Correspondence: Nadine Camougrand, CNRS, Institut de Biochimie et de Genetique Cellulaires (IBGC) (UMR 5095), Université de Bordeaux 2, 1 rue Camille Saint-Saens, 33077 Bordeaux Cedex, France. Tel.: +33 55 699 9045; fax: +33 55 699 9051; e-mail: firstname.lastname@example.org
Mitochondria are essential for oxidative energy production in aerobic eukaryotic cells, where they are also required for multiple biosynthetic pathways to take place. Mitochondria also monitor and evaluate complex information from the environment and intracellular milieu, including the presence or absence of growth factors, oxygen, reactive oxygen species, and DNA damage. It follows that disturbances of the integrity of mitochondrial function lead to the disruption of cell function, expressed as disease, aging, or cell death. It has been assumed that the degradation of damaged mitochondria by an autophagy-related pathway specific to mitochondria (mitophagy), recently found to be strictly regulated, is a fundamental process essential for cell homeostasis. Until now, the main role of mitophagy has been tentatively defined as a ‘house-cleaning’ pathway that allows to eliminate altered mitochondria, but mitophagy may also play a role in the adaptation of the number and quality of mitochondria to new environmental conditions. In yeast, recent data defined two categories of mitophagy actors: ones constitutively required for mitophagy and those with mitophagy-regulatory functions. Situations were also uncovered in normal physiology in which cells utilize mitophagy to eliminate damaged, dysfunctional, and superfluous mitochondria to adjust to changing physiological demands.
Table 1. Basic autophagic nomenclature used in the review
Catabolic process involving the degradation of a cell's own components through the lysosomal/vacuolar machinery
Random cytosolic material including organelles and long-lived proteins is delivered into vacuoles for degradation
The target for vacuolar degradation in this case is specifically selected; it can be one type of organelle (peroxisome, mitochondria, endoplasmic reticulum, ribosomes), or long-lived protein
Sequestration of a portion of cytosol, organelles, or long-lived proteins in a double-membrane vesicle, called an autophagosome, followed by the fusion of the outer membrane of the autophagosomes in the cytoplasm with the vacuole, where their contents are degraded via acidic hydrolases; can be selective or nonselective
Occurs when the vacuole directly engulfs cytoplasmic material by invagination, protrusion, and/or septation of the vacuolar limiting membrane; can be selective or nonselective
Selective vacuolar degradation of mitochondria by an autophagy-related process, which can be separated into macromitophagy and micromitophagy
Selective degradation of mitochondria by macroautophagy
Selective degradation of mitochondria by microautophagy
Although nonselective autophagy plays an essential role in survival by nitrogen starvation in providing amino acids to the cell, selective autophagy is more likely to play a role in the maintenance of cellular structures, occurring under normal conditions as a ‘housecleaning’ process, and under stress conditions by eliminating altered organelles and macromolecular structures (Tolkovsky et al., 2002; Kundu & Thompson, 2005; Lemasters, 2005; Tolkovsky, 2009). Selective autophagy where the target is the mitochondria may be particularly relevant to stress conditions. The mitochondrial respiratory chain is both the main site and the target of reactive oxygen species (ROS) production. Consequently, the maintenance of a pool of healthy mitochondria is a crucial challenge for the cells. The progressive accumulation of altered mitochondria caused by the loss of efficiency of the maintenance process (degradation/de novo biogenesis) is often considered a major cause of cellular aging (Xue et al., 1999; Jazwinski, 2005; Sanz et al., 2006; Terman et al., 2006; Yu & Chung, 2006).
A brief history of mitophagy
The process of selective removal of mitochondria by autophagy was termed mitophagy by Lemasters in 2005, following decades of research. The first report illustrating the possible presence of mitochondria within lysosomes and single- or double-membrane-bounded vesicles (later called autophagosomes) in mammalian cells came from the study of cellular differentiation in the kidneys of newborn mice by Clark (1957). A few years later, Ashford & Porter (1962) observed that the incidence of lysosomes in rat hepatocytes following their exposure to glucagon was higher than normal, and their typical distribution in cells was altered. The peculiar feature of these glucagon-treated cells was that mitochondria were the most common objects identified in encapsulation and lysis. In addition to cytoplasmic components, small vesicles, and ribonucleoprotein particles, almost every lysosome contained mitochondria in various stages of breakdown or hydrolysis. Because the mitochondria, with the exception of those in sequestration organelles, did not appear to be changed, it was difficult to judge whether mitochondria experienced these changes before or only after their sequestration. Beaulaton & Lockshin (1977), studying the ultrastructural changes in intersegmental muscles that occur during programmed cell death in the metamorphosing muscles of the silk moth, suggested that mitochondria could be selectively removed during insect metamorphosis. They remarked that while other organelles looked completely normal, mitochondria developed some functional and morphological alterations, which could activate autophagy, resulting in the almost exclusive engulfment of mitochondria by autophagosomes. Observations of Decker & Wildenthal (1980) suggested that subcellular damage caused by hypoxia and especially recovery from hypoxia can induce important lysosomal responses in the rabbit heart, including ‘controlled’ autophagy of damaged organelles, especially mitochondria, or fusion of autophagic vacuoles with adjacent degenerating mitochondria, depending on the length of the hypoxic episode and reoxygenation. A quantitative ultrastructural study of normal rat erythroblasts and reticulocytes revealed that they completely eliminate their mitochondria (along with the reduction in the content of other organelles such as the endoplasmic reticulum, Golgi apparatus, and ribosomes) during maturation, and suggested that mitochondria are recycled through an autophagy-related process, specifically by their uptake into autophagosomes in late erythroblasts and in reticulocytes (Heynen & Verwilghen, 1982). The regulated clearance of mitochondria remained a well-recognized, but poorly understood aspect of cellular homeostasis for a long time. Recently, it was found that Bcl-2-related protein NIX is required for programmed mitochondrial clearance during reticulocyte maturation (Schweers et al., 2007; Sandoval et al., 2008).
Years of detailed studies on the fate of mitochondria in rat hepatocytes after serum deprivation provided Lemasters et al. (1998) with the opportunity to demonstrate that the opening of the mitochondrial permeability transition pore (mPTP) and the loss of mitochondrial membrane potential (ΔΨm) trigger the induction of autophagy. Subsequent studies revealed that autophagy could selectively remove mitochondria (Elmore et al., 2001). Because the mPTP opening can cause the release of toxic proteins, the authors suggested that autophagy of mitochondria is a protective device used by cells to eliminate harmful/damaged organelles. Lemasters also introduced a model that initiated research on the connection between autophagy and apoptosis, whereby autophagy could serve as a cytoprotective mechanism. For example, in order to preserve the efficiency of the cell, autophagy can target damaged organelles, but if the damage is too extensive, the cell may undergo apoptosis. It has been shown that when cells were induced to undergo apoptosis, but then were prevented from carrying out apoptosis by the concomitant inhibition of caspases and subsequently returned to their normal growth environment, there followed a period during which the entire cohort of mitochondria were selectively eliminated from the cells (Xue et al., 1999, 2001). Interestingly, it was proposed that mitochondria are responsible for initiating their own death.
Analysis of autophagy in the yeast system dates back 18 years, when it was demonstrated that the autophagy morphology in yeast was similar to that documented in mammals (Takeshige et al., 1992). This result was fundamental to enable further studies in this genetic model organism. This same study reported the presence of mitochondria in autophagosomes sequestered with other cytoplasmic constituents following autophagy induction by nitrogen starvation in the presence of a fermentative carbon source of glucose, suggesting a nonselective manner of mitochondrial removal in yeast under these conditions (Takeshige et al., 1992). Vacuole-dependent and selective uptake of mitochondria has been found in yeast Δyme1 mutant cells growing on an ethanol/glycerol medium (Campbell & Thorsness, 1998). Yme1 protein is mitochondrial AAA ATPase that controls the escape of mtDNA from the mitochondria to the nucleus, in the budding yeast Saccharomyces cerevisiae. In its absence, an increased rate of escape of mtDNA in cells was accompanied by the appearance of ‘pinched’ and fragmented mitochondria adjacent to invaginations at the surface of the vacuole. Thus, YME1 was the first identified gene where complete absence triggered a specific mitochondrial turnover. Whether autophagy is involved, and what kind of autophagy, was unclear.
A key point in our understanding of the selective mitochondrial degradation by autophagy (mitophagy) only occurred following analyses of yeast cells shifted to nitrogen starvation or in the stationary phase under respiratory conditions (Kiššováet al., 2004, 2006a, 2007; Tal et al., 2007; Zhang et al., 2007; Kanki & Klionsky, 2008; Deffieu et al., 2009; Journo et al., 2009; Kanki et al., 2009a, b; Okamoto et al., 2009). In the yeasts, the mitochondrial function, morphology, and the extent of autophagic activity depend on the growth conditions. On glucose, yeast obtains energy by fermentation, and the Crabtree effect prevents the differentiation of fully functional mitochondria. However, on a respiratory carbon source (lactate or glycerol/ethanol), mitochondrial respiration is essential for growth and viability, mitochondrial biogenesis is vigorous, and, as a result, cells contain a high number of optimally differentiated mitochondria. Thus, elimination of mitochondria during growth in glucose and in a respiratory carbon source may have to involve various/distinct mechanisms and may also have diverse consequences on the cells. The results obtained have allowed the definition of two categories of actors in mitophagy: ones constitutively required for mitophagy and those with mitophagy regulatory functions.
Mitophagy and ATG genes
The first autophagic mutant, designated as Δatg1, and defective in the accumulation of autophagic bodies in the vacuoles of yeast cells grown in various nutrient-deficient media was isolated by selection using a light microscope by Tsukada & Ohsumi (1993). Seventeen years of focused effort to conduct different genetic screens for autophagy have allowed the identification of 33 autophagy-related genes (ATG) to date. The ATG gene products are needed for the Cvt pathway, nonspecific macro- and microautophagy, and also for selective autophagic pathways.
About half of the Atg proteins form the core machinery essential for the biogenesis of autophagosomes. Other Atg proteins adapt this core machinery to the needs of autophagic subtypes. In S. cerevisiae, the core Atg proteins colocalize at the site of autophagosome biogenesis [autophagosome assembly site (PAS)] and form functional complexes, which assemble hierarchically. The serine/threonine protein kinase, Atg1p, and its regulatory factor Atg13p, form a basic scaffold for PAS assembly during starvation. In rich media, Atg13p is phosphorylated depending on the rapamycin-sensitive TOR kinase. Upon induction of autophagy, Atg13p is dephosphorylated, and together with other proteins, interacts with Atg1p. This stimulates Atg1 kinase activity and modulates the autophagic response. The formation of autophagosomes is initiated by class III phosphoinositide 3-kinase and Atg6p. The combination of several other Atg proteins and two conjugation events is essential for the process by which the double membrane enwraps and sequesters cytosol during autophagy. The first event is the covalent attachment of Atg12p to Atg5p in a ubiquitin-like conjugation process in association with Atg7 and Atg16 proteins. This multiconjugate complex is directly required for vesicle formation and is also a prerequisite for membrane association of Atg8p in a second conjugation reaction. Atg8p mediates membrane tethering and is essential for the formation of normal-sized vesicles. The source of lipids for membrane expansion at the PAS is largely unknown. There are hints of an involvement of the mitochondria, the trans-Golgi network, and the endocytic system cluster (Farre et al., 2009).
At the time the existence of the process of mitophagy was discovered, the possible involvement of Atg proteins in the process was tested, revealing the requirement for Atg1,5–9,12–13 proteins (Kiššováet al., 2004, 2007; Tal et al., 2007; Zhang et al., 2007). However, the absence of these proteins disabled all forms of autophagy. Some Atg proteins are required only for selective autophagy, as in the case of Atg11p, an essential adaptor or scaffold protein for pexophagy (Kim et al., 2001). It was found that the Atg11 protein was also indispensable for mitophagy induced during nitrogen starvation or in the stationary phase under respiratory conditions (Kanki & Klionsky, 2008; Okamoto et al., 2009). The role of this protein was to recruit the degradation-sentenced organelle to the PAS. Very recently, by performing screens on a yeast knockout library for mitophagy-deficient mutants, two different research groups identified the same protein, named Atg32p (Kanki et al., 2009a; Okamoto et al., 2009). This mitochondrial protein was essential for mitophagy, but was not required, in contrast to Atg11p, for other types of selective autophagy or for nonspecific macroautophagy. Furthermore, data suggested that the Atg32 protein interacts with Atg11p and Atg8p, and may act as a mitochondrial receptor specific to mitophagy. Another mitochondrial protein, Atg33, was observed to detect or present aged mitochondria for degradation at the stationary phase. In addition, some Atg proteins specific for bulk autophagy and the Cvt pathway were not essential for mitophagy. These results suggested that mitophagy is a bona fide autophagy-related process that is mechanistically distinct from nonspecific autophagy and the Cvt pathway. Interestingly, there is some discrepancy in the results from these two independent screens possibly because of the different conditions under which the screen was performed (the stationary phase for Okamoto's study vs. a shift to nitrogen starvation media with glucose for Kanki's study), and this can suggest a variation in the mechanism of mitophagy induced under different conditions.
Recent progress in the understanding of mitophagy in yeast indicated that the initial step of the molecular mechanism for the delivery of mitochondria into vacuoles is the interaction of Atg32 and Atg11 proteins. Nevertheless, how this interaction is regulated, what induces it, and the identity of the signal transduction proteins/molecules relating to mitochondrial function(s) involved in this pathway remain unclear. Two potential candidates to be evaluated are mitochondrial proteins Uth1p and Aup1p, both known to be required for the regulation of mitophagy induction under certain conditions.
Regulatory mitophagy players
The first evidence that mitophagy in yeast is genetically controlled has been obtained by work on the UTH1 gene (Camougrand et al., 2004; Kiššováet al., 2004, 2007). The UTH1 gene was identified in two independent genetic screens: the first one was based on the isolation of strains exhibiting an increased life span during starvation stress (Kennedy et al., 1995), and later, it was also identified in the screen designed to find yeast genes, in the absence of which cells showed altered sensitivities to oxidative damage (Bandara et al., 1998). The protein encoded by this gene possessed a double cellular localization (outer mitochondrial membrane and cell wall), and its absence promotes pleiotropic effects at different cellular levels such as aging, oxidative stress response, and mitochondrial biogenesis. Moreover, a strain in which the UTH1 gene was deleted exhibited resistance to the heterologous expression of the mammalian proapoptotic protein Bax (Camougrand et al., 2003). Interestingly, the absence of Uth1p also conferred resistance to rapamycin and nitrogen starvation under both respiratory and fermentative conditions (Kiššováet al., 2004). Under respiratory conditions, nitrogen starvation as well as rapamycin treatment induced early autophagic removal and degradation of mitochondria. Using different approaches including fluorescent and electron microscopy studies in a wild-type strain and a strain with a deleted UTH1 locus, the existence of two distinct processes was revealed for mitochondrial autophagy. The first, a mitochondria-specific process, did not appear in the absence of Uth1p, and was characterized by the early occurrence of direct contacts between mitochondria and vacuoles, possible fusion between these two organelles, and the later appearance of vacuolar vesicles containing almost exclusively mitochondria with only residual cytosol. The second process was present in both wild-type and mutant strains and involved a nonexclusive engulfment of the mitochondria, with a significant proportion of surrounding cytosol at the vacuolar surface, as readily expected from nonselective microautophagy. As a consequence of the deficiency in the first process, the overall mitochondria autophagy appeared to be poorly efficient in the uth1 mutant, but it was not abolished completely (Kiššováet al., 2007; Camougrand et al., 2008). Based on these observations, it is evident that Uth1p has a function in mitophagy distinct from that observed for Atg32p, where its absence prevented mitophagy entirely. The autophagic machinery is fully functional in the absence of Uth1p (Kiššováet al., 2004). It is not clear whether the lack of the Uth1 protein itself is the signal for mitophagy or whether other proteins participate in the essential signal response chain by acquiring abnormal or novel functions in the absence of the primary proteins.
Abeliovich and colleagues have identified and characterized Aup1 protein, a yeast mitochondrial protein phosphatase homolog that functionally interacts with the autophagy-dedicated protein kinase Atg1p (Tal et al., 2007). The absence of Aup1p does not cause a defect in starvation-induced nonspecific macroautophagy; however, Aup1p is required for efficient mitophagy in prolonged stationary-phase incubation in a medium containing lactate as a carbon source. The reduction in mitophagy in the Δaup1 mutant cells is correlated with reduced viability and rapamycin hypersensitivity, suggesting a prosurvival role for Aup1-directed mitophagy.
Aup1p is also involved in the phosphorylation of Pda1p, the α-subunit of the pyruvate dehydrogenase complex (Gey et al., 2008), which is responsible for the formation of acetyl-CoA, and is essential for the tricarboxylic acid (TCA) cycle and respiration. The mitochondrial TCA cycle participates in nitrogen assimilation through the amination of oxaloacetate and α-ketoglutarate, and therefore, mitochondrial malfunction (in the stationary phase) may lead to effects on nitrogen balance and autophagy induction. Recently, it was reported that the onset of mitophagy in the stationary phase leads to the concomitant induction of retrograde pathway target genes (RTG) in an Aup1-dependent fashion (Journo et al., 2009). The authors have suggested that the function of Aup1p in mitophagy could be explained through its regulation of Rtg3-dependent transcription, but additional components will need to be identified in order to understand the specific mechanism by which defective (oxidative damage accumulating) mitochondria are targeted for degradation. One possibility is to test the involvement of mitochondrial Atg33p in Aup1-mediated mitophagy.
Both Uth1p and Aup1p are mitochondrial proteins with localization in different mitochondrial compartments (the outer membrane or the intermembrane space, respectively) that are required for mitophagy, but unnecessary for starvation-induced bulk autophagic activity. Whereas Uth1p seems to be constitutively expressed in yeast cells under various growth conditions (logarithmical and stationary phase, starvation) (Camougrand et al., 2004), the Aup1p level peaks sharply at the beginning of the stationary phase, correlating with the onset of mitophagy (Tal et al., 2007). Interestingly, their effects on the physiological role of the process appear to be different: knockout of UTH1 results in increased viability of cells transferred from a nitrogen-replete medium to a nitrogen starvation medium, both containing lactate as a carbon source, or treated with rapamycin under the respiratory condition, while the absence of AUP1 results in decreased viability of cells in the late stationary phase during respiratory growth. Whether this difference is simply due to noncomparable assay conditions or whether it reflects some fundamental process addressing the roles of each protein remains ambiguous. Interestingly, neither Uth1p nor Aup1p has been found to be essential for the selective degradation of mitochondria in post-logarithmic-phase cells under respiratory conditions (Okamoto et al., 2009) or during nitrogen starvation in glucose-containing media after a shift from complete media supplemented with a respiratory carbon source (Kanki et al., 2009a). It is not known whether Uth1p and Aup1p interact with Atg32p, proposed to be a mitochondrial receptor specific for mitophagy, or some of the other autophagy-related proteins and directly mediate mitophagy. It has been proposed that there is a nonselective UTH1-independent form of mitophagy in yeast (Kiššováet al., 2007), but whether this is Atg32p dependent, and drives the bulk of mitophagy (by microautophagy), also remains to be examined.
Induction and regulation of mitophagy
Mitophagy and cell physiology
There are several conditions shown in yeast that can induce mitophagy: nitrogen starvation, rapamycin treatment, stationary phase, intracellular redox imbalance, and some alteration of the mitochondrial function, such as a decrease in the membrane potential or oxidation of mitochondrial lipids (Fig. 1). Interestingly, selective mitochondrial degradation was discovered for the first time under conditions known to induce nonspecific autophagy: nitrogen starvation and rapamycin treatment (Kiššováet al., 2004). The absence of nitrogen does not induce mitophagy regularly/commonly, because a shift from a complete medium to starvation under sustained gluconeogenic conditions does not lead to mitophagy induction, contrary to the cells shifted to a starvation medium regardless of the carbon source (glucose or a respiratory substrate) after growth under the respiratory condition. It is important to remember that nonselective autophagy is induced to the same extent under both fermentative and respiratory conditions (Kiššováet al., 2004). Intriguingly, cells can possess several redundant mechanistically distinct/similar pathways to eliminate mitochondria, possibly converging to a common point, where alternative pathways can be utilized depending on the situation. Thus, vacuolar delivery of mitochondria in cells grown under respiratory conditions after their shift to a starvation medium with glucose is mediated by macroautophagy (Kanki et al., 2009a), while in starved cells under sustained respiratory conditions mitophagy is carried out in the microautophagic mode (Kiššováet al., 2007; Camougrand et al., 2008). Clearly, the inhibition of mitochondrial biogenesis is likely to interfere with mitochondrial clearance after the shift from a respiratory carbon source to a starvation under fermentative conditions and possibly has an effect on the mode of autophagy. A similar observation was reported in the case of peroxisomal degradation, which, depending on the carbon source in the medium, was effected through micropexophagy or macropexophagy (Sakai et al., 2006). The purpose of mitophagy after a shift from respiratory to fermentative conditions is without any doubt the removal of redundant organelles under new environmental conditions; the goal of such a specific process under starvation under respiratory conditions is not so obvious. There is a possibility that the degradation of some nonessential portion of mitochondria into molecular building blocks and their subsequent availability for biosynthesis means a lesser energetic expense than de novo synthesis for a cell. Starvation may affect mitochondrial metabolism, causing alterations that could trigger this process. The deletion of neither Atg32p nor Uth1p has been found to cause a defect in cell growth and cell death. Moreover, the absence of Uth1p rescues cells from starvation-induced death, suggesting that it is autophagy rather than mitophagy that is required for rescue from starvation.
Mitophagy and redox imbalance
Nonselective bulk autophagy and mitophagy are often induced by the same stimulus (nitrogen starvation and rapamycin), but it has been shown recently that these two events may be regulated independently (Bhatia-Kiššová & Camougrand, 2009; Deffieu et al., 2009). This conclusion was drawn based on the observation of cells starved for nitrogen in the presence or absence of various antioxidants after growth in a medium supplemented with a respiratory carbon source. Here, unlike other tested compounds, the addition of N-acetylcysteine (NAC) completely prevented mitophagy, while the general nonselective autophagic pathway was not affected. It has been demonstrated that the effect of NAC was not related to its ROS-scavenging properties, but rather to its fueling effect of the intracellular glutathione pool, which is indispensable for maintenance of the cellular redox balance (Deffieu et al., 2009). One of the critical factors regulating glutathione metabolism in living cells is the availability of its precursor l-cysteine, because only a small pool of this amino acid is available to sustain a much larger metabolically active glutathione pool. Although NAC can provide some benefit in preventing or reducing the toxicity related to moderate oxidative stress during nitrogen starvation, it was proposed that the availability of cysteine is a fundamental factor in the regulation of glutathione production. Apparently, an increase in the glutathione concentration in starved cells observed after NAC addition does not affect nonselective autophagy, but it has a remarkable effect on the delivery and degradation of mitochondria, as these processes are completely blocked. By manipulating glutathione, it has been possible to identify the relationships between the redox status of the cells and mitophagy. The hypothesis that redox imbalance and intracellular glutathione content modulate mitophagy induction also confirms the observation showing that the addition of ethacrynic acid, known to decrease the glutathione pool in growth medium, accelerates mitophagy in cells when they reach the stationary phase (Deffieu et al., 2009). The stimulation of glutathione synthesis (by NAC) restores NAD(P)H/NAD(P) ratios and may accommodate significant changes in mitochondrial status, which would possibly relieve the requirement for mitochondria elimination. In fact, no selective mitophagy has ever been observed on glucose-grown cells (Kiššováet al., 2007), where the activity of the mitochondria, and the potential danger they might represent, is low. Mitophagy might be finely tuned by subtle changes in the concentration and compartmentalization of glutathione or regulated by thiol-containing proteins targeted by glutathione, and additional investigations are needed to refine these critical points. Moreover, the fact that glutathione is a known powerful cellular antioxidant, and at the same time decreases mitophagy, fits the notion that autophagy is specifically overactivated in the face of an injury to a particular cellular component. In mammals, it has been shown that a mitochondrial injury in CoQ-deficient fibroblasts is accompanied by an increase in mitochondrial degradation by mitophagy, which can be abolished by antioxidants (Rodríguez-Hernández et al., 2009).
Mitophagy and mitochondrial dysfunctions
In yeast, mitophagy can be induced under a nutrient-limited condition as an attempt to eliminate superfluous or eventually altered mitochondria, but was also reported to represent a cellular response designed to face stress conditions causing physiological mitochondrial dysfunction. In one study, the occurrence of the preferential degradation of mitochondria was reported as a consequence of impairing the mitochondrial electrochemical transmembrane potential (ΔΨm) and/or the lack of mitochondrial biogenesis, resulting from anaerobic and heat stress growth of an FMC1 null mutant in fermentable glucose-containing media (Priault et al., 2005). Under these conditions, the collapse of ΔΨm was due to the accumulation of aggregates of a dysfunctional F1 ATPase subunit, and the inability to use glycolytic ATP for the membrane potential maintenance, which leads to cell death. Based on electron-microscopy observations, both macro- and microautophagy were triggered under these conditions, and to understand which pathway mediated mitochondrial turnover will require further work. Another study has reported that alterations in mitochondrial ion homeostasis caused by the inactivation of the K+/H+ exchanger Mdm38p induced both autophagy and selective degradation of mitochondria by microautophagy (Nowikovsky et al., 2007). The deletion of MDM38 caused pleiotropic effects including mitochondrial swelling, loss of ΔΨm, fragmentation, and reduction of cell growth on nonfermentable substrates; however, the results indicate that K+/H+ exchange activity is primarily responsible for mediating mitophagy. Cells lacking Mdm38p were more likely to die in a stationary phase. It is not known whether this form of death was due to mitophagy/autophagy or whether there was a specific K+/H+ exchange defect that is not known. Recently, it was shown that Mdm38p-dependent mitophagy is mediated by Atg32p (Kanki et al., 2009a).
Based on these data, an alteration in the mitochondrial membrane potential can induce mitophagy in yeast. However, the fact that (1) the inactivation of the profission DNM1 gene avoided Mdm38p-mediated mitophagy, although mitochondria still have reduced ΔΨm (Nowikovsky et al., 2007) and (2) chemical disruption of ΔΨm does not induce autophagy or mitophagy (Kiššováet al., 2004; Kanki & Klionsky, 2008) suggests that in yeast, contrary to mammals, a change in the ΔΨm potential may influence, but may not be sufficient for mitophagy induction. In light of these results, mitophagy can serve as a possible cellular surveillance over the control of mitochondrial quality or as a final program initiated when cells cannot recuperate from mitochondrial damage.
In addition to the cases described above, the implication of mitochondrial fission in the mitophagic process has been reported in several other studies. The addition of rapamycin or nitrogen starvation induced fragmentation of the mitochondrial network, followed by selective mitochondrial degradation, but the inhibition of mitochondrial lipid oxidation (Kiššováet al., 2006a) or the maintenance of the cellular redox balance by glutathione pool renewal (Deffieu et al., 2009) that prevents mitophagy did not prevent the initial alteration of the mitochondrial network. In mammals, the inhibition of the fission machinery through DRP1 or FIS1 inactivation decreased mitochondrial autophagy, and resulted in the accumulation of oxidized mitochondrial proteins (Twig et al., 2008; Dagda et al., 2009). Furthermore, these findings suggest that fission, followed by selective fusion segregates dysfunctional mitochondria, allowing their removal by autophagy. Recently, DNM1 has also been identified in one screen as a gene whose absence reduced mitophagy in yeast (Kanki et al., 2009b). Actin and other cytoskeletal proteins act as binding partners for Dnm1p; therefore, it is not surprising that the actin cytoskeleton has been found to be required for some selective types of autophagy (Reggiori & Klionsky, 2005). Further analyses are needed to test the involvement of actin in the regulation of mitophagy.
Mitophagy and aging
Incubation of yeast cells during a prolonged stationary (late logarithmic) phase, especially in the case of growth under aerobic conditions, gives rise to the accumulation of damaged mitochondria. They can generate ROS, release cell death-inducing factors such as cytochrome c into the cytosol, or generally burden the metabolic machinery of the cell by decreasing the efficiency of ATP generation. Thus, the ability to recognize and degrade such impaired mitochondria is a vital task in cells. It has been shown that these conditions induce massive and specific autophagic mitochondrial removal, and several genes have been identified as key players in this process (Tal et al., 2007; Kanki & Klionsky, 2008; Deffieu et al., 2009; Kanki et al., 2009a, b; Okamoto et al., 2009). The function of one protein required for mitophagy under this condition, Aup1p, is required for the survival of cells through a prolonged stationary phase. The absence of another one (Atg32p) does not have a discernible effect on cell viability, and the lack of another involved protein (Uth1p) even improves cell survival. This suggests that each protein can manage a different function in this process, and it is not known whether these proteins occur in a common or a linked generic pathway. Mitochondria under these conditions seemed to be removed by a macroautophagic pathway, including the generation of autophagosomes containing only selected organelles, mitophagosome (Kanki et al., 2009a; Okamoto et al., 2009).
Mitophagy, ROS, and cell death
It has been proposed that although autophagy is actually a prosurvival process, an abnormally high level of autophagy might be responsible for cell death (Kundu & Thompson, 2005; Pattingre et al., 2005). The treatment of yeast cells with rapamycin induces autophagy even under nutrition-rich conditions (Cardenas et al., 1999) and leads to growth arrest and surprisingly rapid loss of viability in both the respiratory and the fermentative case (Kiššováet al., 2004). This drug was also found to induce early selective mitochondrial autophagy by the microautophagic pathway, when added in media supplemented with lactate, and it was entirely prevented by the deletion of the UTH1 gene (Camougrand et al., 2004; Kiššováet al., 2004). Further studies have revealed that rapamycin-induced autophagy was accompanied by the early production of ROS and by the early oxidation of mitochondrial lipids. Inhibition of these oxidative effects by the hydrophobic antioxidant resveratrol largely impaired autophagy of both cytosolic proteins and mitochondria, and delayed subsequent cell death, suggesting that mitochondrial oxidation events may play a crucial role in the regulation of autophagy (Kiššováet al., 2006a). There is evidence that autophagy can be regulated by ROS as supported by data derived from studies in HeLa cells (Scherz-Shouval et al., 2007), where it was demonstrated that starvation stimulates ROS production, namely H2O2, which was essential for autophagy. Furthermore, the cysteine protease hsAtg4 was identified as a direct target for oxidation by H2O2. Another study has reported that inactivation of catalase induced programmed cell death accompanied by autophagy and elimination of altered mitochondria (Yu et al., 2006). Mitochondrial lipid oxidation is not only an indicator of mitochondrial dysfunction, but can also be a hallmark for altered mitochondria because, in contrast to ROS, oxidized lipids are not expected to diffuse rapidly throughout the cell, including to other mitochondria, especially if the mitochondrial network is fragmented as it occurs early during autophagy (Kiššováet al., 2004, 2006a). The combination of mitochondria fragmentation and mitochondrial lipid oxidation might together be an adequate signal indicating that the autophagic machinery should be activated. The fact that the inhibition of lipid oxidation by hydrophobic antioxidants limits autophagy lends further support to a crucial role of membrane lipid oxidation as a trigger for the degradation of altered mitochondria. The poor preventative ability of hydrophilic ROS scavengers suggests that the role of soluble ROS in the signaling remains marginal. The impact of mitophagy on cell survival under this peculiar condition when autophagy itself seems to cause cell death remains unknown. It is also not known whether Atg32p mediates rapamycin-induced mitophagy.
Genomic screens for yeast mutants defective in selective mitochondria autophagy have revealed 53 mitophagy-related genes where absence resulted in a partial or a complete block of the process (Kanki et al., 2009a, b; Okamoto et al., 2009; Kanki & Klionsky, 2010). In addition to 30 mutants that are known to be defective in autophagy and/or the Cvt pathway, another 23 cytosolic and mitochondrial mutants obtained have not been implicated and are involved in diverse cellular processes. Future analyses can lead to a better understanding of the key players in the mechanism(s) and the regulation of mitophagy.
Mitochondria form complex and dynamic networks within cells and are, in fact, a single population when it comes to exchange of genetic information in a cell (Nakada et al., 2001). However, it is possible that at a certain level/amount of mitochondrial injury, there is no complementation of function by the remaining ‘healthy’ mitochondria, and the cell wants to eliminate damaged organelles. Based on the latest data, autophagy-dependent degradation of mitochondria is a fundamental process conserved from yeast to humans. It is conceivable that mitochondria are the primary target of selective autophagy, because they accumulate oxidative damage due to their own byproducts, ROS, and that clearance of dysfunctional mitochondria is important for organelle quality control. Consistent with this idea, the loss of mitochondrial functions promotes organelle degradation and defects in autophagy cause mitochondrial abnormalities, the latter of which might ultimately be associated with cell aging and death. Interestingly, Parkin and PINK1, both Parkinson's disease-associated proteins, promote elimination of impaired mitochondria, which highlights a possible link between mitophagy and neurodegenerative disorders (Narendra et al., 2008, 2010; Dagda & Chu, 2009; Dagda et al., 2009; Vives-Bauza et al., 2010). Moreover, complete autophagic mitochondrial clearance is essential for erythrocyte and reticulocyte maturation, implying a role of mitophagy in cell differentiation (Zhang & Ney, 2010). Accumulation of dysfunctional mitochondria also causes mitochondrial autophagy in yeast. In this organism, mitophagy also takes place in cells during starvation or cultivation in the late stationary phase under respiratory conditions. Furthermore, some results support the view that activation of mitophagy, as a process that protects against mitochondrial alteration, is a means to convert a necrotic form of death into a regulated form of death (Kiššováet al., 2006b).
These reported data indicate that induction and regulation of mitophagy are very complex processes that can exhibit modifications in the cell's attempt to assume the most favorable position in accommodating the different physiological states of the cell and changing environment. Further understanding of the molecular mechanisms underlying the regulation and selectivity of the mitophagic process may prove to be relevant to therapeutic strategies in mitochondrial-related disorders. The riveting challenge for scientists also includes the questions to what extent can mitochondria be removed from cells to still preserve normal cell functioning and what are the components and signalling pathway(s) involved in monitoring and evaluation in cells?
This work was supported in part by grants from Slovak Ministry of Education Contract VEGA1/0264/08, Science and Technology Assistance Agency Contract APVV-0024-07, the CNRS, the Université de Bordeaux 2, the Conseil Régional d'Aquitaine, and Association pour la Reserche contre le Cancer grant 3812. The France/Slovakia collaboration was supported by Egide ECO-NET 10187VF.