Mitophagy describes the selective targeting and degradation of mitochondria by the autophagy pathway. In this process, defective mitochondria are first purged from the mitochondrial network then delivered to the lysosome by the autophagy machinery. Mitophagy has emerged as a key facet of mitochondrial quality control and has been implicated in a variety of human diseases. Disturbances in the cellular control of mitophagy can result in a dysfunctional mitochondrial network with grave implications for high energy demanding tissue. The present chapter reviews the recent advancements in the study of mitophagy mechanisms and regulation.
- mitochondrial dynamics
- mitochondrial dysfunction
- Parkinson's disease
For decades, mitochondria have been recognized as the powerhouses of eukaryotic cells and we now appreciate that they function in other key cellular processes such as calcium signalling and apoptosis. The mitochondrial network is an amazingly dynamic and adaptable organelle system that must remain healthy in order to meet changing demands for ATP. Indeed, disturbances in mitochondrial homoeostasis result in an increasingly damaged and dysfunctional mitochondrial network, leading to a range of human pathologies including ischaemia, diabetes and neurodegeneration . Dysfunctional mitochondria have the potential to generate vast quantities of ROS (reactive oxygen species) and can become ‘leaky’, releasing pro-apoptotic proteins into the cytoplasm, often with catastrophic consequences. If unchecked, deleterious mtDNA (mitochondrial DNA) mutations are also likely to manifest within the network and further compromise mitochondrial activity . Cells must therefore efficiently remove and replace damaged mitochondria with healthy ones in order to maintain their energy supply.
Damaged mitochondria are targeted for degradation in the lysosomes via a specific autophagic pathway called mitophagy . When mitophagy was first described, the distinction between it and the non-selective autophagy pathway was hazy. However, following the discovery of proteins that function solely in the autophagic degradation of mitochondria , it soon became clear that mitophagy has evolved as a unique quality control mechanism. By sequestering dysfunctional (or sometimes redundant) mitochondria into double membrane autophagosomes and trafficking them to lysosomes, cells are able to neutralize any threat posed by the faulty powerhouses. Over the last 10 years, developments in techniques such as live-cell imaging have allowed us to make exciting progress in the study of mitophagy dynamics and regulation.
Mitophagy has been studied in a number of different systems both in vitro and in vivo and it has emerged that mitochondrial degradation can be initiated by different environmental and developmental cues. It is now clear that mitophagy is not solely a defence mechanism triggered by mitochondrial damage. By exploring the molecular mechanisms of mitophagy in more detail, we can begin to understand how it has evolved as an essential feature of both cellular quality control and adaptation.
Mitophagy in yeast
Mitophagy in yeast depends on the delivery of mitochondria to the acidic lysosome-like vacuole for degradation. Yeast initiate mitophagy in response to a number of conditions including nitrogen starvation and rapamycin treatment . They have proven to be important in demonstrating that mitophagy can proceed via tightly regulated mechanisms distinct from the common autophagy pathway. Proteins such as the OMM (outer mitochondrial membrane) protein Uth1p were first demonstrated to regulate mitophagy independently of autophagy under certain conditions . Subsequent genetic screens identified another OMM protein, Atg32, as being essential for mitophagy [6,7]. In order to guide ill-fated mitochondria to the autophagosome, Atg32 binds directly with the resident autophagosome protein Atg8 and also interacts with the previously identified selective autophagy adaptor protein Atg11  (Figure 1). Positive regulation of the Atg32–Atg11 interaction requires MAPK (mitogen-activated protein kinase)-dependent phosphorylation of Atg32 at Ser114 . This phosphorylation event is critical to mitophagy as mutation of the serine residue results in abolishment of the mitophagy response to nitrogen starvation . Although Atg32 has no known higher eukaryote homologue, these seminal studies have certainly removed any doubts surrounding the selective nature of mitophagy.
Work in yeast has also advanced our understanding of how mitophagy is regulated during shifts in metabolism. For example, activation of mitophagy following an exchange of respiratory to fermentative conditions appears to signify a housekeeping role for mitophagy as redundant mitochondria are selected for degradation .
Mitophagy during erythropoiesis
A striking example of mitophagy being employed to remove healthy yet redundant mitochondria occurs during mammalian red blood cell development (erythropoiesis). Mature red blood cells are devoid of mitochondria following the selective elimination of their entire mitochondrial population by a distinct mitophagy pathway . During normal erythroid terminal differentiation, expression of the OMM BNIP3-like protein Nix is augmented  and this is necessary for complete mitochondrial depletion [12,13]. Importantly, an absence of Nix causes defective erythroid maturation and anaemia in mice .
Similar to Atg32 in yeast, Nix confers the selectivity for mitochondrial engulfment by the autophagosome. As discussed in Chapter 7 of this volume, Nix directly interacts with autophagosomal marker proteins LC3 (light-chain 3) and GABARAP (γ-aminobutyric acid receptor-associated protein) via its LIR (LC3-interacting region) motif  allowing the recruitment and formation of autophagosomes around the mitochondria (Figure 2). Interestingly, however, rescue of mitophagy in Nix−/− maturing blood cells is only partially suppressed by mutation of the LIR motif and recently a newly characterized 16-amino-acid cytoplasmic region of Nix called the MER (minimal essential region) has been described as critical for Nix activity . Although LC3 does not interact with the MER , identification of other possible mitophagy-related interaction partners will improve our mechanistic understanding of mitochondrial clearance in developing red blood cells. Besides interacting directly with autophagosome components, Nix can also regulate mitophagy by facilitating mitochondrial membrane depolarization and ROS generation . A Nix-regulated increase in ROS production may provide a key step in mitophagy by enhancing autophagosome biogenesis via mTOR [mammalian (also known as mechanistic) target of rapamycin] inhibition .
Nix-orchestrated mitophagy during erythropoiesis is arguably an example of pre-programmed mitochondrial clearance. How then do mammalian cells react to unexpected mitochondrial damage in order to preserve a healthy mitochondrial population? This question appears to be particularly significant in respiring differentiated cells such as neurons and hepatocytes. Given the pathological consequences of mitochondrial dysfunction in such tissue, extensive research has begun to elucidate the mechanisms that exist to selectively target and eradicate damaged mitochondria.
Parkin-mediated mitophagy and beyond
Mutations in the E3-ubiquitin ligase parkin  and PINK1 [PTEN (phosphatase and tensin homologue deleted on chromosome 10)-induced putative kinase 1]  were found to cause autosomal recessive forms of PD (Parkinson's disease); however, it was not initially clear how these two proteins contributed to the severe pathogenesis. The spotlight was cast on a possible role in mitochondrial quality control when in vivo studies demonstrated that these mutations manifest as mitochondrial dysfunction. Drosophila harbouring PINK1/parkin mutations suffer dopaminergic neuron loss and locomotion defects, associated with abnormal mitochondria [19,20]. Subsequently, elegant Drosophila genetics were employed to demonstrate that parkin functions downstream from PINK1 in the same pathway [20,21].
The question remained of how mutations in PINK1 and parkin might result in the accumulation of dysfunctional mitochondria. An answer was provided when Narendra et al.  discovered that parkin normally functions to target damaged mitochondria for degradation by mitophagy. Live-cell imaging of cultured cells expressing fluorescently tagged parkin revealed the remarkably dynamic recruitment of cytosolic parkin to damaged mitochondria. These parkin-decorated mitochondria are then sequestered into autophagosomes and delivered to lysosomes for degradation. Astoundingly, many parkin-overexpressing cells can clear their entire mitochondrial population after the induction of global mitochondrial damage by treatment with mitochondrial depolarizing drugs.
Questions remain of how physiologically relevant such wholesale examples of mitophagy are, not least because more bioenergetically active cells have been demonstrated to resist parkin-mediated mitophagy in response to depolarizing drugs . In that study, cells with a greater dependence on oxidative phosphorylation were found to inhibit parkin translocation to depolarized mitochondria . Important discrepancies also exist between the functions of endogenous and overexpressed parkin . Further work examining the role of endogenous parkin, particularly in disease-vulnerable neurons, is essential. Surprisingly, parkin-knockout mice do not display hallmark PD phenotypes such as reduced motor ability and decreased neurological function . It remains unclear why parkin-deficient fly and mouse models vary so much in the severity of Parkinsonism symptoms. Nevertheless, both in vitro and in vivo experiments continue to be essential for further dissection of the parkin/PINK1 mitophagy pathway.
In a healthy mitochondrion, PINK1 is rapidly turned over by constitutive proteolytic processing. However, following insult and dissipation of the inner membrane potential (mtΔΨ), which is required to drive ATP production and mitochondrial protein import, PINK1 becomes stabilized on the OMM permitting it to recruit its binding partner parkin [26,27]. PINK1 interaction with parkin is necessary for parkin self-association and activation of parkin's ubiquitin ligase activity . Once activated, parkin ubiquitinates targets on the OMM allowing the clustering of damaged mitochondria and their engulfment by autophagosomes (Figure 3).⇓
Parkin is not the only ubiquitin ligase shown to direct mitophagy. Increased activity of the ubiquitin ligase Mul1 results in enhanced mitophagy during skeletal muscle wasting. Suppression of Mul1 maintains mitochondrial number and partially prevents muscle wasting in mice exposed to muscle-wasting stimuli . Finally, gp78 (glycoprotein 78) is another E3 ubiquitin ligase recently identified to drive depolarization-induced mitophagy in a parkin-independent manner . Interestingly, gp78 is ER (endoplasmic reticulum)-associated and, upon mitochondrial depolarization, its ubiquitination activity recruits LC3 to engulf mitochondria in close proximity to the ER .
Exactly how autophagosomes are recruited to parkin-decorated mitochondria remains an intriguing question. The adaptor protein p62 can bind both ubiquitin and LC3 and is found to accumulate on mitochondria that have been polyubiquitinated by parkin. Despite this, although a role for p62 in mitochondrial clustering looks likely , conflicting reports exist as to whether p62 is actually essential for the recruitment of autophagosomes to mitochondria (or vice versa) [31,32]. As we have seen with Atg32 in yeast and Nix in mammalian red blood cells, mitophagy does not always depend on cytosolic adapter proteins for recruitment of the autophagosome. Indeed, hypoxia-induced mitophagy also requires the direct interaction between an integral OMM protein FUNDC1 and autophagosomal LC3. FUNDC1 binds LC3 via its LIR motif and this interaction is enhanced on FUNDC1 dephosphorylation during hypoxic conditions . Hypoxia-induced mitophagy therefore offers another mechanism whereby mitochondria are targeted for mitophagy by regulated and direct interaction between OMM and autophagosome proteins.
Regardless of targeting, further work is certainly required to elucidate the actual origins of the ‘mitophagosome’. As discussed in Chapter 3 of this volume, several organelles have so far been implicated in autophagosome biogenesis including the mitochondria themselves . It has also been suggested recently that the autophagic isolation membrane can form de novo on depolarized mitochondria as opposed to being recruited from elsewhere in the cytoplasm . There is much left to learn about the role played by mitochondria in autophagosome biogenesis, however, it is now apparent that mitochondria can also regulate their own degradation by strict control of their morphology. The remainder of the present chapter focuses on the exciting progress being made in the understanding of how, and why, mitophagy is regulated by mitochondrial dynamics.
Mitophagy regulation by mitochondrial dynamics
It is impossible to discuss mitophagy without exploring the regulatory role played by mitochondrial dynamics. Mitochondrial fusion and fission have been implicated in many of the classical mitochondrial-associated cellular pathways including calcium signalling, apoptosis and the cell cycle. To explore how mitochondrial dynamics regulates mitophagy we must first introduce the fission and fusion machinery.
Mitochondrial dynamics: a balance between fusion and fission
Mitochondrial fission is driven by the dynamin-related protein Drp1 , and OMM proteins Fis1  and Mff . Unlike Fis1 and Mff, the GTPase Drp1 is a cytoplasmic protein that localizes to sites of mitochondrial fission in a rigorously regulated manner. Drp1 is the target of several key post-translational modifications including sumoylation, ubiquitination and phosphorylation . These modifications either enhance or disrupt Drp1 activity in a number of ways such as by regulating Drp1 translocation to the mitochondria or by mediating Drp1 assembly into spiral structures at the future sites of mitochondrial division.
The mitochondrial fusion machinery also consists of three key proteins: the OMM GTPases Mfn1 (mitofusin 1) and Mfn2  and the IMM (inner mitochondrial membrane) protein OPA1 (optic atrophy 1) . Although structurally similar, Mfn1 and Mfn2 have been associated with separate functions. Of note, Mfn2 tethers mitochondria to the ER at contact sites that have been shown recently to play an important role in autophagosome biogenesis . On the IMM, OPA1 is regulated by constitutive and inducible proteolytic cleavage and besides driving IMM fusion, OPA1 also serves to remodel cristae; key for ATP production and apoptosis .
Linking fission with mitophagy
It is generally accepted that mitochondrial fission is a pre-requisite for mitophagy in many mammalian cell types. As others have pointed out, it is logical for an elongated mitochondrion of over 5 μm in length to be chopped up before engulfment by 0.5-μm-diameter autophagosomes . But beyond this, it is clear that the fission and fusion machinery offer invaluable quality control and can help determine whether a mitochondrion is destined for degradation. Twig et al.  first demonstrated elegantly that fission can generate fragmented mitochondria with a low membrane potential and reduced levels of OPA1. These mitochondria therefore have an inherent inability to re-fuse with the healthy mitochondrial network. Following their expulsion from the mitochondrial network, the mitochondria are finally removed by mitophagy.
In parkin-mediated mitophagy, parkin translocation to depolarized mitochondria triggers the ubiquitination and proteasome-dependent degradation of the mitofusins , thus further isolating dysfunctional mitochondria from the rest of the network. In other pathways, Mfn2 is targeted by the muscle-wasting mitophagy ubiquitin ligase Mul1  and both Mfn1 and Mfn2 are ubiquitinated by the mitophagy regulator gp78 . Therefore obstructing mitochondrial fusion by mitofusin degradation appears to be a common step in mitophagy initiation.
Mitochondrial parkin recruitment also causes degradation of the kinesin motor adaptor protein Miro . This detaches kinesin from the mitochondrial surface leading to an arrest of mitochondrial motility that may facilitate mitophagy. Such a mechanism is particularly significant in neurons where mitochondrial transport is key for the distribution of functional mitochondria along axons to synapses. So by preventing the transport of damaged and potentially toxic mitochondria to these vital domains, PINK1 and parkin may be limiting the risk posed by dysfunctional mitochondria. In addition, Miro localizes to ER–mitochondrial contact sites and it remains to be seen whether disruption of ER–mitochondrial tethering is a key step in initiating mitophagy .
Fusion protects mitochondria from mitophagy
Given the importance of mitochondrial fission in assisting mitophagy, is it conceivable that mitochondria can resist mitophagy under certain conditions by inhibiting fission and promoting fusion? Owing to metabolic and bioenergetic demands, it sometimes becomes the case that mitophagy must be kept in check in order to maintain a respiring mitochondrial population. For example, mitophagy is blocked in yeast that are fed on a carbon source making mitochondria essential for metabolism . An exciting possibility is that the metabolic status of a cell can feed into the mitochondrial dynamics machinery. Evidence for this comes from recent reports that describe how mitochondria undergo fusion and are spared from mitophagy during starvation and thus are able to maintain ATP levels. Cells can therefore protect their precious mitochondria during starvation-induced macroautophagy by maintaining an elongated mitochondrial network [49,50]. In both studies it was found that starvation-induced activation of PKA (protein kinase A) leads to inhibitory phosphorylation of fission factor Drp1 at Ser637 and dephosphorylation at Ser616. This blocks the translocation of Drp1 to the mitochondria and hence inhibits mitochondrial fission.
Despite recent advances, there is a long way to go if we are to truly understand the cross-talk between mitochondrial dynamics and mitophagy and how this affects disease pathology. Mitochondrial elongation depends on the type of nutrient starvation indicating that tight control of mitochondrial dynamics depends on intricate signalling pathways yet to be properly defined . It is also important to bear in mind that mitochondrial networks differ from cell type to cell type. In yeast, for example, mitophagy occurs independently of mitochondrial fission . Interestingly, on simulation of fasting by nutrient deprivation in the presence of glucagon, mitophagy is actually enhanced in GFP–LC3-expressing primary mouse hepatocytes . Morphologically, the mitochondria do not appear fused in these hepatocytes and autophagosomes are even found to form around mitochondria that have an intact membrane potential . This provides strong evidence that mitochondrial depolarization is not always a pre-requisite for the selective removal of mitochondria.
Mitophagy is fundamental to mitochondrial quality control and homoeostasis and its importance is demonstrated by the pathological consequences of its mis-regulation. Since its initial recognition as a bona fide example of a specific autophagy pathway, the sheer diversity of mitophagy regulatory mechanisms continues to surprise. Cells regulate mitochondrial degradation not only through control of the autophagy machinery, but also via delicate tuning of mitochondrial fusion and fission. It remains to be seen whether other cellular processes linked with mitochondria also have a role to play in mitophagy regulation. Calcium, for instance, has emerged as a key modulator of autophagy . Given the central role played by mitochondria in calcium signalling , it too could have important functions in mitophagy control.
Better understanding of the mechanisms involved in mitochondrial quality control will hopefully hold serious therapeutic potential for the number of devastating human diseases associated with mitochondrial dysfunction.
• Mitophagy is the selective degradation of dysfunctional or redundant mitochondria by the autophagy machinery.
• Disturbances in mitochondrial quality control have been implicated in a range of human pathologies including neurodegeneration.
• Mitophagy-specific proteins exist that target mitochondria for degradation by directly or indirectly interacting with autophagosomes, these include Atg32 in yeast and Nix in mammalian red blood cells.
• The PINK1/parkin pathway selectively targets damaged mitochondria and its disruption has been implicated in early-onset forms of PD.
• Mitochondrial dynamics play a key role in regulating mitophagy; fission isolates damaged mitochondria from the network, whereas fusion protects healthy mitochondria from engulfment by autophagosomes.
- © The Authors Journal compilation © 2013 Biochemical Society