototypical adrenergic receptors as exemplars (Fig. 1), β₁-AR coupling to G_s stimulates adenylyl cyclase and mediates positive cardiac inotropy (the vigor of myocardial contraction) and chronotropy (the rate of cardiac contraction), α₂-AR coupling to G_i inhibits adenylyl cyclase and suppresses synaptic neurotransmitter release, and α₁-AR coupling to G_q/11 activates phospholipase C and provokes vasoconstriction. Thus, a century-long investigative path beginning with the observation that crude extracts of adrenal medulla provoke acute tachycardia and hypertension led to our modern understanding of one of the most widespread stimulus-response systems in biology, responsible for such diverse physiology as the fight-or-flight adrenergic response and the senses of vision and smell. The structural determinants of G-protein coupled 7-TM receptor signaling and functional details beyond the scope of this manuscript have been elegantly reviewed by Nobel Laureate Brian Kobilka et al. (7).

Figure 1.

Mix-and-match of neurohormone receptors and effectors for functional diversity and tissue-specificity of response. Ribbon models of receptors were produced using swissmodel.expasy.org. Blue is amino terminus; red is carboxy terminus. Heterotrimeric G-proteins subunit color: Gα, green; Gβ, red; Gγ, purple.

NEUROHORMONES, BIOMECHANICS, AND CARDIAC STRESS ADAPTATION

Clinical therapeutics based on mimicking, interrupting, or otherwise modulating neurohormone synthesis or signaling are a cornerstone of modern medical therapy. Indeed, therapeutic interdiction of the adrenergic and renin-angiotensin systems has become the foundation for current pharmacological care of cardiac hypertrophy and heart failure. However, for decades after Powell and Slater (8) observed that dichloroisoprenaline (a chemical analog of norepinephrine) inhibited adrenergic receptors, and Nobel Laureate Sir James Black subsequently developed prototype β-adrenergic receptor antagonists (i.e., β-blockers) including propranolol (9–11), adrenergic blockade was proscribed in heart failure. Propanolol, the first β-blocker to be widely employed, found immediate therapeutic use in acute ischemic heart disease because it suppressed myocardial work (approximated by the product of heart rate and systolic blood pressure), thereby diminishing myocardial oxygen requirements to levels that could be accommodated by a diseased coronary arterial tree. The negative inotropic and chronotropic effects of cardiac β-adrenergic receptor blockade (12) were, by unassailable mechanistic logic and reports of adverse patient experiences to sympatholytic therapy (13, 14), viewed as contraindications to β-AR blockade in heart failure. Even advocates of β-blocker therapy cautioned against their use in heart failure: “Patients with left ventricular insufficiency should not be given a β-receptor blocking drug without prior administration of digitalis and diuretics” (15).

It is worth briefly considering how β-blocker therapy went from being contraindicated to a first-line, life-prolonging therapy in heart failure. When β-blockers were first developed and introduced into clinical practice, conventional wisdom regarding the regulation of integrated cardiovascular function was deeply rooted in biophysics, within the field of “cardiac mechanics.” Both acute cardiac pump function and chronic structural adaptation of the heart were understood in terms of two 19th century mathematical formulae: “Ohm’s law” was used to describe how the heart pump and its vascular conduits interact to maintain tissue perfusion, and the “Laplace relationship” described how ventricular remodeling (i.e., changes in wall thickness and/or chamber dimension) affected overall ventricular pump function (Fig. 2).

Figure 2.

How Ohm’s law and the Laplace equation can explain cardiac adaptation.

Ohm’s law is most recognized for its application to electrical circuits, but as applied to the cardiovascular system it describes how blood flow (I, or cardiac output) is determined by the pressure differential across the vascular bed (E, which for the systemic circulation is the difference between systemic arterial pressure and left atrial/pulmonary capillary wedge pressure) and vascular resistance (R, which increases as the integrated cross sectional area of the arteriolar bed decreases, i.e., during vasoconstriction) (Fig. 2). When tissue perfusion pressure (E) is constant/normal, Ohm’s law predicts that pump function/cardiac output (I) will decrease as vascular resistance (R) increases. This relationship can be used to understand the norepinephrine-mediated adrenergic fight or flight response: In the context of great emotional excitation, intense physical stress or acute massive blood loss, the imperatives are to maintain tissue perfusion pressure (E) by increasing systemic vascular resistance (R) via vasoconstricting α1-AR, and to enhance cardiac output (I) via the inotropic effects of myocardial β1-AR (see Fig. 1). According to Ohm’s law, an isolated decrease in cardiac output, as with primary cardiac insufficiency or acute massive blood loss, lowers tissue perfusion (Fig. 2). The integrated cardiovascular system reacts by invoking the adrenergic fight-or-flight response, including elaboration of epinephrine and norepinephrine. Thus, increased β1-AR mediated myocardial inotropy and chronotropy enhances pump function while intense α1-AR mediated vasoconstriction reduces blood flow to skin and periphery, thereby preferentially shunting the cardiac output to major organs and brain. Peripheral vasoconstriction and increased heart rate resulting from adrenergic stimulation underlie some of the classic physical signs of systolic heart failure, such as cool and pale extremities with a rapid and thready pulse. However, the Laplace relationship shows that increased systolic blood pressure from peripheral vasoconstriction further stresses the heart by increasing afterload (Fig. 2). Although a transient increase in cardiac afterload would likely be tolerated by a normal heart during fight-or-flight, increased afterload will further compromise cardiac pump function in primary cardiac insufficiency caused by, for example, acute myocardial infarction and viral or toxic myocarditis. Taken together, these concepts provided support for the notion that heart failure patients are “critically dependent on sympathetic activity … to maintain cardiac output” (15) and therefore should avoid β-blockers.

The Laplace relationship (Fig. 2) as applied to the left ventricle (LV) describes how chamber geometry and arterial pressure help determine LV ejection performance, i.e., the efficiency with which the heart pumps blood into its arterial conduits. In the context of LV systolic function, wall stress (σ) represents afterload, i.e., the intrinsic and extrinsic forces that oppose ventricular ejection; increased wall stress/afterload impedes ventricular ejection. The Laplace relationship predicts that afterload/systolic wall stress will increase as LV (systolic) pressure (p) increases (i.e., hypertension) and as LV ventricular size (r) increases (i.e., dilated cardiomyopathy), to the detriment of pump function. Conversely, wall stress will favorably decrease as LV wall thickness (h) increases, as in compensated cardiac hypertrophy (16, 17). Importantly, pressure is a variable in both the Laplace and Ohm relationships and is regulated in part via neurohormone-mediated vasoconstriction. The phenomenon wherein hearts modify their chamber geometry in response to different pathophysiological conditions is called cardiac remodeling and is driven in large part by hypertrophic growth or programmed death of constituent cardiac myocytes. The next section reviews the role of neurohormone signaling in these cellular events.

THE NEUROHORMONAL HYPOTHESIS OF CARDIAC HYPERTROPHY AND HEART FAILURE

As introduced, early observations that circulating epinephrine and norepinephrine levels were increased in heart failure and that β-blocker administration to heart failure patients caused acute functional deterioration, led to the widely accepted notion that failing hearts rely upon the positive inotropic, chronotropic, and vasoconstrictor effects of catecholaminergic stimulation to sustain cardiac pump output and peripheral blood pressure (18–22). The counter-hypothesis was that neurohormonal activity contributed to heart failure, which suggested that interrupting neurohormone signaling could be therapeutic. Various contrarian mechanisms were proposed in support of therapeutic trials of neurohormonal blockade in heart failure, including the idea that myocardial β-adrenergic receptor downregulation from heart failure catecholamine excess could be reversed by gentle β-AR therapy, i.e., “resensitization” (23–25). Another putative mechanism held that catecholamines and other vasoconstricting neurohormones could directly stimulate cardiac myocyte hypertrophy, independent of mechanical stress produce by increased hemodynamic load. This “neurohormonal hypothesis” was formalized by Milton Packer (26, 27), but the concept is founded upon a series of paradigm-changing studies enabled by Paul Simpson’s development of an experimental platform in which the response of tissue cultured cardiac myocytes to neurohormonal activation could be studied in the absence of altered mechanical loading (28–30). Simpson (28) initially observed that epinephrine provoked hypertrophy of unloaded neonatal rat cardiac myocytes in culture. Others subsequently adapted neonatal rat or embryonic chick cardiac myocyte systems to demonstrate analogous prohypertrophic effects of angiotensin II (31, 32) and endothelin I (33, 34). Later, Sadoshima and Izumo (35) applied neonatal rat cardiac myocytes to deformable substrates, revealing the existence of a stretch-stimulated endogenous renin-angiotensin system within cardiac myocytes, and an autocrine/paracrine system that linked mechanical stress to local, endogenous cardiomyocyte neurohormone release. Taken together, these studies showed the sufficiency of neurohormonal activation through multiple independent pathways to induce cardiac myocyte hypertrophy in vitro.

Neonatal rat cardiac myocytes in tissue culture are unlikely to react to epinephrine, angiotensin II or endothelin I in a manner that exactly recapitulates the in vivo responses of functioning hearts. Thus, it remained possible that hemodynamic stress was the dominant direct in vivo stimulus for cardiac remodeling, and that neurohormonal effects were required to sustain in vivo cardiac pump function and blood pressure as originally envisioned. Alternately, neurohormones might play a subordinate prohypertrophic role compared with the dominant effects of increased hemodynamic load. These possibilities required in vivo testing, but the straightforward study of administering the neurohormones continuously over a period of time (as with an osmotic mini-pump) and determining if that provoked cardiac remodeling could not provide an answer because the vasoconstricting effects of prohypertrophic neurohormones increases blood pressure (p), leading us back to increased biomechanical stress and the Laplace and Ohm equations.

A new experimental approach was needed to bypass the confounding vasoconstrictor effects of systemic neurohormonal stimulation and validate the hypothesis that in vivo neurohormonal stimulation of cardiac myocytes directly stimulates cardiomyocyte hypertrophy and ventricular remodeling. The development of cardiac-targeted transgenesis in mice, using promoter constructs employing cardiomyocyte-specific genes to drive cardiomyocyte-directed transgene expression (36–38), provided experimental alternatives (Fig. 3). One idea was to overexpress neurohormone receptors in mouse hearts, which was anticipated to constitutively increase signaling through that neurohormone signaling pathway in the absence of the neurohormones themselves. Indeed, transgenic receptor overexpression proved useful for exploring the qualitative and quantitative consequences of myocardial β-adrenergic receptor signaling (39–42). However, receptor expression met with variable success for prohypertrophic neurohormones (43–46), possibly because G-protein coupled receptor (GPCR) signaling is tightly regulated: chronic receptor activity induces functional receptor uncoupling and physical receptor downregulation in a negative feedback loop that ultimately suppressed the signaling pathways being experimentally activated (reviewed in Ref. 47). The solution to uncovering direct neurohormonal effects on the heart without experimental confounders was to circumvent both the prohypertrophic neurohormones and their individual GPCR receptors through transgenic cardiac myocyte-targeted expression of the common effector for prohypertrophic neurohormones and their respective receptors [i.e., epinephrine (α1-AR), angiotensin II (AT1R) and endothelin (ET) receptors], the Gq signaling protein (Fig. 3). The expectation was that overexpression of the alpha subunit of heterotrimeric Gq (Gαq), which is the component of the heterotrimer that binds both GPCRs and their common immediate downstream effector phospholipase C (48), would tonically activate signaling downstream of all three receptors without evoking ligand-mediated receptor desensitization or downregulation. The Dorn laboratory was the first to do this, and the result was what has been described as “pressure overload hypertrophy without pressure overload” (49, 50). Despite normal left ventricular systolic pressure, cardiomyocyte Gq overexpression recapitulated left ventricular wall thickening, the increase in cardiac myocyte cross-sectional area, β-AR desensitization, and reexpression of fetal cardiac genes that are hallmarks of pressure overload hypertrophy (49, 51–53). Together, these findings provided strong support for the notion that Gq signaling through myocardial neurohormone receptors is sufficient to stimulate in vivo cardiac hypertrophy in the absence of any abnormal hemodynamic load or stress. Subsequent studies employing a Gq inhibitory peptide (54) or cardiac-directed ablation of the genes encoding Gαq and functionally overlapping Gα11 (55) revealed suppression of pressure overload hypertrophy when cardiomyocyte Gq signaling was interrupted. Thus, Gq signaling was shown to be both necessary and sufficient for pressure-overload cardiac hypertrophy in mice.

Figure 3.

Experimental approaches used for in vivo testing of the neurohormonal theory of cardiac hypertrophy/heart failure.

The natural history of cardiac hypertrophy when the inciting hemodynamic stress is not relieved is progression to dilated cardiomyopathy with overt heart failure, called “decompensation.” In vivo Gq manipulation mechanistically linked “adaptive” cardiac hypertrophy to Gq-mediated neurohormone signaling, but did not reveal how or why this mechanism for functional compensation ultimately fails. In biomechanical terms of the Laplace equation: Why does the increase in ventricular wall thickness (h) that normalizes wall stress (σ) in the face of pressure overload (p) ultimately decompensate? Because development of transgenic Gq-mediated cardiac hypertrophy was a cellular event wherein enlargement and transcriptional reprogramming of individual cardiomyocytes collectively produced hypertrophic ventricular remodeling, we asked whether maladaptive ventricular dilation might likewise have its mechanistic foundations within individual cardiomyocytes. The idea that adaptive ventricular hypertrophy predisposes to maladaptive dilated cardiomyopathy was strengthened by the observation that a modest left ventricular pressure stimulus that would not normally cause heart failure provoked dilated cardiomyopathy in Gq transgenic mice (56). Moreover, a human GNAQ gene polymorphism that increases expression of Gαq was linked to accelerated mortality among African American patients with heart failure (57). These correlative studies implied detrimental synergy between hemodynamic stress and Gq-mediated neurohormone signaling.

The mechanism for maladaptive ventricular dilation in Gq mice was uncovered in Gq transgenic mice with greater Gq signaling activity than the original “compensated hypertrophy” Gq mouse line (58). These dual heterozygous Gq mice (carrying two independent Gq transgenes on different chromosomal loci) developed spontaneous adverse cardiac remodeling, i.e., left ventricular chamber dilation and wall thinning in the absence of any abnormal extracardiac hemodynamic stress. The underlying cellular mechanism mediating the transition to decompensated dilated cardiomyopathy (and also for a unique peripartum cardiomyopathy that afflicted Gq transgenic dames) was identified as apoptotic cardiac myocyte drop-out linked to markedly increased expression of proapoptotic factors (59). Likewise, a constitutively active Gαq mutant in unloaded cultured neonatal rat cardiac myocytes provoked apoptosis (58). Induction of proapoptotic genes was not a nonspecific feature of all forms of hypertrophy, but seemed to be specifically evoked by pathological hypertrophy signaling through the neurohormone/Gq signaling pathway (59, 60). Thus, the very process of neurohormone-stimulated cardiac myocyte hypertrophy, which can initially compensate for increased ventricular loading via the Laplace mechanism, carries within it changes in gene expression that culminates in programmed cardiac myocyte death, maladaptive ventricular remodeling and overt heart failure.

Collectively, these results provided mechanistic support for clinical trials of neurohormone antagonists in heart failure. Ironically, the major clinical trials demonstrating therapeutic efficacy of neurohormonal blockade were largely complete before neurohormonal mechanisms leading to cardiac hypertropy and failure were understood (Fig. 4). The first proof of concept study of neurohormonal blockade in heart failure used the α1-AR blocker prazocin (a vasodilator) to unload the heart. As reported in 1980 by Colucci et al. (61), prazocin-treated patients with severe heart failure showed initial improvement in objective metrics for left ventricular performance as well as symptoms. However, the benefits diminished over time, presumably because of increased activity of the renin-angiotensin system (RAS). Indeed, interrupting the RAS pathway with angiotensin converting enzyme (ACE) inhibitors captopril or enalapril was subsequently shown to be both symptom-relieving and life-prolonging as stand-alone therapy for heart failure of various causes (62–65). Perhaps more surprising were observations that β-AR antagonists (“β-blockers”) used in combination with ACE inhibitors further reduced hospitalizations and prolonged survival in chronic heart failure (66–68). By contrast, endothelin A receptor blockade with either darusentan or bosentan added to conventional therapy with ACE inhibitors and β-blockers showed no benefit for cardiac remodeling, heart failure symptoms or outcomes (69, 70). Thus, interrupting AT1 and β-AR signaling were proven to be strikingly beneficial in heart failure and their administration is the standard of modern medical care.

Figure 4.

Approximate timeline of basic and clinical studies testing neurohormone effects in cardiomyocyte hypertrophy and clinical heart failure. Key references are provided, color-coded by neurohormone system. Struck-through references indicate a negative study.

CARDIAC MALADAPTATION, NEUROHORMONES, AND MITOCHONDRIA

The sections above describe the convoluted conceptual path from early recognition that neurohormones regulate acute myocardial function via the Gs-mediated fight-or-flight response to a scientific understanding that neurohormonal responses determine chronic cardiac adaptation via Gq-mediated transcriptional programs for adaptive cardiomyocyte hypertrophy and maladaptive programmed cardiomyocyte death. The current section delves more deeply into the role of mitochondria as central regulators of cardiomyocyte fate in myocardial hypertrophy and the transition to dilated cardiomyopathy.

Hearts have the highest metabolic rate and greatest mitochondrial density of any mammalian organ. It has been calculated that an average human heart, which is ∼40% mitochondria by mass, will generate and consume ∼5kg of ATP in a single minute (71, 72). The reliance of myocardium upon mitochondrial ATP suggests that mitochondrial fitness can be an important determinant of cardiac health and integrated cardiovascular function, both under normal conditions and when the heart is damaged, diseased or chronically overloaded. Thus, it is not surprising that metabolic remodeling, wherein the normal adult myocardial preference for fatty acids as mitochondrial substrates reverts to a fetal myocardium-like preference for carbohydrates, is recognized in many myocardial disease (73–76).

Mitochondria are essential to cardiac pump functioning and stress-adaptation, but are unique in that they can either sustain or terminate the life of cells within which they reside. Although mitochondrial metabolism creates life-sustaining ATP via oxidative phosphorylation, mitochondria are also the “gatekeepers” for programmed cell death via apoptotic and nonapoptotic pathways (77–80). The roles played by mitochondria in apoptosis and programmed necrosis, and the respective roles of caspase signaling and activation of the mitochondrial permeability transition pore (MPTP) in these programmed cell death pathways, have been extensively reviewed elsewhere (81–83) and so will not be reproduced here. Rather, we address the question of how the transformation of cardiac myocyte mitochondria from cell-sustaining ATP generators into cell-killing engines of programmed death contributes to cardiac maladaptation.

Transcriptional and epitranscriptional reprogramming during pathological (as opposed to exercise-induced or physiological) (17) cardiac hypertrophy was initially recognized primarily for reexpression of embryonic heart genes, the so-called “fetal gene expression program” (84–90). However, microarray profiling of mRNAs differentially expressed in hypertrophied/compensated (i.e., nonfailing) Gq hearts revealed abnormally high levels of multiple transcripts linked to apoptosis (60). Of particular interest was Bnip3l or Nix (91), a Bcl2 family protein originally identified by Greenberg as proapoptotic (92). Contextually, Nix was upregulated in Gq and pressure-overload hypertrophies, whereas closely related proapoptotic Bnip3 was upregulated after ischemic myocardial injury (93). Cardiomyocyte-targeted transgenic Nix overexpression proved sufficient to induce cardiomyocyte apoptosis, provoking adverse ventricular remodeling and dilated cardiomyopathy (91, 94). Moreover, germ-line or cardiomyocyte-specific genetic Nix ablation suppressed cardiomyocyte apoptosis and attenuated ventricular dilation in Gq-mediated and pressure overload hypertrophy (95); Bnip3 plays an analogous role mediating cardiomyocyte apoptosis after postischemic cardiac injury (96). Together, these results showed that mitochondrial death protein Nix is transcriptionally induced during compensated hypertrophy, and revealed that Nix is both necessary and sufficient for maladaptive hypertrophy remodeling leading to dilated cardiomyopathy.

Conventional apoptosis is mediated by a well-characterized sequence of events: proapoptotic Bcl2-family proteins form macromolecular pores in mitochondrial outer membranes, permitting release of sequestered cytochrome C (mw 12,000 Da) into the cytosol. Extramitochondrial cytochrome C interacts with cytosolic Apaf-1 to form the multimeric apoptosome, which binds and activates caspase 9 in a ternary complex (97–100). Both Nix and Bnip3 localize to mitochondria and interact with proapoptotic Bax to initiate “intrinsic” pathway mitochondrial apoptosis mediated by the caspase cascade (92, 101) (Fig. 5).

Figure 5.

Hypertrophy-associated programmed death pathways induced by Gq-coupled neurohormone stimulation. Signaling events connecting a prohypertrophic, Gq-coupled receptor to apoptotic and necrotic programmed cell death are shown as described in the text. Transcriptional upregulation of Nix is not specifically shown, but is part of the hypertrophy-associated change in gene expression profile. cyt C, cytochrome C; IP3R, IP3 receptor; MCU, mitochondrial calcium uniporter; MPTP, mitochondrial permeability transition pore; PKC, protein kinase C; PLC, phospholipase C. Nix ribbon diagram was produces using swissmodel.expasy.org; red is Nix carboxy terminal transmembrane domain. Red box is expanded in Fig. 7D.

It is worth considering why potent proapoptotic factors such as Nix and Bnip3 are expressed in terminally differentiated cardiac myocytes that, if lost, cannot readily be replaced. One possibility is that apoptosis, which plays a central role in determining cardiac chamber and valve structure in developing embryonic hearts (102), is simply part of a fetal transcriptional program that is reexpressed in myocardial diseases. Alternately, Nix may have a necessary physiological function, independent of apoptosis. Indeed, although the majority of Nix localizes to mitochondrial outer membranes where it is positioned to activate caspase-dependent apoptosis, a substantial fraction of Nix localizes instead to endo/sarcoplasmic (ER/SR) reticulum where it can, through poorly understood mechanisms, activate calcium uptake and increase ER/SR calcium levels (103, 104) (Fig. 5). Mitochondria are physically associated with ER/SR and can sense ER/SR calcium release, which in the case of cardiomyocyte SR is cyclic, regulated by Gs-coupled β1-AR (105, 106), and is a primary determinant of cardiomyocyte contraction and cardiac failure (107). An increase in SR calcium stores such as that induced by Nix can therefore sensitize mitochondria to SR calcium crosstalk (108, 109). SR calcium release increases with the frequency of cardiomyocyte contraction (increased heart rate) provokes enhanced mitochondrial calcium uptake that accelerates calcium-sensitive mitochondrial metabolic pathways, resulting in accelerated ATP production (110, 111). In this way, Nix expression might help functionally compensate for greater demands on the cardiac pump. However, this same mitochondria-SR calcium crosstalk mechanism can, if of sufficient magnitude, activate the mitochondrial permeability transition pore (MPTP) leading to permeabilization of outer mitochondrial membranes and a catastrophic loss of mitochondrial structural integrity that halts or reverses ATP production (112) (Fig. 5). Nix is unusual in that it can activate both intrinsic caspase-dependent apoptosis and MPTP-dependent “programmed necrosis” pathways in mouse hearts (104, 113). Consequently, therapeutic cardiomyocyte salvage in neurohormone-dependent pressure overload hypertrophy aimed at preventing the transition to dilated cardiomyopathy and heart failure will likely require interrupting pathways leading to both mitochondria-mediated apoptosis and MPTP-mediated programmed cell death.

MITOCHONDRIAL DYNAMICS PROTEINS AND CARDIAC MALADAPTATION

Accumulated data support an important role for modulated mitochondrial metabolism and mitochondria-mediated programmed cell death in neurohormone-stimulated responses to cardiac injury and chronic hemodynamic stress. In addition to their unique dual roles in cell-sustaining ATP production and cell-terminating apoptosis/MPTP-mediated necrosis, mitochondria are exceptional in their ability to structurally remodel and relocate, referred to as “mitochondrial dynamism.” Dynamic behavior of mitochondria has been appreciated for over a century (114), but is highly contextual. For example, cultured fibroblasts are a preferred experimental system for studying mitochondrial fusion and fission because fibroblast mitochondria exist as an interconnected collective that undergoes almost continuous structural remodeling via these two dynamic processes (115, 116). On the other hand, fibroblasts are a poor platform to study mitochondrial motility because only a few fibroblast mitochondria undergo directed subcellular transport at any given time. Thus, mitochondrial motility is more readily assessed in neuronal processes wherein, under basal conditions, up to 30% of mitochondria are undergoing active transport (116–119).

Compared to fibroblasts and neurons, cardiomyocyte mitochondria appear relatively hypodynamic (120, 121). Moreover, cardiomyocyte mitochondria exist as many physically distinct organelles interspersed between myofilaments rather than as a filamentous network. While mitochondrial fission and fusion can certainly play roles in heart disease, as reviewed elsewhere (122–124), physically interconnected mitochondrial networks that undergo fusion/fission-mediated physical remodeling are not present in cardiac myocytes. Rather, cardiac myocyte mitochondria exhibit transient physical connectivity referred to as “kiss and run,” and interorganelle communications which, instead of being mediated by complete organelle fusion, involve reversable formation of tunneling nanotubes (125–127) and possibly other mechanisms (128, 129). Furthermore, mitochondrial turnover via fusion and fission is slower in cardiac myocytes than in other cell types (130). Mitochondrial motility is even less important to myocardial homeostasis than fusion and fission because the tightly organized intracellular architecture of cardiac myocytes restrains subcellular organelle transport. Nevertheless, the principal effectors of mitochondrial dynamics, including fusion proteins mitofusins (MFN) 1 and 2 and optic atrophy 1 (OPA1), and the fission protein dynamin-related protein 1 (DRP1), are expressed at high levels in mammalian myocardium (131–134). Moreover, cardiomyocyte-specific ablation of the genes encoding each of these mitochondrial fusion/fission proteins induces striking pathological cardiac phenotypes, revealing the necessity for mitochondrial fusion and fission proteins during both cardiac development and normal adult heart functioning (130, 135–140) (Fig. 6). The inference was that mitochondrial dynamics factors might be functioning in a noncanonical manner in the heart, i.e., that they are functioning in additional ways than just mediating mitochondrial dynamism (141).

Figure 6.

Schematic depiction of cardiac phenotypes provoked by cardiomyocyte-targeted genetic mitofusin manipulation. Arrow represents developmental stage.

The prototypical noncanonical function for a mitochondrial dynamics protein is mitofusin (MFN) 2 tethering of mitochondria to endo/sarcoplasmic reticulum. As introduced, physical approximation of mitochondria to ER/SR facilitates calcium transport from reticular structures to neighboring mitochondria, thereby modulating both mitochondrial metabolism and MPTP-mediated cell death (142, 143) (reviewed in Refs. 108, 109, 144). Indeed, mitochondrial calcium uptake through the uniporter (MCU), which requires at least micromolar concentrations of calcium that are not normally achieved in cytoplasm (145, 146), is thought to require close physical connections between mitochondria and ER/SR that form privileged mitochondrial-reticular microdomains through which calcium can transfer between organelles without diffusing into the cytoplasm (147, 148). Multimeric MFN2 is embedded within outer mitochondrial membranes from which it extends out into cytosolic space to dimerize in trans with mitofusins (1 or 2) on adjacent mitochondria (Fig. 7); this is its canonical function in mitochondrial-mitochondrial tethering and is required for subsequent outer membrane fusion (149–151). However, a minor fraction of MFN2 localizes to ER/SR, enabling mitochondrial-ER/SR tethering (152) that helps form calcium microdomains for trans-organelle calcium signaling (153–155). An evaluation of the consequences of cardiac-targeted MFN2 gene ablation on cardiomyocyte SR-mitochondria physical approximation and trans-organelle calcium signaling supported the concept that MFN2 promotes functional microdomains which aid mitochondrial sensing of SR calcium release (143). However, in this study MFN2 ablation produced only a transient delay in the mitochondrial reaction to SR calcium release prompted by increased work. The conclusion was that MFN2 tethering of mitochondria to SR contributes to the bioenergetic feedback response during physiological stress, but this aspect of MFN2 biology does not readily explain the development of severe cardiac phenotypes in mitofusin-deficient embryonic and adult hearts.

Figure 7.

Mitofusin 2 hypothetical active structure and putative multimeric structures. A: human MFN2 monomer in extended conformation with labelled structural domains. TM is transmembrane domain. B: comparison with Gαq shows a similar GTPase domain but absence of dynamin-like core and TM structures present in MFN2. C, left: putative MFN2 cis-homodimer, envisioned to assemble into dynamin-like rosettes on organelle membranes. D: cross section of how two MFN2 rosettes might interact in trans across cytoplasmic space tethering outer mitochondrial membranes (OMM) to sarcoplasmic reticulum (SR). This panel is greatly expanded view of red box in Fig. 5.

An important clue to alternate roles played by mitochondrial dynamics factors in vertebrate and invertebrate myocardium was the discovery that, under defined conditions, MFN2 is a central effector of the PINK1-Parkin pathway of mitophagic mitochondrial quality control (156). PINK1 is a kinase imported into, and then immediately degraded by, healthy mitochondria. This kinase is stabilized only in damaged mitochondria, wherein it can initiate mitophagy signaling through the PINK-Parkin pathway. Parkin is an E3 ubiquitin ligase activated by PINK1 in damaged mitochondria (157). When phosphorylated by mitochondrial PINK1, MFN2 exchanges an ability to interact with MFN1 and MFN2 on other mitochondria (e.g., during mitochondrial fusion) for an ability to bind cytosolic Parkin, thus initiating mitophagy (158–160). The consequence of Parkin recruitment to mitochondria by MFN2 and other “Parkin receptors” is polyubiquitination of a hundred or more outer mitochondrial membrane proteins, which attracts autophagosomes that engulf the damaged organelle and transport it to lysosomes for decomposition (157). Thus, unphosphorylated MFN2 actively promotes mitochondrial fusion, but is inactive for mitophagy (and vice versa).

The discovery of MFN2 functioning in mitophagy and the delineation of a mechanism by which it regulates fusion versus Parkin binding suggested an experimental means for distinguishing the functional impact of MFN2-mediated mitochondrial fusion versus mitophagy in mouse hearts in vivo: Cardiomyocyte-directed expression of MFN2 mutants engineered either to mimic or prevent PINK1-mediated phosphorylation of fusion/mitophagy switch sites. Conventional wisdom was that PINK1-Parkin mitophagy was the dominant mechanism for mitochondrial quality control, wherein individual damaged mitochondria initiate a pathway culminating in their selective suicide (157). If MFN2 was an essential intermediate between PINK1 and Parkin in mitophagy, then suppression of MFN2-mediated mitochondrial Parkin recruitment should interrupt mitophagy, resulting in accumulation of abnormal and damaged mitochondria due to defective quality control at the organelle level. However, this is not what was observed. Rather, suppression of MFN2-Parkin binding elicited a perinatal developmental defect characterized by cardiomyocyte retention of fetal-type mitochondria and failure to replace them with adult-type mitochondria, as is normal within a few days to weeks after birth (161). Failure to transition from fetal heart carbohydrate-based mitochondrial metabolism to postnatal fatty-acid based mitochondrial metabolism was incompatible with life. These studies uncovered a previously unappreciated function for PINK1-Parkin mediated mitophagy as the process mediating the normal developmentally regulated generalized turnover of mitochondria that enables the transition from fetal to adult myocardial metabolism. Indeed, at least for adult mouse myocardium, the PINK1-Parkin mitophagy pathway does not appear to play a major role in homeostatic mitochondrial quality control, but instead is a stress-induced pathway activated after cardiac injury (156, 162–164).

NONCANONICAL FUNCTIONING OF MITOCHONDRIAL FUSION PROTEINS IN EXTRACARDIAC DISEASES

The results described in the previous section reveal unexpected mechanisms by which mitochondria contribute to myocardial disease. It seems intuitively obvious that mitochondrial metabolic dysfunction would compromise pumping efficiency of this most mitochondrial-rich organ. Yet, what has been observed are mitochondrial pathways to cardiac disease, including programmed cardiomyocyte death and interrupted mitophagy, in which altered ATP production is a secondary, collateral or unrelated event. Although investigating context-driven noncanonical functioning by mitochondrial dynamics factors in the heart is still in its early stages, the notion that hypodynamic myocardial mitochondria may be uniquely suited to uncovering functioning of mitochondrial dynamics proteins beyond fusion and fission gives rise to the concept that organ-specific mitochondrial dynamic physiology can exist beyond the heart.

It seems noteworthy that no clinical cardiac-specific diseases are yet known to be caused by mutations of MFN1, MFN2, OPA1, or DRP1. Rather, the cardiac-disease associations are almost entirely in animal models using gene knockout techniques. Clinically, MFN1- linked human disease has not been described. OPA1 mutations are most recognized as causing autosomal dominant optic atrophy (165–168) (although a homozygous missense OPA1 mutation is reported to cause mitochondrial DNA depletion syndrome with lethal infantile encephalopathy and hypertrophic cardiomyopathy in two siblings) (169). De novo heterozygous mutations in the DNM1 gene that encodes the mitochondrial fission protein DRP1 were reported as a very rare cause of Developmental and Epileptic Encephalopathy type 31 (170), but are not an identified cause of common genetic diseases. Indeed, the only relatively common genetic condition linked to mutations of a mitochondrial dynamics protein is the predominantly pediatric peripheral neuropathy, Charcot-Marie-Tooth disease type 2 A (CMT2A). CMT2A is caused by ∼100 different mutations of MFN2 (171, 172). MFN2 mutations have also been identified as a cause of rare genetic lipodystrophy and variant forms of dominant optic atrophy (173, 174). Excellent reviews of MFN2 and CMT2A are available for the interested reader (175–178). Here CMT2A is discussed as the prototype disease incontrovertibly caused by a disturbance of mitochondrial dynamism.

Because CMT2A is caused by mutations in a mitofusin, it is widely considered to be a disease of interrupted or incomplete mitochondrial fusion (179–181). However, CMT2A characteristically has the greatest pathological impact on the distal portions of nerves innervating the upper and lower extremities, i.e., the sections of the longest nerves in the body that are farthest from the neuronal cell bodies, or soma. Because mitochondrial fusion is a ubiquitous process that is seemingly essential to every cell type and tissue, it is curious that a mutational fusion defect would have such a specific pathophysiological impact. Indeed, systemic interruption of mitochondrial fusion in mice by ablating either the Mfn1 or Mfn2 genes was embryonic lethal (182) and conditional ablation of Mfn2 in mouse hearts, livers and brains evokes dysfunction of the respective organs (183–187). Exclusivity of clinical disease with mutations of MFN2 (vs. MFN1) and specificity of disease within distal long peripheral neurons of CMT2A patients with MFN2 mutations suggested to some that neurons might rely more than other cell types on MFN2 versus MFN1 for mitochondrial fusion (188, 189). Support for the idea that MFN1 cannot functionally compensate for mutant MFN2 specifically in neurons was provided by reports that cultured dermal fibroblasts of patients with CMT2A did not exhibit the expected fragmented mitochondria (190, 191). However, absence of mitochondrial pathology in cultured fibroblasts of patients with CMT2A was recently discovered to be a function of the fibroblast culture conditions, and not attributable to any distinct resistance to MFN2 CMT2A mutations in nonneuronal cells (116). An alternate hypothesis that could explain unique neuronal sensitivity of long peripheral neurons to MFN2 mutations is that, as in the heart, MFN2 has a noncanonical function with specific biological relevance to long neurons. Accumulating data suggest that this extra function is regulation of mitochondria motility.

Synaptic release and reuptake of neurohormones is an energy intensive process requiring mitochondrial ATP that, because ATP is chemically unstable at physiological temperature and pH, must be generated locally by mitochondria located near neuronal synapses (192, 193). Since mitochondrial biogenesis must occur near the neuronal nucleus, wherein over 99% of genes encoding mitochondrial proteins are encoded and transcribed, delivery of fresh mitochondria to distal neuro-muscular or neuroneuronal junctions requires that mitochondria be transported from neuronal soma and through axons. For this reason, abnormal mitochondrial motility compromises neuronal signaling, repair and regeneration and is a commonly described feature of multiple neurodegenerative diseases (180, 194).

The best understood mechanism of mitochondrial motility involves calcium-regulated binding of outer mitochondrial membrane MIRO proteins to Trafficking kinesin-binding (TRAK/Milton) 1 and 2 adaptor proteins that couple mitochondria and other cellular cargo to dynein and kinesin-1 molecular motors for directed transport along microtubules (195–200). Severe mitochondrial dysmotility is a hallmark of neuronal die-back in CMT2A (201, 202). Baloh first suggested a mechanistic link between CMT2A MFN2 mutations and mitochondrial transport by showing that 1) MFN2 normally binds MIRO1 and MIRO2; and 2) CMT2A MFN2 mutants exhibit defective mitochondrial transport (183, 203, 204). Collectively, the observational data support a regulatory role for MFN2 in mitochondrial motility, perhaps by priming the critical MIRO-TRAK interaction (205). Confirming this hypothesis will require more fully delineating the key determinants and biological consequences of MFN2-MIRO interactions.

The most compelling evidence supporting both a central pathological role for mitochondrial dysmotility in CMT2A caused by MFN2 mutations and a key role for MFN2 as a regulator of axonal mitochondrial trafficking may be the ability of small molecular activators of mitofusins to reverse mitochondrial dysmotility and neuromuscular degeneration in preclinical CMT2A models (202). Reprogrammed CMT2A patient motor neurons, cultured CMT2A mouse neurons, and neuronal axons of CMT2A mouse sciatic nerves all exhibit nearly complete loss of mitochondrial trafficking through neuronal processes (206). To the extent studied, mitochondrial dysmotility in CMT2A appears agnostic to different MFN2 CMT2A mutations (206). Importantly, pharmacological mitofusin activation with either fusogenic peptides (207) or structurally dissimilar small molecules (206, 208–210) has normalized CMT2A mitochondrial dysmotility. In vivo, mitofusin activation reversed functional and neuroelectrophysiological metrics of neuromuscular degeneration and promoted neuronal regrowth by restoring mitochondrial occupancy at axonal growth buds and neuromuscular junctions (206). Taken together, these results implicate MFN2 in neuronal mitochondrial transport/delivery, thereby further expanding its recognized biological functions.

MFN2 multifunctioning seems attributable to its ability to pair with protein partners depending upon cellular and pathophysiological context (Fig. 8). Given the central importance of PINK1-MFN2-Parkin mitophagy to cardiac myocytes, versus mitochondrial transport for neurons, this concept suggests that a hypothetical MFN2 mutation impairing mitophagy and sparing mitochondrial motility would evoke cardiomyopathy instead of neuropathy. Likewise, a hypothetical MFN2 mutation that suppresses mitochondrial fusion without impairing motility might produce either milder neuropathy than is typical for CMT2A or a different spectrum of organ involvement. Because the focus on MFN2 mutations has been almost entirely on their suppression of mitochondrial fusion, significant effort will be required to test these notions.

Figure 8.

Schematic depiction of the relationship between MFN2 protein partnering, function, and disease manifestation when disrupted.

THE “BIG PICTURE”

In this overview, I have employed two seemingly different cellular processes, neurohormonal signaling and mitochondrial dynamics, as examples of how biological pathways can be inextricably but differentially linked in cardiac and neurological disease. In heart disease, neurohormonal signaling is an important proximal effector of adaptation to hemodynamic stress, mediated in large part by induction of transcriptional programs leading to cardiomyocyte hypertrophy and apoptosis. Furthermore, perinatal cardiac development seems uniquely dependent upon mitochondrial replacement mediated by MFN2-Parkin for metabolic adaptation to the extrauterine milieu. Thus, increased signaling through neurohormone receptor-effectors in response to mechanical injury or stress is a proximate cause of pathological changes to biological pathways regulating replacement of cardiomyocyte mitochondria by mitophagy and inducing programmed cardiomyocyte death. In neurological diseases caused by primary damaging mutations or secondary dysregulation of mitochondrial dynamics factors the same connections exist, but the pathological cycle is reversed. As in CMT2A caused by MFN2 mutations, suppression of mitochondrial transport along neuronal processes interrupts neuronal signaling because cytotoxic senescent or damaged mitochondria accumulate at neuronal synapses, thereby impairing neurohormone release/reuptake and ultimately provoking neuronal die-back. A more complete understanding of how cardiac myocytes and neurons rely in different ways upon ubiquitous protein signaling factors, such as heterotrimeric G-proteins and MFN2 dynamin GTPases reviewed here, may provide insights into organ system selectivity in genetic and nongenetic disease.

GRANTS

This work was supported by National Institutes of Heart/National Heart, Lung, and Blood Institute (NIH/NHLBI) R35 HL135736 and NIH/National Institute of Neurological Disorders and Stroke (NINDS) R42 NS115184 and R42NS113642.

DISCLOSURES

G.W.D. is an inventor on patents that cover the use of small molecule mitofusin activators to treat neurodegenerative diseases and is the founder of Mitochondria in Motion, Inc., a Saint Louis based biotech R&D company that aims to enhance mitochondrial trafficking and fitness in neurodegenerative diseases.

AUTHOR CONTRIBUTIONS

G.W.D. prepared figures; drafted manuscript; edited and revised manuscript; approved final version of manuscript.