Posttranslational modifications of mitochondrial fission and fusion proteins in cardiac physiology and pathophysiology

Abstract

Mitochondrial fragmentation frequently occurs in chronic pathological conditions as seen in various human diseases. In fact, abnormal mitochondrial morphology and mitochondrial dysfunction are hallmarks of heart failure (HF) in both human patients and HF animal models. A link between mitochondrial fragmentation and cardiac pathologies has been widely proposed, but the physiological relevance of mitochondrial fission and fusion in the heart is still unclear. Recent studies have increasingly shown that posttranslational modifications (PTMs) of fission and fusion proteins are capable of directly modulating the stability, localization, and/or activity of these proteins. These PTMs include phosphorylation, acetylation, ubiquitination, conjugation of small ubiquitin-like modifier proteins, O-linked-N-acetyl-glucosamine glycosylation, and proteolysis. Thus, understanding the PTMs of fission and fusion proteins may allow us to understand the complexities that determine the balance of mitochondrial fission and fusion as well as mitochondrial function in various cell types and organs including cardiomyocytes and the heart. In this review, we summarize present knowledge regarding the function and regulation of mitochondrial fission and fusion in cardiomyocytes, specifically focusing on the PTMs of each mitochondrial fission/fusion protein. We also discuss the molecular mechanisms underlying abnormal mitochondrial morphology in HF and their contributions to the development of cardiac diseases, highlighting the crucial roles of PTMs of mitochondrial fission and fusion proteins. Finally, we discuss the future potential of manipulating PTMs of fission and fusion proteins as a therapeutic strategy for preventing and/or treating HF.

INTRODUCTION

Mitochondria are highly dynamic organelles with the ability to divide (fission), combine (fusion), and move along the cytoskeleton to interact with other organelles, which determines cell-specific mitochondrial morphology, intracellular distribution, and function in the mammalian cells (54, 109, 197, 202). These mitochondrial dynamics are mainly controlled by the balance between fission and fusion, both processes regulated by the relative activities of mitochondrial fission and fusion machineries. The core components of the mammalian machineries are members of the dynamin family of large GTPases that can bind and hydrolyze guanosine triphosphate (GTP). A GTPase, dynamin-related protein 1 (Drp1, also called DLP1), and its receptors [mitochondrial fission factor (Mff; 65, 138) and mitochondrial dynamics proteins of 49 and 51 kDa (MiD49 and MiD51, respectively; 141, 210)] dictate mitochondrial fission, whereas the GTPases mitofusin 1 and 2 (Mfn1 and Mfn2, respectively) and optic atrophy protein-1 (OPA1) mediate mitochondrial fusion (18, 150, 192).

The relative activities of fission and fusion proteins are likely determined by several factors: 1) the total expression levels of these proteins, controlled by transcription, translation, protein stability, and degradation; 2) the expression levels of these proteins at the mitochondria [e.g., Drp1 can be recruited from the cytosol to the outer mitochondria membrane (OMM) in response to cellular stimuli]; and 3) the overall activities of the GTPases, which likely depend on GTP availability, inter/intra-protein-protein interactions, and/or protein expression levels present at the mitochondria. These three factors interdependently modulate the balance of fission and fusion (Fig. 1). For instance, under physiological conditions, the expression levels of these proteins may influence their localization and GTPase activity at the mitochondria, whereas the localizations of the proteins affect their activities at the mitochondria. Thus, this interdependence points to overall protein activities as the integrative regulators of mitochondrial morphology. Recent studies have increasingly shown that posttranslational modifications (PTMs) of fission and fusion proteins are capable of directly modulating their stability, localization, and/or activity (19, 55, 126, 136). Therefore, understanding the PTMs of fission and fusion proteins may allow us to understand the complexities that determine the balance of fission and fusion that ultimately dictate mitochondrial morphology in various cell types and organs including cardiomyocytes (CMs) and the heart (Fig. 2, A–D).

Fig. 1.

Regulation of mitochondrial morphology by posttranslational modifications (PTMs) of mitochondrial fission and fusion proteins. Schematic model of the three main regulators (protein expression, localization at the mitochondria, and overall GTPase activity) of mitochondrial fission and fusion, which interdependently modulate the balance of fission and fusion.

Fig. 2.

Subpopulations of mitochondria in adult cardiomyocytes. A: illustration of a single adult cardiomyocyte and diagram of different subpopulations of mitochondria. B: illustration of longitudinal section of the area indicated in A including subsarcolemmal mitochondria. C: illustration of longitudinal section of the area indicated in A including perinuclear mitochondria. D: illustration of longitudinal section of the area indicated in A including interfibrillar mitochondria. Note that interfibrillar mitochondria face the sarcoplasmic reticulum (SR). T tubule, transverse tubule. E: transmission electron microscope image of the longitudinal section of a papillary muscle obtained from an adult rabbit ventricle. Arrows show the location of subsarcolemmal mitochondria. Scale bars, 5 µm. F: transmission electron microscope image of the short axis of a papillary muscle obtained from an adult rabbit ventricle. Blue area indicates location of the nucleus. Red areas indicate perinuclear mitochondria. Yellow areas indicate subsarcolemmal mitochondria. Scale bar, 5 µm.

On the basis of studies using cell culture systems, mitochondrial dynamics are suggested as an important modulator for bioenergetics, reactive oxygen species (ROS) generation, spatiotemporal dynamics of Ca²⁺ signaling, cell growth, and cell death (54, 109, 197, 202). Since mammalian hearts, which are enriched in mitochondria (≅30% of total volume) mostly from CMs, have a high demand for ATP to constantly pump blood through the body (128), it is reasonable to suggest that mitochondrial fission and fusion are important for regulating ATP production in CMs so that they may acquire the energy necessary for performing their function. Transmission electron microscope images show that adult CMs have highly organized and largely static mitochondria that are densely packed 1) into the intermyofibrillar spaces with one or more mitochondria surrounding each sarcomere (i.e., interfibrillar mitochondria) (Fig. 2, D and E), 2) just beneath the plasma membrane (i.e., subsarcolemmal mitochondria) (Fig. 2, B, E, and F), and 3) at the perinuclear region (i.e., perinuclear mitochondria) (Fig. 2, C and F), which likely restricts mitochondrial movement and formation of networks (8, 133, 152, 190). Thus, the activity of mitochondrial dynamics and their physiological relevance to mitochondrial and cellular function in CMs are still under debate. Emerging evidence using state-of-the-art imaging techniques suggests that cardiac mitochondria (especially interfibrillar mitochondria) form mitochondrial networks inside the myofibrils and may possess at least slow fusion activity under physiological conditions. For instance, several groups, using confocal microscopy and a mitochondrial matrix-targeted photoactivable green fluorescent protein, reported that interfibrillar mitochondria form local networks and undergo OPA1- and Mfn1-dependent fusion events to share mitochondrial matrix content with neighboring mitochondria in both adult skeletal and cardiac muscles (52, 53, 151). However, the green fluorescent protein diffusion/sharing events were slower (≥ 3 min) and less frequent in adult CMs compared with nondifferentiated myoblasts or neonatal CMs (52, 53). Using three-dimensional electron microscopy, Balaban’s group also proposed that interfibrillar mitochondria in both skeletal and cardiac muscles communicate through junctions (68, 69, 148), which allows for rapid electrical coupling though mitochondrial networks (103). However, it is still unknown whether there is continuity in the matrix along all of the conducting elements (68, 69, 148). Moreover, the frequency and timescale of fission events in adult CMs remain unclear since the techniques for detecting mitochondrial fission events in live adult CMs are still experimentally challenging. Importantly, abnormal mitochondrial morphologies, especially small, disorganized interfibrillar mitochondria, are frequently observed in the myocardium of both human patients (43, 95, 154, 169) and animal models of cardiac diseases, such as heart failure (HF; 11, 13, 25, 36, 87, 167, 198). These observations indicate that mitochondrial fragmentation in CMs frequently occurs under chronic pathological conditions as has been observed in other human diseases including neurodegenerative diseases, diabetes, and cancer (5, 26). This suggests a link between mitochondrial fragmentation and cardiac pathologies, although the physiological relevance of mitochondrial fission and fusion in adult CMs is still unclear. Consequently, multiple questions still remain: 1) Is this mitochondrial fragmentation caused by the enhancement of mitochondrial fission activity and/or the inhibition of fusion activity? 2) How are the PTMs of mitochondrial fission and fusion proteins involved in these fission and fusion processes? 3) Is the abnormality of cardiac mitochondrial morphology involved in the development of cardiac pathology or just an adaptation/maladaptation to pathological changes in the heart?

It is conceivable that mitochondrial fusion and fission impact all cell types in the heart, including cardiac fibroblasts and vascular endothelial and smooth muscle cells in addition to CMs. Indeed, the roles and functional relevance of mitochondrial fission and fusion proteins in fibroblasts [e.g., embryonic fibroblasts (179)], vascular smooth muscle cells [e.g., pulmonary artery smooth muscle cells (116)], and endothelial cells [e.g., kidney microvascular endothelial cells (194)] in other organs/tissues have emerged. However, the number of publications investigating these cell types in the heart (i.e., cardiac fibroblasts and smooth muscle and endothelial cells from coronary arteries) is extremely limited compared with those focusing on CMs [see review (189)]. Therefore, in this review, we summarize present knowledge regarding the function and regulation of mitochondrial fission and fusion only in CMs, with a focus on the PTMs of each mitochondrial fission/fusion protein. We also discuss the molecular mechanisms underlying abnormal mitochondrial morphology in CMs and their contribution to the development of cardiac diseases, highlighting the crucial role of PTMs of mitochondrial fission and fusion proteins in CMs.

OVERVIEW: MITOCHONDRIAL MORPHOLOGY AND FISSION AND FUSION PROTEINS IN CARDIOMYOCYTES

As described in the introduction, unlike noncardiac cells, adult CMs have highly organized and globular-shaped mitochondria, which can be categorized into three distinct subpopulations depending on their location and function [see also review (133); Fig. 2]. The majority of these mitochondria are classified as interfibrillar mitochondria, and a smaller population are categorized as subsarcolemmal or perinuclear mitochondria. The interfibrillar mitochondria are aligned adjacent to the myofibrils, with one or two mitochondria lying along each sarcomere [i.e., between the 2 transverse tubules (T tubules); Fig. 2, A, D, and E]. These mitochondria are involved in ATP generation for contractile function and Ca²⁺ signaling. The subsarcolemmal mitochondria are arranged just beneath the sarcolemma and may provide ATP required for ion channel function and cell signaling (Fig. 2, A, B, E, and F). The perinuclear mitochondria form clusters around the nucleus and may provide ATP for transcription and other nuclear activities (Fig. 2, A, C, and F). These specific and compact intracellular mitochondrial arrangements may restrict mitochondrial mobility in adult CMs compared with other cell types, although they still maintain their networks and interaction throughout the myofibrils (see introduction). In the heart and CMs, mitochondrial fission and fusion proteins are highly expressed, including Drp1, Mff, Mfn1, Mfn2, and OPA1, although their stoichiometric ratio may differ during development (e.g., neonatal CMs vs. adult CMs) or under various pathological conditions (e.g., HF; 11, 25, 88; Table 1).

Table 1.

Altered expression levels of mitochondrial fission and fusion proteins in cardiac disease animal models

Reference	Species	Model	Mitochondrial Morphology	Drp1	Mfn1	Mfn2	OPA1
Chen et al. (25)	Rat	9 wk post-MI-induced HF	Fragmented mitochondria	→	→	→	↓
Javadov et al. (88)	Rat	12 wk post-MI-induced HF	Fragmented mitochondria	→	ND	↓	→
Jiang et al. (92)	Rat	8 wk post-MI-induced cardiac injury	ND	↑	→	↓	↓
Littlejohns et al. (110)	Mouse	Cardiac I/R injury in 20–21-wk high-fat diet model	Fragmented mitochondria	↑	→	↑	↓
Cao et al. (15)	Rat	Cardiac injury in 8-wk high-fat diet model	ND	→	↓	→	↓
Maneechote et al. (115)	Rat	Cardiac I/R injury	Swollen mitochondria	↑	ND	→	→

Drp1, dynamin-related protein-1; HF, heart failure; I/R, ischemia-reperfusion; Mfn1 and Mfn2, mitofusin 1 and 2, respectively; MI, myocardial infarction; ND, not determined; OPA1, optic atrophy protein-1; →, remained unchanged; ↑, increased; ↓, decreased.

Initially, the importance of mitochondrial fission and fusion proteins in maintaining cellular functions has been investigated using cultured cell lines and/or nonexcitable cells (54, 90, 107, 150). Recently, the physiological and pathological importance of mitochondrial fission and fusion proteins in humans has emerged from clinical reports showing that individuals who possess a mutation in these proteins exhibit various clinical phenotypes, especially related to excitable cell functions. These include 1) neurological disorders in Drp1 mutations (21, 58, 127, 174, 188), Mfn2 mutations linked to Charcot-Marie-Tooth disease type 2A (61), and OPA1 mutations linked to autosomal dominant optic atrophy (2, 38, 42) and 2) the mixture of neurological and skeletal muscle disorders in OPA1 mutations linked to Behr syndrome (77). Finally, thanks to recent studies using primary adult CMs and knockout (KO) mouse lines (Table 2), many laboratories have started to precisely assess the roles of fission and fusion proteins in cardiac physiology and their contributions to cardiac pathologies. However, there are multiple technical challenges involved in researching mitochondrial morphology and dynamics using adult CMs that still need to be overcome. For instance, the observation of mitochondrial morphology/dynamics (especially fission events) in live CMs under light/confocal microscopy (52, 53) is still challenging, which necessitates the use of electron microscopy under fixed conditions (68, 69, 148). In addition, it is technically difficult to maintain the integrity of cellular membrane structures such as T tubules (Fig. 2D) in cultured adult CMs for the time periods required for genetic manipulations (e.g., 24–48 h; 54, 139, 200). Thus, the utilization of adult CMs/hearts freshly isolated from KO mouse lines may limit the observable information for investigating the roles of PTMs in mitochondrial fission and fusion proteins. Therefore, it is reasonable to extract the information about cardiac-specific PTMs of mitochondrial fission and fusion proteins from various “cell models for adult CMs” that possess a similar structure of signaling pathways, despite the fact that their cellular membrane integrity and mitochondrial morphology/dynamics at resting conditions are different from those in adult CMs. These include H9c2 cardiac myoblasts (25, 91, 177, 200, 204), the HL-1 mouse atrial CM cell line (56, 147, 206), primary neonatal CMs (12, 45, 91, 114, 211), human induced pluripotent stem cell-derived CMs (79), and embryonic stem cells-derived CMs (97; see also the “Notes” columns in Tables 3 and 4).

Table 2.

Effects of in vivo knockout/ablation of the mitochondrial fission and fusion proteins in the heart

Reference	Targeted Protein	Model	Cardiac Phenotype	Mitochondrial Morphology	Mitochondrial Function	Other Notes
Kageyama et al. (94); Song et al. (179)	Drp1	CM-specific (Myh6 nuclear-directed Cre) KO after birth	Neonatal cardiomyopathy (94); modestly enlarged hearts (179)	Connected or enlarged mitochondria (94)	Mitochondrial respiration ↓ (94); Parkin-independent mitophagy signaling ↑ (94)	Died at 1–1.5 wk (94); died by 6 wk (179)
Song et al. (179)	Drp1	Inducible CM-specific (Myh6 MER-Cre-MER) KO (132)	At 6–7 wk after Drp1 deletion, DCM, CM necrosis, cardiac fibrosis, and HF	Elongated and enlarged mitochondria	mPTP opening ↑; Parkin-dependent mitophagy signaling ↑
Ikeda et al. (82)	Drp1	Inducible CM-specific (α-MHC-MER-Cre-MER) homozygous KO	At 4–8 wk after Drp1 deletion, cardiac hypertrophy, CM apoptosis, cardiac fibrosis, and HF	Elongated mitochondria	ATP ↓; mPTP opening ↑; ROS ↑; autophagic flux ↓	Died at 8–13 wk after Drp1 deletion
Ikeda et al. (82); Shirakabe et al. (176)	Drp1	CM-specific (α-MHC) heterozygous KO	Cardiac function at 12 wk →; cardiac hypertrophy at 5 days after TAC ↑; HF at ~4 wk after TAC	Elongated mitochondria; after TAC, enlarged mitochondria at ~24 h and ~4 wk and fragmented mitochondria at 3–5 days	Susceptibility to I/R injury ↑; after 3–5 days of TAC, mitophagy ↑, ATP ↓, and mitochondrial respiration ↓	Homozygous mice: embryonic lethal
Ishihara et al. (85)	Drp1	Muscle-specific (Mck) KO	Neonatal cardiomyopathy	Connected or enlarged mitochondria	Mitochondrial respiration ↓	Died at 1–1.5 wk
Chen et al. (23)	Mff	Conventional KO	At 13–14 wk, DCM, cardiac fibrosis, and HF	Mitochondrial length →; mitochondria number ↓	Mitochondrial respiration ↓; mitophagy ↑	Neuromuscular defects; died at ~13 wk
Papanicolaou et al. (144)	Mfn1	CM-specific (Myh6) KO	Cardiac function →	Fragmented mitochondria	Mitochondrial respiration →; ROS-induced mPTP opening and CM death ↓
Chen et al. (28)	Mfn1	CM-specific (Myh6 nuclear-directed “turbo” Cre) KO	Cardiac function at 6–8 wk →	Mean distance between SR and mitochondria →	During pacing and β-adrenergic stimulation, mitochondrial Ca2+ uptake → and oxidation of NAD(P)H and FADH2 →
Papanicolaou et al. (142)	Mfn2	CM-specific (α-MHC) KO	Modest LV hypertrophy; mild LV systolic dysfunction; recovery after I/R ↑	Pleomorphic and enlarged mitochondria	Mitochondrial respiration →; time to reach Ca2+-induced mPTP opening ↑
Chen et al. (28)	Mfn2	CM-specific (Myh6 nuclear-directed “turbo” Cre) KO	Cardiac function at 6–8 wk →	Enlarged mitochondria; mean distance between SR and mitochondria ↑	During pacing and β-adrenergic stimulation, mitochondrial Ca2+ uptake ↓, oxidation of NAD(P)H and FADH2 ↑, and mitochondrial ROS ↑	No apoptosis
Chen et al. (30)	Mfn1 and Mfn2	Inducible CM-specific (Myh6 MER-Cre-MER) DKO	DCM during 5 wk; HF after 7–8 wk	Fragmented mitochondria	Mitochondrial respiration ↓
Papanicolaou et al. (143)	Mfn1 and Mfn2	CM-specific (Myh6) DKO	At 1 wk, DCM and HF	At 1 wk, spherical and heterogeneously sized mitochondria and swollen mitochondria	ATP →; early mediators of autophagy ↑	Died at 6–16 days
Chen et al. (30); Kasahara et al. (97)	Mfn1 and Mfn2	CM-specific (Nkx2.5, which induces robust gene recombination before E9.5 in CMs) DKO	CM differentiation at E9.5 ↓ (97); cardiac development at E12.5–13.5 ↓ (97)	Fragmented mitochondria in embryonic heart (97)	Mitochondrial functions were normal (97)	Died after E9.5 (30); died after E13.5 (97)
Hall et al. (75)	Mfn1 and Mfn2	Inducible CM-specific (Myh6 MER-Cre-MER) DKO	At ~4 wk after DKO, cardiac function →, I/R injury ↓, and contractile function ↓	Fragmented mitochondria; mean distance between SR and mitochondria ↑	Mitochondrial respiration ↓; time to reach Ca2+-induced mPTP opening ↑; mitochondrial Ca2+ uptake during I/R ↓
Piquereau et al. (153)	OPA1	Heterozygous mutation (329–355del, OPA1+/− )	Cardiac function at 6 mo →; LV hypertrophy after TAC ↑	Enlarged mitochondria; cristae disorganization	Mitochondrial respiration →; time to reach Ca2+-induced mPTP opening ↑
Chen et al. (27)	OPA1	Heterozygous mutation (Q285 Stop, OPA1+/−)	Cardiomyopathy and HF at 12 mo	Disorganized and fragmented mitochondria ↑; cristae structure ↓	Mitochondrial respiration ↓; ATP ↓; ROS ↑	No apoptotic CM death; homozygous mice: embryonic lethal

CM, cardiomyocyte; DCM, dilated cardiomyopathy; DKO, double knockout; Drp1, dynamin-related protein-1; E9.5, embryonic day 9.5; HF, heart failure; I/R, ischemia-reperfusion; KO, knockout; LV, left ventricular; Mck, muscle creatine kinase; MER, modified estrogen receptor; Mff, mitochondrial fission factor; Mfn1 and Mfn2, mitofusin 1 and 2, respectively; α-MHC, α-myosin heavy chain; mPTP, mitochondrial permeability transition pore; Myh6, myosin, heavy polypeptide 6 cardiac muscle-α; Nkx2.5, NK2 homeobox 5; OPA1, optic atrophy protein-1; ROS, reactive oxygen species; SR, sarcoplasmic reticulum; TAC, transverse aortic constriction; →, remained unchanged; ↑, increased; ↓, decreased.

Table 3.

Posttranslational modifications of Drp1 in cardiomyocytes/hearts

Position	Type of Modification	Upstream Molecules	Effect on Mitochondrial Morphology	Notes
K532, K535, K558, K568, K594, K597, K606, and K608 (59)	SUMOylation	SENP5 (99, 175)	DeSUMOylation: enlarged and swollen mitochondria (99); fragmented mitochondria (175)	Figueroa-Romero et al. (59) detected SUMOylation sites in noncardiac cells
T585 and T586	O-GlcNAcylation	ND (O-GlcNAc transferase?)	O-GlcNAcylation: fragmentation (66)
S616	Phosphorylation	CDK1/cyclin B (206); PKCδ (113, 206); ERK1/2 (203); CaMKII (200); ROCK (12); RIP1 (164)	Phosphorylation: fragmentation (113, 203, 206) Dephosphorylation: elongation (206) S616A: fragmentation (91); no significant changes (12)	Xu et al. (200) detected Drp1 phosphorylation at S616 and S637 in adult CMs, but the effect of CaMKII on mitochondrial morphology was analyzed in H9c2 cardiac myoblasts; Lu et al. (113) used the neonatal CMs for detecting Drp1 phosphorylation at S616 but did not provide any direct evidence of PKCδ-dependent Drp1 phosphorylation; Yu et al. (203) detected phosphorylation of Drp1 at S616 using an in vitro kinase assay, and mitochondrial morphological changes were analyzed in H9c2 cells; Zaja et al. (206) used the HL-1 cell line derived from the AT-1 mouse atrial myocyte tumor lineage for detecting Drp1 phosphorylation at S616; Jhun et al. (91) detected Drp1 phosphorylation at S616 in H9c2 cells and heart trusses, but the effect of S616 phosphorylation on mitochondrial morphology was analyzed in H9c2 cells
S637	Phosphorylation	PKA (35); PKD (91); Pim-1 (45); CaN (35, 149, 173, 206)	Phosphorylation: fragmentation (91) Dephosphorylation: fragmentation (149, 173, 206) S637A: no significant changes (91)	Zaja et al. (206) used the HL-1 CM line for detecting Drp1 phosphorylation at S637; Jhun et al. (91) detected Drp1 phosphorylation at S637 in H9c2 cells, neonatal CMs, and heart trusses, but the effect of S637 phosphorylation on mitochondrial morphology was analyzed in H9c2 cells and neonatal CMs; Cribbs and Strack (35) detected S637 phosphorylation in the heart by in vivo isoproterenol injection and exercise, but they did not observe mitochondrial morphology in CMs; Din et al. (45) showed that Pim-1 decreases Drp1 expression and increases S637 phosphorylation, which results in mitochondrial elongation

Information about sites of conjugation of small ubiquitin-like modifier (SUMO) proteins (SUMOylation) was collected from PhosphoSitePlus (80). CaN, Ca²⁺-dependent phosphatase calcineurin; CDK1, cyclin-dependent kinase; CM, cardiomyocyte; Drp1, dynamin-related protein-1; ND, not determined; O-GlcNAc, O-linked-N-acetyl-glucosamine; O-GlycNAcylation, O-linked-N-acetyl-glucosamine glycosylation; Pim-1, proto-oncogene serine/threonine-protein kinase; RIP1, receptor-interacting protein-1; ROCK, Rho-associated protein kinase; SENP5, sentrin/SUMO-specific protease 5.

Table 4.

Posttranslational modifications of mitochondrial fusion proteins

Protein	Type of Modification (Position)	Upstream Molecules	Effect on Mitochondrial Morphology	Detected/Tested in CMs?	Notes
Mfn1	Phosphorylation (T562)	ERK (156)	Phosphorylation: fragmentation (156)	No
Mfn1	Acetylation (K491)	ND (145)	Acetylation-deficient K491R: aggregation (145) Acetylation-mimetic K491Q: elongation (145)	No
Mfn1	Ubiquitination (ND)	MARCH5 (145)		No	Mfn1 levels are maintained by MARCH5-mediated quality control on acetylated Mfn1 (145)
Mfn2	Phosphorylation (S27)	JNK (105)	S27D: fragmentation (105)	No	S27 phosphorylation of Mfn2 increased ubiquitination and degradation of Mfn2 by the E3 ubiquitin ligase Huwe1 (105)
Mfn2	Phosphorylation (S442)	PKA (212)	S442A/S442D: no change (212)	No
Mfn2	Phosphorylation (T111 and S442)	PINK1 (29)	ND	Yes (29)	PINK1-dependent phosphorylation of Mfn2 induces Parkin translocation to the OMM upon membrane depolarization, which subsequently promotes Parkin-mediated ubiquitination of Mfn2 in adult CMs (29)
Mfn1/Mfn2	Ubiquitination (In Mfn2 sequence: K406, K416, K420, K719, K720, K730, and K737)	Parkin (29, 67, 70, 117, 185); USP30 (205)	Ubiquitination: reduced mitochondrial number (185); distance between ER and mitochondria ↑ (117, 185) Nondegradative ubiquitination: restore a normal mitochondrial network (205)	Yes (29)	Parkin-mediated ubiquitination of Mfn2 leads to its degradation, resulting in the selective removal of damaged mitochondria by mitophagy in adult CMs (29)
OPA1	Proteolysis (S1 or S2 cleavage sites; see Fig. 5, A and B, for positions)	Paraplegin (84); PARL (32); YME1L (3, 72, 180, 191); OMA1 (1, 3, 6, 51, 78, 93, 159, 191, 209)	YME1L KO: mitochondrial fragmentation (3, 180, 191) YME1L-OMA1 DKO: maintain L-OPA1 amount, normal mitochondrial morphology (191) OMA1 KO: increased the number of tubular mitochondria (159)	Yes (1, 191)	OMA1 KO alone protects against cardiac dysfunction in various HF mouse models (1); YME1L KO promotes DCM and HF, and YME1L-OMA1 DKO rescues this cardiac phenotype (159, 191)
OPA1	Acetylation (K926 and K931)	SIRT3 (166)	SIRT3 KO: hyperacetylation of OPA1 and mitochondrial fragmentation SIRT3 overexpression: maintain mitochondrial morphology and networks; protect from doxorubicin-mediated cell death	Yes (166)	Acetylation- and deacetylation-mimetic mutants were tested only in OPA1-null MEFs, not in CMs (166)

Information about ubiquitination sites was collected from Sarraf et al. (168) and Chen and Dorn (29). CMs, cardiomyocytes; DCM, dilated cardiomyopathy; DKO, double knockout; ER, endoplasmic reticulum; HF, heart failure; Huwe1, E3 ubiquitin-protein ligase HUWE1; KO, knockout; L-OPA1, membrane-anchored long form of optic atrophy protein-1; MARCH5, E3 ubiquitin-protein ligase MARCH5; MEFs, mouse embryo fibroblasts; Mfn1 and Mfn2, mitofusin 1 and 2, respectively; ND, not determined; OMA1, overlapping with the m-AAA protease 1 homolog (a zinc metalloprotease); OMM, outer mitochondrial membrane; OPA1, optic atrophy protein-1; paraplegin, an m-AAA metalloprotease; PARL, presenilins-associated rhomboid-like protease; PINK1, phosphatase and tensin homolog (PTEN)-induced putative kinase 1; S1 and S2, sites 1 and 2, respectively; SIRT3, NAD-dependent deacetylase sirtuin-3; USP30, ubiquitin COOH-terminal hydrolase 30; YME1L, ATP-dependent zinc metalloprotease YME1L1.

POSTTRANSLATIONAL MODIFICATIONS OF MITOCHONDRIAL FISSION PROTEINS IN THE HEART

Overview of the Structure of Mitochondrial Fission Proteins and Their Posttranslational Modifications

Mitochondrial fission in mammalian cells (including CMs) is mediated by 1) an ~80-kDa cytosolic soluble protein, Drp1 (also called DLP1; Fig. 3), which was originally discovered as a member of the dynamin family of GTPases that drive cellular membrane remodeling, and 2) its receptors at the OMM [e.g., Mff (65, 138), MiD49, and MiD51, respectively (141, 210); see also introduction]. Human fission factor 1 (hFis1, 17 kDa) is a small OMM protein with a COOH-terminal single transmembrane (TM) domain, which was also proposed to act as a Drp1 receptor in mammalian cells on the basis of initial studies using yeast (123). Indeed, earlier reports showed that overexpression of hFis1 induces mitochondrial fragmentation in mammalian cells (86, 182, 201), but subsequent studies have claimed that KO or knockdown of hFis1 in mammalian cells does not affect Drp1 translocation/association to mitochondria (101, 112, 135, 137, 140). Although these studies do not exclude a possible role for hFis1 in mitochondrial fission independent of the recruitment of Drp1 to the OMM, we mainly focus on the function of the recently discovered Drp1 receptors (i.e., Mff, MiD49, and MiD51) rather than hFis1 in this review. It has been suggested that the relative contributions of each Drp1 receptor for overall fission activity may differ across cell types and tissues (112, 140). Mff is likely the primary Drp1 receptor at the OMM in the heart, since loss of Mff in vivo causes a significant decrease in mitochondrial fission and exacerbates cardiomyopathy (23; Table 2). Structurally, Drp1 is composed of a highly conserved NH₂-terminal GTPase domain that provides mechanical force, a middle domain involved in higher-order self-assembly, a variable domain (also known as B insert) that functions in protein-protein interactions, and a GTPase effector domain (GED) that interacts with the NH₂-terminal domains to regulate GTPase activity (64, 90, 196; Fig. 3). Importantly, B insert normally blocks the Mff binding sites on Drp1 and retains this protein as a dimer at the cytosol (64, 196). Once B insert binds to cardiolipin at the OMM, this allows for the exposure of the Mff binding site of Drp1 for Drp1-Mff interaction (111, 135). The binding of Drp1 dimers to Mff allows Drp1 assemblies to form functionally active Drp1 oligomers on the OMM as a helical ring structure, which promotes the GTP-mediated constriction of mitochondrial tubules (33, 163).

Fig. 3.

Schematic diagram of dynamin-related protein-1 (Drp1) structure and distribution of posttranslational modifications in cardiomyocytes and hearts. Structure-based domain architecture of human Drp1 isoform 1 is shown on the basis of the published data of Drp1 isoform 2 given by Fröhlich et al. (64). All the posttranslational modifications listed in this diagram are based on reports using native cardiomyocytes or heart tissues (see also Table 3). Information on sites of conjugation of small ubiquitin-like modifier (SUMO) proteins (SUMOylation) was collected from PhosphoSitePlus (80). BSE, bundle signaling element; CaN, Ca²⁺-dependent phosphatase calcineurin; GED, GTPase effector domain; O-GlcNAc, O-linked-N-acetyl-glucosamine; O-GlycNAcylation, O-linked-N-acetyl-glucosamine glycosylation; Pim-1, proto-oncogene serine/threonine-protein kinase; RIP1, receptor-interacting protein-1; ROCK, Rho-associated protein kinase; SENP5, sentrin/SUMO-specific protease 5; VD, variable domain.

Multiple PTMs of Drp1 have been detected by mass spectrometry-based proteomics, and some of them have been further investigated in cell systems to understand their roles in mitochondrial and cellular functions under physiological and pathophysiological conditions [see our recent review (90)]. These PTMs include phosphorylation, ubiquitination, conjugation of small ubiquitin-like modifier (SUMO) proteins (SUMOylation), S-nitrosylation, and O-linked-N-acetyl-glucosamine glycosylation (O-GlcNAcylation), which can modulate functional properties of Drp1 such as its subcellular localizations, protein-protein interactions (e.g., Drp1-Mff binding), protein stability, and GTPase activity (see also Fig. 1). Among these PTMs, the phosphorylation, SUMOylation, and O-GlcNAcylation of Drp1 have been reported in CMs and hearts (90; Table 3 and Fig. 3). Importantly, since the B insert region contains multiple sites for phosphorylation (Fig. 3), SUMOylation, and O-GlcNAcylation, these PTMs of Drp1 in particular may induce a conformational change of the B insert that controls the binding affinity of Drp1 with Mff and ultimately the mitochondrial fission activity in CMs. In addition to Drp1, there are several reports elucidating the PTMs of Drp1 receptors [e.g., AMP-activated protein kinase (AMPK)-dependent phosphorylation of Mff (49, 187) and E3 ubiquitin-protein ligase MARCH5 (MARCH5)-dependent ubiquitination of MiD49 (199)] and their influence on mitochondrial fission activity in noncardiac cells. Recently, Qi’s group reported the concomitant phosphorylation of Mff and Drp1, possibly by ERK1/2 and PKCδ, respectively, during succinate stimulation in neonatal CMs, suggesting that the functional consequences of increased fission in CMs may be derived from the combination of the effects of phosphorylation of both Drp1 and Mff (113). Future studies are required to precisely delineate the effects of PTMs of Drp1 and Mff separately under cardiac stress signaling.

Phosphorylation and O-GlcNAcylation of Drp1 in Cardiomyocytes and Cardiac Pathology

In this section, we introduce data regarding the function and regulation of mitochondrial fission proteins from adult CMs and cell models for adult CMs (see overview: mitochondrial morphology and fission and fusion proteins in cardiomyocytes). Putative Drp1 inhibitors including mitochondrial division inhibitor 1 (Mdivi-1) have been widely applied to inhibit the GTPase activity of Drp1 in cells and in vivo, which frequently exhibits protective effects from mitochondrial injury and cell death [see review (162)]. However, these beneficial effects of these drugs are likely via nonspecific inhibition of complex I (10), and thus the data obtained by the use of Mdivi-1 may not be suitable for understanding Drp1 function or the roles of PTMs of Drp1. Therefore, to precisely understand the roles of PTMs of Drp1 in CMs, we present our discussion here based on the accumulating data using the mutants of Drp1 that mimic or diminish PTMs of Drp1 as well as the experimental results from genetic modification of upstream signaling pathways, rather than the reports using pharmacological inhibition of Drp1 activity and upstream signaling. A full list of PTMs of Drp1 currently reported from all cell types and tissues is given in our recent review (90). The PTMs of Drp1 specifically tested or detected in CMs/hearts are summarized in Table 3.

Two phosphorylation sites on Drp1, Ser616 and Ser637 within the B insert, are well reported in both noncardiac and cardiac cells [see also our recent review (90)]. Ser616 has been shown to be phosphorylated by cyclin-dependent kinase 1 (CDK1/cyclin B; 116, 183, 184, 206), ERK1/2 (98, 160, 172, 203), PKCδ (47, 113, 157, 206), CaMKII (200), and Rho-associated protein kinase (ROCK; 12). On the other hand, the phosphorylation levels of Ser637 are regulated by PKA (20, 35, 71, 100), CaMKIα (76), PKD (91), Ca²⁺-dependent phosphatase calcineurin (CaN; 17, 149, 173), and protein phosphatase 2 (PP2A; 44, 119). Additionally, Ser637 is located within a consensus motif for ROCK and has been found to be phosphorylated by ROCK in endothelial cells (194).

Ser616 phosphorylation promotes mitochondrial fission, which is consistent with reports from multiple groups using various noncardiac cells (90). The next question is whether Ser616 phosphorylation of Drp1 under pathological conditions modulates mitochondrial morphology and function in CMs and hearts. Wang and Sheu’s group have recently proposed that CaMKII-dependent phosphorylation of Drp1 at Ser616 activates noncanonical functions of Drp1 [e.g., regulation of transient mitochondrial permeability transition pore (mPTP) opening and oxidative phosphorylation (208)] in adult CMs (200). However, the possibility remains that these effects can be mediated via the canonical function of Drp1 (i.e., increased mitochondrial fission via Ser616 phosphorylation) since the effect of Ser616 phosphorylation on mitochondrial morphology has only been tested in H9c2 cardiac myoblasts (91, 200) and neonatal CMs (12, 113) and has not yet been investigated in adult CMs. Further studies are required to quantitatively assess whether the canonical (i.e., mitochondrial fission) and noncanonical functions of Drp1 are linked to or occur independently of each other and how Ser616 phosphorylation regulates both functions in CMs under pathological conditions such as HF.

Unlike Ser616 phosphorylation, the impact of Ser637 phosphorylation on mitochondrial morphology is highly controversial in both noncardiac and cardiac cells (81). Overexpression of phosphomimetic or dephosphomimetic mutants of Drp1 (Ser637Asp and Ser637Ala, respectively) is frequently utilized to understand the functions of Ser637 sites in various cell systems but has resulted in conflicting outcomes for mitochondrial morphology: expression of Ser637Asp promotes mitochondrial elongation (17, 20, 35) or fragmentation (76), and Ser637Ala promotes fragmentation (17, 35), elongation (76), or no significant changes in mitochondrial morphology (20, 91; Table 3). Moreover, another PTM of Drp1, O-GlcNAcylation at Thr585 and Thr586, has been shown to result in dephosphorylation of Drp1 at Ser637 via an unknown mechanism, which promotes Drp1 translocation to the OMM and mitochondrial fragmentation under hyperglycemic conditions in neonatal CMs (66; Table 3). We also showed that PKD-dependent phosphorylation of Ser637 induces Drp1 association to the OMM, resulting in mitochondrial fragmentation under G_q protein-coupled receptor stimulations including α₁-adrenoceptor (α₁-AR) stimulation (10 μM phenylephrine for 30 min to 1 h) in neonatal CMs, which initially increases mitochondrial respiration but later promotes mPTP opening and apoptotic signaling activation (91). There are several possible causes for this discrepancy in the field. First, the effects of Ser637 phosphorylation and dephosphorylation have been tested in various cell types that have distinct mitochondrial shapes/networks at resting conditions and possess different sets of upstream signaling molecules (i.e., kinases and phosphatases) for regulating the phosphorylation state of Ser637. Second, because of variability of signaling cascades between different cell types, we cannot exclude the possibility of the coexistence of additional PTMs within Drp1 caused by different upstream molecules in addition to Ser637 phosphorylation. Here, we discuss further the discrepancy of the functions of Ser637 sites in CMs. Cribbs and Strack first showed that acute β-AR stimulation by an injection of the β-AR agonist isoproterenol in animals and exercise (forced swimming) promotes Ser637 phosphorylation in Drp1 in the heart in vivo, under the assumption that this phosphorylation occurs via PKA based on observations from noncardiac cells (35; Table 3). Indeed, the sequences around Ser637 match the consensus motif of PKA. Next, Wang’s group, using isolated adult CMs, confirmed that 12-h incubation with isoproterenol exhibits increased Ser637 phosphorylation by PKA (but not via CaMKII; 200). However, they could not detect increased Ser637 phosphorylation after 2-wk isoproterenol infusion in vivo, and the functional consequences of the Ser637 phosphorylation were not investigated both in CMs and in vivo in this publication (200). These observations can be partly attributed to 1) the activation of the Ca²⁺-dependent phosphatase CaN in the later phase of chronic β-AR stimulation (146) and/or 2) the activation of PP2A via the β₂-AR-G_i protein pathway (108), which may dephosphorylate Drp1 at Ser637, similar to observations in noncardiac cells and neonatal CMs (35, 44, 119, 149). Thus, the phosphorylation levels of Ser637 by PKA/CaN pathways may vary time dependently during chronic β-AR stimulation (i.e., the time course of HF development) in vivo. We also reported that PKD-mediated phosphorylation at Ser637 under G_q protein-coupled receptor stimulation promotes mitochondrial fragmentation (91). Our observation is consistent with the previous results of CaMKIα- and ROCK1-dependent Drp1 phosphorylation at Ser637 in noncardiac cells (76, 194). However, these results show the opposite effect of what PKA does via Ser637. This discrepancy may be partly due to the involvement of different upstream molecules in different AR subtype stimulations in CMs (β-AR vs. α₁-AR) that may not only regulate the phosphorylation levels of Ser637 in Drp1 but also modulate the phosphorylation levels of other site(s) within Drp1 or other mitochondrial fission/fusion machinery proteins. In line with this idea, we also tested β-AR- and PKA-mediated changes in Drp1 phosphorylation, its translocation to the OMM, and mitochondrial morphology (91). Indeed, β-AR stimulation activates PKA and increases Drp1 phosphorylation at Ser637, which promotes Drp1 translocation to mitochondria in H9c2 cells, similar to the case of PKD-dependent Drp1 phosphorylation at Ser637 (91). However, β-AR stimulation did not show any significant changes in mitochondrial morphology. To further investigate the molecular mechanisms producing this discrepancy between the effects of PKA and PKD on mitochondrial fission in H9c2 cells, we assessed the changes in phosphorylation levels of Drp1 at Ser616 before and after α₁-AR stimulation using a dephosphomimetic mutant of Drp1, Drp1-Ser637Ala. Importantly, α₁-AR stimulation significantly decreases Ser616 phosphorylation in cells overexpressing a Drp1-Ser637Ala mutant. This result indicates that phosphorylation at one site within Drp1 may prevent or initiate the phosphorylation or dephosphorylation at the other sites by changing the access of other kinases or phosphatases, since Ser616 and Ser637 may be proximal within the three-dimensional structure of Drp1 (64). Another speculation may be the requirement of basal Ser637 phosphorylation for maintaining Ser616 basal phosphorylation. Further experiments using Drp1 mutants with single or double mutations at Ser616 and Ser637 (17) in CMs may clarify the relative roles of these two sites in mitochondrial fission more precisely. In summary, it is likely that the phosphorylation status of Ser637 under physiological and/or pathological conditions participates in the regulation of mitochondrial fission in CMs as seen in Ser616 phosphorylation.

SUMOylation of Drp1 in Cardiomyocytes and Cardiac Pathology

Drp1 is also shown to be modified by the conjugation of SUMO proteins (i.e., SUMOylation), which occurs at multiple nonconsensus lysine sites within its B insert including Lys594, Lys597, Lys606, and Lys608 (59; Fig. 3 and Table 3). SUMOylation of Drp1 has been extensively studied in noncardiac cells, but its physiological and pathological relevance to mitochondrial and cellular functions is still controversial. For instance, McBride’s group showed that SUMOylation of Drp1 by mitochondrial-anchored protein/SUMO ligase (MAPL) stabilizes its oligomeric form on the OMM resulting in increased mitochondrial fission during apoptosis (155, 195), whereas deSUMOylation of Drp1 by sentrin/SUMO-specific protease 5 (SENP5) reduces mitochondrial fission assessed by knockdown of SENP5 (214). Later, they also reported that SENP5 translocation from the nucleus to the mitochondria and deSUMOylation of Drp1 at the onset of mitosis also resulted in increased Drp1 oligomerization, followed by mitochondrial fragmentation (213). On the other hand, Henley’s group showed that SENP3 (a SUMO-2/3-specific protease)-mediated deSUMOylation of Drp1 plays a major role in regulating mitochondrial fission (73). Currently, there are only few reports proposing significant roles of SUMOylation and deSUMOylation of Drp1 in CMs/hearts (99, 175), especially focusing on SENP5. Importantly, SENP5 was reported as only SENP that is overexpressed in human HF (99). In transgenic mice with CM-specific overexpression of SENP5 (99), decreased SUMO-2/3-modified Drp1 induces enlarged (or swollen) mitochondria and CM apoptosis, leading to cardiomyopathy and HF (99). Moreover, Calvert’s group found that the expression of SENP5 increases in conventional DJ-1 [also known as Parkinson’s disease autosomal recessive, early onset 7 (Park7)] KO mice, and DJ-1 deficiency exacerbates acute ischemia-reperfusion (I/R) injury partly via decreased SUMO-2/3-modified Drp1 (175). Importantly, decreased SUMO-2/3-modified Drp1 is preferentially localized in the mitochondrial fraction without changing the whole cell expression level of Drp1, suggesting that the SUMOylation state of Drp1 regulated by SENP5 likely influences its subcellular localization (i.e., mitochondrial translocation) rather than Drp1 protein expression in CMs, consistent with the results from one of the reports from McBride’s group using noncardiac cells (213). However, SENP5 may target mitochondrial fission/fusion proteins other than Drp1 in these in vivo models (99, 175), which could cause the unbalance of SUMOylation-deSUMOylation in other fission/fusion proteins that may ultimately impact mitochondrial morphology and/or function in vivo. Further studies are required to understand the relative contributions of SUMOylationof Drp1 to the regulation of mitochondrial morphology and function compared with other PTMs of Drp1 under HF. Since the SUMO conjugation pathways are likely to be potential targets in the prevention and treatment of HF (113, 118), the physiological and pathological relevance of SUMOylation with regard to mitochondrial fission and fusion proteins (see also roles of posttranslational modifications of mitochondrial fission and fusion proteins in mitochondrial fragmentation under heart failure) in CMs is a topic to explore in the future.

POSTTRANSLATIONAL MODIFICATIONS OF MITOCHONDRIAL FUSION PROTEINS IN THE HEART

Overview of the Structure of Mitochondrial Fusion Proteins and Their Posttranslational Modifications

Because mitochondria are double-membraned organelles, mitochondrial fusion requires initial fusion of the OMM, mediated by Mfn1 and Mfn2, followed by fusion of the inner mitochondrial membrane (IMM), which occurs mainly via OPA1 (18, 150, 192). The precise mechanisms underlying the coupling of OMM and IMM fusion machinery remain unclear, but the main driving force for mitochondrial fusion is GTP hydrolysis catalyzed by the appropriate GTPase. Mfn1 and Mfn2 are highly homologous dynamin-related GTPases (~80% similarity in humans) and share a similar molecular structure, consisting of an NH₂-terminal GTPase domain, a heptad-repeat 1 domain (HR1), two TM domains, and a COOH-terminal heptad-repeat 2 domain (HR2; 161; Fig. 4A). Mfn1 and Mfn2 also possess a unique topology by spanning the OMM twice with their TM domains, which results in a short loop within the intermembrane space (IMS) and leaves major parts of the protein, including the GTPase domain, HR1, and HR2, exposed to the cytosol (Fig. 4B). Between HR1 and the TM domains, Mfn2 (but not Mfn1) possesses a proline-rich domain, which is likely responsible for Mfn2-specific protein-protein interactions (Fig. 4, A and B). Based on recent reports of mammalian Mfn1 structure (16, 158), the proposed model for the OMM fusion step is as follows (Fig. 4C): 1) HR2s initiate the tethering of adjacent mitochondria by forming HR2-HR2 dimers; 2) GTPase domains of Mfn proteins may contact each other across OMM tubules, suggesting that GTP-dependent oligomerization is also involved in the initial tethering in addition to the HR2-HR2 interaction; and 3) the membranes are pulled together by a power stroke dependent on GTP hydrolysis, allowing for the fusion of the two OMMs [see also review (61)]. Although this model was developed on the basis of structural information about Mfn1, it is likely to be applicable to Mfn2. Indeed, several groups have shown that the HR2 of Mfn interacts with the HR2 of other Mfn molecules to form homodimers (Mfn1-Mfn1 or Mfn2-Mfn2) and/or heterodimers (Mfn1-Mfn2) to initiate mitochondrial fusion (22, 102, 161). These results suggest that Mfn1 and Mfn2 are partially redundant in their function, as either homodimer can suffice to initiate mitochondrial fusion (61). However, several functions that differ between Mfn1 and Mfn2 have been reported. This may be partly due to the difference in 1) GTPase activities between Mfn1 and Mfn2 (recombinant Mfn1 exhibits an approximately eightfold higher GTPase activity than Mfn2; 83) and 2) subcellular expression patterns [e.g., in addition to the OMM, Mfn2 (but not Mfn1) is localized in the endoplasmic reticulum (ER)/sarcoplasmic reticulum (SR) and tethers ER/SR to the mitochondria; 28, 39]. On the basis of KO studies of Mfn2 in cell systems and in vivo (Table 2), in addition to its unquestioned role in OMM fusion, it has been proposed that Mfn2 serves as a key regulator of 1) ER-mitochondrial Ca²⁺ coupling (60), 2) ER stress response machinery at ER-mitochondria contact sites (124, 129, 171), 3) mitochondrial bioenergetics (170), and 4) axonal transport of mitochondria (120). Moreover, in addition to the interaction of HR2-HR2 and GTPase dimerization described above, Dorn’s group reported the importance of intramolecular binding of HR1 and HR2 in Mfn2 (62). They proposed that there are two states of Mfn2 structure: 1) at resting conditions, HR2 is restrained by antiparallel intramolecular binding to HR1, which forms the nonpermissive state; and 2) in the tethering-permissive state, destabilization of HR1-HR2 binding unfolds and extends HR2 into the cytosol, followed by the classic model of the HR2-HR2 interaction.

Fig. 4.

Schematic diagram of structure and distribution of posttranslational modifications of mitofusin 1 and 2 (Mfn1 and Mfn2, respectively), and mechanism of outer mitochondrial membrane (OMM) fusion by Mfn proteins. A: structure-based domain architectures of human Mfn1 and Mfn2 are depicted on the basis of the published data given by Franco et al. (62; see also Table 4). TM, transmembrane domain; HR, heptad-repeat domain. B: Mfn2 topology at the OMM, with two TM domains, a small bridge structure at the intermembrane space, and large structures at the cytosolic space. C: representative cartoon shows Mfn2-Mfn2 interaction. Initial tethering is mediated between heptad-repeat 2 domains (HR2s) from the opposite OMM (step 1). Next, dimerization of GTPase domains occurs (step 2), and power stroke from GTPase hydrolysis promotes OMM fusion (step 3). CMs, cardiomyocytes; IMM, inner mitochondrial membrane; PINK1, phosphatase and tensin homolog (PTEN)-induced putative kinase 1; PR, proline-rich domain.

IMM fusion is regulated by the IMM-associated GTPase OPA1 in mammalian systems (7, 104, 107). OPA1 consists of an NH₂-terminal mitochondrial-targeting sequence (MTS), a TM, an HR, a GTPase domain, a middle domain, and a COOH-terminal GED (Fig. 5A). The NH₂-terminal TM of OPA1 is embedded in the IMM, and the main part of the protein is located in the IMS, allowing it to interact with the loop structure of Mfn proteins for successful coordination to complete mitochondrial fusion (31, 181). The exceptional complexity of the regulation of this protein compared with other mitochondrial fission and fusion proteins is due to the existence of multiple alternative splicing variants [8 variants in humans involving exons 4, 4b, and 5b (40)] and proteolytic processing in the mitochondria (Fig. 5, B and C). Since all OPA1 proteins have an MTS at the NH₂ terminus, all OPA1 proteins can be imported into mitochondria via the OMM and IMM translocases. Subsequently, all OPA1 proteins can be proteolytically cleaved in the mitochondria at the NH₂-terminal side, which forms combinations of membrane-anchored long forms (termed L-OPA1) and soluble short forms of OPA1 (termed S-OPA1; 7, 107; Fig. 5B). Historically, protease presenilins-associated rhomboid-like protease (PARL; 32) and the m-AAA metalloprotease paraplegin (84) were proposed as the main molecules responsible for OPA1 processing (Table 4). However, later studies found that cells lacking PARL or paraplegin maintain normal OPA1 processing (50, 72). The present established model for OPA1 processing is mediated via the function of two mitochondrial proteases, the i-AAA metalloprotease ATP-dependent zinc metalloprotease YME1L1 (YME1L; 3, 72, 180, 191) and the zinc metalloprotease, overlapping with the m-AAA protease 1 homolog (OMA1; 1, 3, 51, 78, 191; Table 4 and Fig. 5, A and B). All variants of OPA1 contain exon 5 with the site 1 (S1) cleavage site targeted by OMA1, whereas half of variants contain the site 2 (S2) cleavage site processed by another peptidase, YME1L (Fig. 5, B and C). Since cleavage at S1 is incomplete without the exon 4b-encoded sequence, some of the variants result in both uncleaved L-OPA1 and cleaved S-OPA1 (Fig. 5C). S-OPA1 is distributed in the IMS because of the lack of a membrane-anchoring TM, but it still retains its GTPase domain (Fig. 5B). S-OPA1 may associate with the IMM via interactions with L-OPA1 integrated in the IMM. The next question is whether there are different roles between L-OPA1 and S-OPA1. Regarding the disputed function of OPA1 (i.e., IMM fusion), the relative contributions of OPA1 isoforms are still controversial. Originally, Chan’s group reported that both L-OPA1 and S-OPA1 are required for successful fusion between mitochondria (180). However, other groups have shown that L-OPA1 alone is sufficient for mitochondrial fusion, whereas S-OPA1 alone is not capable of forming fusion machinery or has a minor role in mitochondrial fusion (3, 84, 106, 186). On the basis of KO studies in cell systems, OPA1 is also proposed to have other functions in addition to mitochondrial fusion. These include 1) maintenance of mtDNA content, 2) regulation of mitochondrial bioenergetics, and 3) cristae organization [see reviews (40, 107)]. Originally, the proposed mechanism underlying the regulation of cristae tightness (i.e., the narrowness of the width of the intracristal space formed between 2 cristae membranes) and bioenergetics by OPA1 is via 1) trans-interaction of L-OPA1 or 2) L-OPA1 oligomerization facilitated by S-OPA1 as a molecular staple (40, 107), which regulates the assembly and stability of the respiratory chain supercomplexes as well as respiratory efficiency (34). However, these models may require significant revision based on recent findings, using an OPA1 KO cell line with selective expression of each OPA1 isoform (41, 106). Using this system, two groups confirmed that S-OPA1 has a minor role or no function in mitochondrial fusion and L-OPA1 is sufficient for recovering mitochondrial fusion in OPA1 KO cell lines (41, 106). Moreover, Yoon’s group showed that expression of OPA1 variant 5, which produces only S-OPA1 (see Fig. 5C), is sufficient to maintain mtDNA, respiratory complexes, and cristae structure, indistinguishable from the functions of L-OPA1 (106). Zanna’s group tested all eight variants and found no differences in the effect of bioenergetics between each variant (41). These results indicate that S-OPA1 alone is sufficient for the maintenance of mitochondrial bioenergetics, which supports the original idea from Scorrano’s group that the mitochondrial fusion function of OPA1 is molecularly distinct and independent from cristae maintenance by OPA1 (63).

Fig. 5.

Schematic diagram of optic atrophy protein-1 (OPA1) structure and variant-dependent proteolysis. A: structure-based domain architecture of human OPA1 is depicted. Numbers in the domains indicate the exon number. Red exons are involved in the formation of splice variants. B: variant-dependent proteolysis and the structures of the membrane-anchored long and soluble short forms of OPA1 (L-OPA1 and S-OPA1, respectively). C: summary table of variant-dependent proteolysis. GED, GTPase effector domain; IMM, inner mitochondrial membrane; IMS, intermembrane space; MPP, mitochondrial processing peptidase; MTS, mitochondrial-targeting sequence; OMA1, overlapping with the m-AAA protease 1 homolog; S1 and S2, sites 1 and 2; TM, transmembrane domain; YME1L, i-AAA metalloprotease.

In addition to the KO/ablation studies both in cell systems and in vivo (see Table 2), PTMs of Mfn1, Mfn2, and OPA1 have also been investigated (Table 4). Multiple PTMs of human mitochondrial fusion proteins such as phosphorylation (in Mfn1, Mfn2, and OPA1), acetylation (in Mfn1, Mfn2, and OPA1), methylation (in Mfn1 and OPA1), and ubiquitination (in Mfn1, Mfn2, and OPA1) have been detected by mass spectrometry-based proteomics as shown in online open databases [e.g., PhosphoSitePlus (80)]. However, the regulation of mitochondrial fusion proteins by PTMs remains largely uncharacterized compared with Drp1. For instance, the following PTMs of Mfn1 and Mfn2 have been investigated in noncardiac cells: 1) phosphorylation (in Mfn1 and Mfn2), 2) acetylation (in Mfn1), 3) ubiquitination (in Mfn1 and Mfn2), and 4) cysteine oxidation (in Mfn2; Table 4 and Fig. 4A). Currently, there is a single report of PTMs of Mfn2 (phosphorylation and ubiquitination) in CMs/hearts, which will be discussed in the following sections. Although Mfn1 is highly expressed in the heart compared with Mfn2 (57), there are no reports of PTMs of Mfn1 in CMs/hearts. The investigation of PTMs of OPA1 mainly focuses on proteolytic processes as described above. Most of the prior studies use genetic manipulations of the responsible proteases to modify the L-OPA1-to-S-OPA1 ratio in cells or in vivo (3, 72, 78, 180, 191) or the overexpression of specific OPA1 variants and/or a mutant OPA1 variant 1 that lacks S1 sites, which forms only L-OPA1 in the mitochondria (41, 106). In the following section, we mainly discuss the physiological and pathological relevance of the proteolytic cleavage of OPA1 in CMs and hearts and also introduce PTMs of OPA1 specifically reported in CMs.

Phosphorylation and Ubiquitination of Mfn2 in Cardiomyocytes and Cardiac Pathology

In the heart, Mfn1 and Mfn2 are required for early embryonic (30, 97) and postnatal development (143). Moreover, double KO (DKO) of both Mfn1 and Mfn2 in adult mice showed dramatic changes in mitochondrial morphology, respiration, and contractile function, resulting in HF and CM death (28, 75; see Table 2). Similar results were observed from CM-specific Mfn2 single-KO mice, but their cardiac phenotype was milder than that of the DKO (28, 142). Although Mfn1 expression is higher than that of Mfn2 in the heart (57), CM-specific Mfn1 KO does not cause any significant changes in cardiac function (28, 143). Given that both Mfn1 and Mfn2 are capable of inducing mitochondrial fusion, these observations indicate that 1) Mfn1 cannot take the place of (or substitute for) Mfn2-specific functions such as mitochondria-SR tethering and mitophagy; 2) Mfn2 can compensate mitochondrial fusion activity under Mfn1 deletion; and 3) inhibition of Mfn2-specific functions by Mfn2 KO may contribute to the mechanisms of cardiac pathology observed in the transgenic mice. Indeed, Mfn2 KO increases the distance between mitochondria and SR (28) and delays in Ca²⁺-induced mPTP opening (142). Moreover, Chen and Dorn showed that CM-specific Mfn2 KO promotes cardiomyopathy via impaired mitophagy with the accumulation of damaged mitochondria (29). They also showed that Mfn2 serves as a Parkin receptor on damaged mitochondria, which promotes mitochondrial recruitment of Parkin; phosphatase and tensin homolog (PTEN)-induced putative kinase 1 (PINK1)-dependent Mfn2 phosphorylation, possibly at Thr111 and Ser442, facilitates Parkin binding to Mfn2, followed by its Parkin-mediated ubiquitination of Mfn2 (Table 4 and Fig. 4A). Thus, ablation of Mfn2 in mouse CMs prevented depolarization-induced translocation of Parkin to the mitochondria and suppressed mitophagy. Historically, Mfn2 was shown to interact with the E3 ubiquitin ligase MARCH5 (96, 125), but evidence for neither direct ubiquitination nor degradation of Mfn2 was observed in these initial publications. Later, Taanman’s group reported the ubiquitination of both Mfn1 and Mfn2 in a PINK1/Parkin-dependent manner upon induction of mitophagy (67), indicating that inhibition of the PINK1/Parkin/Mfn1/2 pathway may contribute to the pathogenesis of Parkinson’s disease. Thus, the report from Chen and Dorn in CMs (29) is indeed well recapitulated by the findings of Taanman’s group in neurons (67). Similarly, on the basis of in silico consensus sequence analysis, several groups showed that Mfn2 can be phosphorylated by PKA at Ser442 in noncardiac cells (24, 212), which is the same site as the predicted PINK1-dependent phosphorylation site on Mfn2 in CMs (Table 4 and Fig. 4A). These results suggest that the function of Mfn2 is posttranslationally regulated in CMs and hearts. However, future studies may be required to further address the following questions: 1) Does the activity of the PINK1/Parkin pathway change under cardiac pathological conditions such as HF, myocardial infarction, and I/R? 2) Since Parkin expression is lower in the heart compared with in neurons (46), what is the relative contribution of Mfn2-mediated mitophagy compared with other mechanisms such as Drp1-mediated mitophagy (82, 164, 176)? 3) Since PKA signaling is one of the predominant cell-signaling pathways in CMs and also important in the development of cardiac pathologies (9), can PKA phosphorylate Mfn2 similar to PINK1? 4) Do PTMs of Mfn2 modulate other Mfn2 functions such as mitochondria-SR interactions under physiological and pathological conditions? Finally, since there are several reports regarding PTMs of Mfn1 (145, 156) and their functions in noncardiac cells (but not in CMs) and Mfn1 is highly expressed in CMs (57), it should be noted that the investigation of PTMs of Mfn1 and their functional relevance in CMs may also be an important topic to explore in the future.

Proteolytic Cleavage of OPA1 in Cardiomyocytes and Cardiac Pathology

As described above, proteolytic cleavage of OPA1 is regulated by YME1L and OMA1 (see posttranslational modifications of mitochondrial fusion proteins in the heart, Overview of the Structure of Mitochondrial Fusion Proteins and Their Posttranslational Modifications). In noncardiac cells, it has been shown that cellular stress, mitochondrial membrane depolarization, and KO of YME1L can activate the other protease, OMA1, which promotes the conversion of L-OPA1 into S-OPA1 and subsequent mitochondrial fragmentation (3, 6, 93, 159, 209; Table 4). Later, Langer’s group investigated the role of OPA1 processing on mitochondrial morphology in the CMs and cardiac function in the heart by generating CM-specific YME1L and OMA1 KO mouse lines (191). First, they confirmed via conventional KO that YME1L, but not OMA1, is essential for embryogenesis. Next, using cardiac-specific YME1L KO mice, they found that the ablation of YME1L activates OMA1 and accelerated OPA1 proteolysis (see also Fig. 5, B and C), which promotes mitochondrial fragmentation and altered cardiac metabolism, followed by dilated cardiomyopathy and HF. Importantly, these cardiac and mitochondrial dysfunctions can be rescued by OMA1 KO because of the prevention of excessive OPA1 cleavage. The authors conclude that L-OPA1 is sufficient to maintain heart function, and OMA1 is a critical regulator of CM survival and the maintenance of mitochondrial function via regulating proteolytic cleavage of OPA1. Finally, Enríquez’s group also reported that OMA1 KO alone protects against cardiac dysfunction in various HF mouse models (1). These reports raise the intriguing possibility that the proteolytic cleavage of OPA1 is critical for the maintenance of mitochondrial morphology and function in the heart and dysregulation of this process contributes to the pathogenesis of HF. However, since proteases such as YME1L and OMA1 likely have multiple target proteins in CMs, it is difficult to completely exclude the possibility that the phenotypes observed in these mouse lines are mediated by the alterations in the proteolysis of these other proteins in CMs, but not solely via OPA1 cleavage. Future studies are required to delineate the differential roles of L-OPA1 and S-OPA-1 in CMs by reexpressing specific OPA1 variants and/or a mutant OPA1 variant 1 (41, 106; see Fig. 5C) in the OPA1-ablated mouse lines (see Table 2).

Acetylation and O-GlcNAcylation of OPA1 in Cardiomyocytes and Cardiac Pathology

In addition to proteolysis, several other PTMs of OPA1 are reported such as acetylation and O-GlcNAcylation (Table 4). Gupta’s group first found the hyperacetylation of OPA1 in cardiac pressure overload animal models (166). In addition, they found that a mitochondrial deacetylase, NAD-dependent deacetylase sirtuin-3 (SIRT3), is capable of directly deacetylating OPA1 at its GED (Lys926 and Lys931; see Fig. 5A), which increases its GTPase activity. Moreover, they showed that CM-specific SIRT3 KO promotes hyperacetylation of OPA1 and mitochondrial fragmentation in the heart. Finally, they confirmed that SIRT3-dependent activation of OPA1 protects CMs from doxorubicin-mediated cell death by maintaining mitochondrial morphology and networks. Later, using H9c2 cells, De Rasmo’s group proposed that PKA signaling, which is one of the major cell-signaling pathways in CMs (9), can modify SIRT3 activity and regulate OPA1 activity (177). The reduction of the mitochondrial cAMP pool promotes SIRT3 protein degradation, followed by hyperacetylation of OPA1, and also triggers the proteolytic cleavage of L-OPA1 and subsequent CM apoptosis. These observations suggest that cardiac stress such as pressure overload or chronic AR stimulation induces OPA1 hyperacetylation, which may contribute to the reduction of the activity of mitochondrial fusion machinery in CMs under HF in vivo. Since SIRT3 KO or overexpression changes the acetylation levels of multiple proteins in the CMs, experimental designs such as reexpressing deacetylation- or acetylation-mimetic OPA1 mutants in the OPA1 ablation animal models (see Table 2) may be appropriate to further elucidate the role of acetylation on OPA1 at Lys926 and Lys931 in CMs and hearts. Similarly, although O-GlcNAcylation of OPA1 was detected under hyperglycemic conditions in neonatal CMs (114), further studies using O-GlcNAcylation site mutants of OPA1 may be necessary to precisely dissect the role of this PTM for OPA1 function in CMs.

ROLES OF POSTTRANSLATIONAL MODIFICATIONS OF MITOCHONDRIAL FISSION AND FUSION PROTEINS IN MITOCHONDRIAL FRAGMENTATION UNDER HEART FAILURE

As described in the introduction, abnormal mitochondrial morphologies, especially small, disorganized interfibrillar mitochondria, are a hallmark feature of failing hearts in both human patients (43, 95, 154, 169) and animal HF models (11, 13, 25, 36, 87, 167, 198). As summarized in this review, mitochondrial dynamics have been widely studied using KO/overexpression systems (Table 2) and/or manipulation of PTMs of mitochondrial fission and fusion proteins (Tables 3 and 4), which has helped advance the understanding of the pathogenesis and molecular mechanisms of HF. In this section, we integrate the information listed from overview: mitochondrial morphology and fission and fusion proteins in cardiomyocytes, posttranslational modifications of mitochondrial fission proteins in the heart, and posttranslational modifications of mitochondrial fusion proteins in the heart and comprehensively discuss how PTMs of mitochondrial fission and fusion proteins affect mitochondrial morphology and function under HF (Fig. 6).

Fig. 6.

Posttranslational modifications (PTMs) of mitochondrial fission and fusion proteins by cardiac pathological signaling promote heart failure but also serve as compensatory mechanisms for protecting cardiomyocytes (CMs). DLP1, dynamin-related protein-1; Mfn2, mitofusin 2; mPTP, mitochondrial permeability transition pore; OPA1, optic atrophy protein-1; ROS, reactive oxygen species.

Enhancement of Mitochondrial Fission Activity under Heart Failure

As described in the introduction, overall fission activity is determined by at least the following three factors: 1) the total expression levels of fission proteins in the cell; 2) the expression levels of these proteins at the mitochondria; and 3) the activities of GTPases (see also Fig. 1). First, increased Drp1 protein expression is observed in several animal models of HF (but not all models), which may promote mitochondrial fission (Table 1). Second, as shown in posttranslational modifications of mitochondrial fission proteins in the heart, Overview of the Structure of Mitochondrial Fission Proteins and Their Posttranslational Modifications, the phosphorylation status of Drp1 regulates the translocation of Drp1 from the cytosol to the OMM (or the binding affinity to the Drp1 receptor Mff), which increases the amount of Drp1 at the OMM. AR signaling, hypoxia, and hyperglycemic signaling in the CMs, which are critical in HF pathogenesis, myocardial infarction, and diabetic cardiomyopathy, respectively, and oxidative stress-sensitive signaling (e.g., CaMKII, PKCδ, and PKD activation) are likely involved in this mechanism (Table 3 and Fig. 6). For instance, Shirakabe and colleagues reported increased Drp1 translocation to mitochondria in a mouse model of nonischemic HF generated by transverse aortic constriction (176). Since these Drp1 phosphorylations likely do not modulate the GTPase activity of Drp1 in vitro or in situ (35), increased mitochondrial fission activity under chronic HF is mainly a result of increased levels of Drp1 at the OMM (see also Fig. 1). Moreover, SENP5 is overexpressed in human HF (99), and SENP5-dependent deSUMOylation of Drp1 may also promote the translocation of Drp1 to the mitochondria under pathological conditions in the heart, followed by mitochondrial fragmentation and CM apoptosis (175; see also posttranslational modifications of mitochondrial fission proteins in the heart, SUMOylation of Drp1 in Cardiomyocytes and Cardiac Pathology). Finally, we also need to take into consideration that Drp1 receptors at the OMM (e.g., Mff) also receive PTMs and may contribute to Drp1 retention at the OMM, although there is currently less information regarding the PTMs of Drp1 receptors compared with Drp1 (see also posttranslational modifications of mitochondrial fission proteins in the heart, Overview of the Structure of Mitochondrial Fission Proteins and Their Posttranslational Modifications).

The last question is whether GTPase activity of Drp1 itself changes under HF. There are numerous reports that genetic and pharmacological inhibition of GTPase activity of Drp1 (see also posttranslational modifications of mitochondrial fission proteins in the heart, Phosphorylation and O-GlcNAcylation of Drp1 in Cardiomyocytes and Cardiac Pathology) is beneficial for protecting CMs and hearts from mitochondrial dysfunction [e.g., I/R injury (134, 207)]. However, these observations do not directly indicate that the GTPase activity of Drp1 itself increases under cardiac pathological conditions. Further studies are needed to directly measure the GTPase activity of Drp1 in CMs/hearts under various pathological conditions or in animal models of HF. For instance, Nishimura and colleagues recently found that the GTP-binding activity of Drp1 is significantly increased in a mouse model of myocardial infarction (131), indicating that the molecular mechanisms underlying enhanced mitochondrial fission activity during HF may differ in various types of cardiac stress (e.g., ischemic vs. nonischemic).

In summary, future experiments should be designed to quantitatively address the relative contributions of 1) Drp1 expression levels, 2) Drp1 translocation to the OMM, and 3) GTPase activity of Drp1 to the overall activity of fission machinery in CMs under HF.

Inhibition of Mitochondrial Fusion Activity under Heart Failure

As shown in posttranslational modifications of mitochondrial fusion proteins in the heart, mitochondrial fusion is mediated via Mfn1, Mfn2, and OPA1. Although Mfn1 is highly expressed in the CMs, PTMs of Mfn1 have not been well investigated. Parkin-mediated ubiquitination of Mfn2 leads to its degradation, resulting in decreased fusion activity, and promotes mitochondrial fragmentation, which leads to the selective removal of damaged mitochondria by mitophagy in adult CMs (29; Fig. 6; see also roles of posttranslational modifications of mitochondrial fission and fusion proteins in mitochondrial fragmentation under heart failure, Mitochondrial Fragmentation and Mitochondrial Dysfunction under Heart Failure). Parkin-dependent ubiquitination of Mfn2 is likely involved in the decreased Mfn2 expression observed in some HF animal models (Table 1), but further studies are required to understand how Parkin/PINK1/Mfn2 can be activated under HF.

Fusion activity of OPA1 is mainly regulated by the amount of L-OPA1, although the role of S-OPA1 in CMs is still under debate. Since the stress response protease OMA1 regulates OPA1 proteolysis and subsequently modulates the amount of L-OPA1 in the heart (1, 191), activation of OPA1 proteolysis is a possible mechanism for decreasing fusion activity in HF (Fig. 6). Moreover, hyperacetylation of OPA1 possibly via decreased activity of PKA and SIRT3 may also be involved in the activation of OPA1 proteolysis under HF (177).

Mitochondrial Fragmentation and Mitochondrial Dysfunction under Heart Failure

Overall, the mitochondrial fragmentation frequently observed in HF is likely attributed to the following two aspects that simultaneously occur via multiple signaling pathways activated under HF (Fig. 6): 1) increased mitochondrial fission activity (see roles of posttranslational modifications of mitochondrial fission and fusion proteins in mitochondrial fragmentation under heart failure, Enhancement of Mitochondrial Fission Activity under Heart Failure) and 2) decreased fusion activity (see roles of posttranslational modifications of mitochondrial fission and fusion proteins in mitochondrial fragmentation under heart failure, Inhibition of Mitochondrial Fusion Activity under Heart Failure). The next question is how mitochondrial fragmentation and/or these two aspects via PTMs of mitochondrial fission and fusion proteins mentioned above cause cardiac mitochondrial dysfunction under HF.

It is still largely unknown whether mitochondrial fission and fusion events influence the beat-to-beat-based regulation of physiological excitation-contraction/metabolism coupling in CMs. On the other hand, it is well documented and demonstrated that mitochondrial fragmentation occurs under both acute and chronic cardiac stress (see introduction) and pharmacological inhibition of the GTPase activity of Drp1 protects CMs and hearts from mPTP opening (82, 115, 134, 173, 200, 207). Although it is still unclear how Drp1 is involved in mPTP opening, several possible molecular mechanisms have been proposed: 1) direct association of Drp1 to the mPTP structure at the OMM triggers mPTP opening; 2) indirect association of Drp1 with other OMM proteins indirectly facilitates mPTP opening (208); and/or 3) there is a link between mitochondrial morphology including mitochondrial cristae structures and contact site stabilization [155; see also reviews (74, 178)]. Recent studies strongly suggest that mitochondrial ATP synthase (F_oF₁ ATP synthase) is a critical component of mPTP and the c-subunit of mitochondrial ATP synthase serves as a pore-forming subunit of mPTP (74, 122). However, the molecular identity of the OMM-spanning component of mPTP is still unknown. Therefore, it is currently very challenging to investigate the possibility of a direct association of Drp1 with the OMM-spanning component of mPTP. Second, since Drp1 can associate with other OMM-bound proteins and also increase mPTP-independent OMM permeability, it is plausible that Drp1 indirectly facilitates mPTP activity via its association with other OMM proteins (121). However, a link between macrostructural and microstructural changes in mitochondrial morphology is the most reasonable of these three possibilities since inhibition of GTPase activity of Drp1 efficiently blocks mPTP opening. Mitochondrial fragmentation by Drp1 may initially increase the efficiency of electron transport at the electron transport chain (ETC) by affecting the ultrastructural and spatial organization of the respiratory chain and ATP synthase (i.e., cristae remodeling), thus stimulating forward electron flow through the ETC and increasing ATP production (34). Indeed, several groups including ours have shown that respiration increases under short-term α₁-AR stimulation or high-glucose treatment (89, 91, 204). Moreover, Wang and Sheu’s group showed that pharmacological and genetic inhibition of the GTPase activity of Drp1 inhibits the ETC activity and the frequency of transient mPTP opening (208). Under well-coupled conditions, stimulation of forward flow by itself would oxidize the ETC and thereby lower ROS levels (130). However, increased electron flow at the ETC may oxidize NADH and NADPH and thus deplete the antioxidative capacity (4). Another possibility is that mitochondrial fragmentation may structurally limit the size of the matrix cavity and mitochondrial network (see also introduction), which contributes to the decrease in the amount of antioxidative enzymes in the matrix. Thus, chronic mitochondrial fragmentation serves as one of the upstream mechanisms for increased mitochondrial ROS, mitochondrial membrane depolarization, and enhanced long-lasting mPTP opening under HF (Fig. 6).

Changes in the levels of PTMs of mitochondrial fission and fusion proteins also alter mitochondrial function in addition to maintenance of mitochondrial morphology. First, dysregulation of mitochondrial cristae formation (or loss of its structure) under HF is well reported in addition to the alteration of global mitochondrial morphology (i.e., mitochondrial fragmentation), which also contributes to inefficient energy production in cardiac mitochondria under HF (13, 25, 36). As discussed in posttranslational modifications of mitochondrial fusion proteins in the heart, Proteolytic Cleavage of OPA1 in Cardiomyocytes and Cardiac Pathology and Acetylation and O-GlcNAcylation of OPA1 in Cardiomyocytes and Cardiac Pathology, this is partly due to the dysregulation of OPA1 activity via proteolysis and alteration of the L-OPA1-to-S-OPA1 ratio. However, it is still unclear why the existence of S-OPA1 is insufficient for maintaining the cristae structures in cardiac mitochondria under HF (1, 191), which is clearly different from observations in noncardiac cells (41, 106). Further studies are needed to address whether other IMM proteins that stabilize cristae morphology [e.g., coiled-coil-helix-coiled-helix domain containing 3 (CHCHD3) and mitofilin (37)] also alter its function during HF in addition to OPA1. Although OMM association of Drp1 is also proposed as a direct regulator of mitochondrial bioenergetics in adult CMs (193), further studies are required to identify the mechanisms of how events in the OMM (i.e., Drp1 translocation) are transmitted to the ETC (which is located at the IMM) and how PTMs of Drp1, especially its phosphorylation, contribute to this noncanonical function of Drp1. Finally, mitochondria-SR interaction may be altered during HF, which contributes to the alterations of Ca²⁺ and ROS cross talk between these two organelles and the regulation of mitochondrial bioenergetics in CMs (14; see also posttranslational modifications of mitochondrial fission proteins in the heart, Overview of the Structure of Mitochondrial Fission Proteins and Their Posttranslational Modifications, and Fig. 2D). Since the importance of Mfn2 tethering functions in CMs has been tested only under Mfn2 KO conditions (28), next steps include exploring whether PTMs of Mfn2 influence mitochondria-SR tethering by changing the three-dimensional structure of Mfn2 or the expression levels of Mfn2 at the contact sites.

The last question that needs to be addressed is whether PTMs of mitochondrial fission and fusion proteins are involved in the development of cardiac pathology or just an adaptation/maladaptation to pathological changes in the heart. As described above, mitochondrial fragmentation may initially work as a compensatory mechanism for matching up with the increased energetic demands in failing hearts due to energy depletion or the inefficient energy production (e.g., increased glycolysis) during HF. However, this may soon turn into a detrimental pathway that increases ROS and mPTP opening, thus activating apoptotic signaling. In addition, mitochondrial fragmentation and the PTMs of fission and fusion proteins may act as an important compensatory mechanism, eliminating damaged mitochondria (i.e., mitophagy, see also posttranslational modifications of mitochondrial fusion proteins in the heart, Overview of the Structure of Mitochondrial Fusion Proteins and Their Posttranslational Modifications and Phosphorylation and Ubiquitination of Mfn2 in Cardiomyocytes and Cardiac Pathology; Fig. 6). Indeed, Sadoshima’s group has recently proposed that Ser616 phosphorylation of Drp1 and subsequent mitochondrial fragmentation are involved in Ras-related protein Rab-9A (Rab9)-associated mitophagy after ischemic stress (164; Table 3). Although multiple technologies have recently been developed, especially the use of fluorescent-labeling techniques, quantitative measurements of mitophagy activity in cells and tissues are still challenging. This includes visualization of 1) colocalization and recruitment of accessory factors [e.g., microtubule-associated protein 1A/1B-light chain 3 (LC3; 82, 94, 176) and Rab9 (164)] by immunostaining and/or overexpression of the fluorescent protein-fused factors and 2) the changes in pH in mitochondria when they are incorporated into lysosomes by overexpressing mitochondria-targeted (mt) acid-stable fluorescent protein [e.g., mt-Keima; see also reviews (48, 165)]. Importantly, it should be noted that PTMs of mitochondrial fission and fusion proteins change the mitochondrial morphology/network and the mitochondrial membrane potential, which directly or indirectly affects these fluorescence-based measurements for mitophagy. For instance, mt-Keima concentration in the mitochondrial matrix may be increased simply by mitochondrial fission events that limit the space available for the diffusion of mt-Keima in the mitochondrial network. This would result in changes of its fluorescence value independent of pH change. The other thing to consider is how to titrate the fluorescence value of mt-Keima in mitochondria with different membrane potentials, since mt-Keima may change its fluorescence value in depolarized mitochondria (but not in the process of mitophagy) independent of lysosomal association. The development of new technologies/tools is crucial to dissect the events that alter mitochondrial morphology, membrane potential, and pH to more quantitatively measure the mitophagic flux. This will help advance our understanding of the molecular mechanisms linking PTMs of fission and fusion proteins and mitophagy in CMs under cardiac stress.

In summary, PTMs of mitochondrial fission and fusion proteins by cardiac pathological signaling may serve as a cause of dysregulated mitophagy and mitochondrial dysfunction and simultaneously work as a compensatory mechanism for mitochondrial and cellular function in CMs and hearts (Fig. 6).

SUMMARY AND CONCLUDING REMARKS

As summarized in this review, there has been substantial progress in understanding the detailed molecular mechanisms of how PTMs of mitochondrial fission and fusion proteins modulate important mitochondrial functions such as mitochondrial dynamics, bioenergetics, Ca²⁺ transport, ROS generation, mPTP activity, and mitophagy in noncardiac cells. Moreover, thanks to recent studies using cultured CM model cells, adult CMs, and cardiac-specific KO mouse lines, the biological importance of mitochondrial fission and fusion proteins in cardiac physiology and pathologies has been gaining recognition. Changes in the activity of mitochondrial fission and fusion proteins via their PTMs in CMs likely contribute to the development of mitochondrial and cellular dysfunctions and also serve as compensatory mechanisms to protect the CMs from cardiac stress. Therefore, modulations of these PTMs of fission and fusion proteins may have promise as potential therapeutic targets for preventing and/or treating cardiac disease. However, it is still up for debate whether altered levels of PTMs of mitochondrial fission and fusion proteins are simply the reflection of overall impairment of cellular function at the end stage of these diseases; some types of PTMs of mitochondrial fission and fusion proteins may be the consequence of the activation of multiple cell-signaling pathways under HF and may not have functional relevance. Approaches such as quantitative measurements of mitochondrial morphology/dynamics, mitochondrial Ca²⁺ handling, bioenergetics, mitochondrial ROS levels, and mPTP activity in cultured CM model cells and adult CMs, combined with the expression of mutant proteins that prevent or enhance PTM effects, may provide clear answers as to whether these PTMs contribute to the regulation of mitochondrial morphology and functions. Finally, in addition to cardiac-specific KO/ablation animals, the development of knock-in mouse lines carrying PTM-mimetic or PTM-removal mutants of fission and fusion proteins may further address the physiological and pathological importance of these modifications in vivo.

GRANTS

This work was supported by Brown University Karen T. Romer Undergraduate Teaching and Research Award (to S. M. Adaniya), American Physiological Society (APS) 2017 Shih-Chun Wang Young Investigator Award (to J. O-Uchi), National Heart, Lung, and Blood Institute Grant R01-HL-136757 (to J. O-Uchi), American Heart Association (AHA) Grant 16SDG27260248 (to J. O-Uchi), National Institute of General Medical Sciences Grant U54-GM-115677 (to B. S. Jhun), and AHA Grant 18CDA34110091 (to B. S. Jhun). B. S. Jhun is a recipient of the 2018 New Investigator Award from the APS Cell and Molecular Physiology Section.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.M.A., J.O-U., Y.K., and B.S.J. prepared figures; S.M.A., J.O-U., and B.S.J. drafted manuscript; S.M.A., J.O-U., M.W.C., Y.K., and B.S.J. edited and revised manuscript; S.M.A., J.O-U., M.W.C., Y.K., and B.S.J. approved final version of manuscript.