Autophagy is a fundamental cellular process promoting survival under various environmental stress conditions. Selective types of autophagy have gained much interest recently as they are involved in specific quality control mechanisms removing, for example, aggregated proteins or dysfunctional mitochondria. This is considered to counteract the development of a number of neurodegenerative disorders and aging. Here we review the role of mitophagy and mitochondrial dynamics in ensuring quality control of mitochondria. In particular, we provide possible explanations why mitophagy in yeast, in contrast with the situation in mammals, was found to be independent of mitochondrial fission. We further discuss recent findings linking these processes to nutrient sensing pathways and the general stress response in yeast. In particular, we propose a model for how the stress response protein Whi2 and the Ras/PKA (protein kinase A) signalling pathway are possibly linked and thereby regulate mitophagy.
- general stress response
- mitochondrial dynamics
Quality control of mitochondria by selective autophagy
Autophagy is a highly conserved catabolic process in eukaryotic cells. It mediates the transport of proteins, lipids and even entire organelles into the vacuole/lysosome for hydrolytic degradation. Thus it allows the cell to adapt to changing nutrient conditions and to respond to different types of stress. Based on morphological characteristics two types of autophagy can be distinguished: macro- and micro-autophagy. During macroautophagy cytoplasmic components are sequestered via double-membrane compartments, termed autophagosomes, which then fuse with the vacuole/lysosome. Microautophagy involves the direct uptake of the components by the vacuolar/lysosomal boundary membrane [1,2]. Autophagy can be bulk, that is, non-specific, but also selective types of autophagy have been reported, among them the Cvt (cytoplasm-to-vacuole targeting) pathway, pexophagy or mitophagy [3–5]. Studies in yeast and other fungi have identified 35 autophagy-related (ATG) genes so far. Most of them are required for both selective and non-selective autophagy, but some, such as Atg11, Atg20 and Atg24, are specifically required for selective types of autophagy [6,7].
The selective degradation of mitochondria via autophagy is termed mitophagy. The mechanism and regulation of this process are only poorly understood. Two groups in parallel performed a genomic screening for yeast mutants defective in mitophagy using a library of non-essential deletion strains [8,9]. They identified approximately 40 genes involved in diverse pathways, such as membrane trafficking, protein modification/degradation, lipid metabolism or mitochondrial metabolism. Atg32 was found, in both screenings, to promote mitophagy. It is not required for bulk autophagy or other types of selective autophagy. Atg32 has been proposed to act as a mitochondrial receptor that during mitophagy interacts with Atg11, a known adaptor protein for selective types of autophagy, and thereby recruit the autophagy machinery to mitochondria [9,10]. In mammalian cells, PINK1 and parkin, both linked to Parkinsons's disease, are involved in targeting depolarized mitochondria to the autophagosome . The mitochondrial kinase PINK1 recruits the E3 ubiquitin ligase parkin to the mitochondrial outer membrane, mediating ubiquitination of mitochondrial proteins such as VDAC (voltage-dependent anion channel) or mitofusin [12,13]. During terminal erythrocyte differentiation, NIX has been identified as a selective receptor for mitophagy interacting with LC3 proteins, essential components for autophagosome formation .
Mitochondria form a highly dynamic network constantly undergoing fusion and fission events [15,16]. Several neurodegenerative diseases such as Parkinson's disease or autosomal dominant optic atrophy type I are associated with alterations in mitochondrial dynamics . It has been proposed that fusion and fission promote mitochondrial quality control . On the one hand they allow mitochondrial content mixing and thereby contribute to the integrity and homogeneity of the mitochondrial network. In addition, mitochondrial dynamics could isolate dysfunctional mitochondria from the intact network, helping to target them for autophagic degradation. Indeed, in mammalian cells mitophagy is impaired when mitochondrial fission is blocked and dysfunctional mitochondria accumulate . However, in yeast the role of mitochondrial dynamics in mitophagy and mitochondrial quality control has been debated over the last few years.
The function and morphology of mitochondria are linked
It has been proposed that mitochondrial dynamics and morphology are linked to the bioenergetic state of mitochondria; but how this is regulated on a molecular level remained unclear for quite some time. The dynamin-like GTPases Mgm1 (in yeast) and OPA1 (in mammals) were found to be key players in this process. Both are essential for mitochondrial fusion in the respective organisms [20,21]. Mgm1 exists in two isoforms, which are present in roughly equal amounts and which are localized in the inner membrane/intermembrane space [22,23]. The short isoform (s-Mgm1) is generated by limited intramembrane proteolysis by the rhomboid protease Pcp1, whereas the large isoform (l-Mgm1) is only processed by matrix processing peptidase, removing the N-terminal mitochondrial targeting sequence [23,24]. Interestingly, both isoforms have to be present to allow respiratory growth and to maintain mitochondrial DNA and mitochondrial morphology [23,25]. The formation of s-Mgm1 was shown to depend on a functional protein import machinery and on the ATP level in the matrix. Reduced ATP level led to a decreased level of s-Mgm1, increased levels of l-Mgm1 and to fragmentation of mitochondria in vivo . A model for alternative topogenesis of Mgm1 was proposed linking the bioenergetic state of mitochondria to the ratio of the two Mgm1 isoforms and consequently to mitochondrial morphology.
Similarly, in mammalian cells different isoforms of OPA1 regulate mitochondrial dynamics in response to bioenergetics. The OPA1 protein exists in at least five mitochondrial isoforms. Mitochondrial dysfunction, for example, induced by dissipation of the membrane potential, leads to an increased proteolytic processing of the large OPA1 isoforms into the small OPA1 isoforms and subsequently to mitochondrial fragmentation [27,28]. Taken together, alternative topogenesis of Mgm1 in yeast and stress-induced proteolytic processing of OPA1 in mammals provide a molecular mechanism to specifically inhibit fusion competence when mitochondrial function is impaired. This would help us to distinguish functional from dysfunctional mitochondria based on their morphology. On the one hand this would lead to a spatial separation of damaged mitochondria from the intact network to minimize further damage. On the other hand it could enable the removal of these isolated mitochondria presumably by mitophagy. Indeed, mitophagy was shown to be increased in various systems of mitochondrial dysfunction in mammals and in yeast [19,29–31].
Mitophagy in yeast is independent of mitochondrial fission
In case segregation of dysfunctional mitochondria from the remaining intact network would target them for degradation, mitochondrial fission is predicted to be a prerequisite for mitophagy. Indeed, Twig et al.  could show that in mammalian cells Drp1-mediated fission is required for the degradation of mitochondria and that blocking of autophagy led to accumulation of oxidized mitochondrial proteins . In yeast, conflicting results on the role of fission in mitophagy have been reported. In particular the involvement of the fission-promoting dynamin-like GTPase Dnm1, the orthologue of mammalian Drp1, remains under debate. In a genomic screening, Δdnm1 was found to show reduced levels of mitophagy, whereas the deletion of the other known fission factors Fis1, Mdv1 and Caf4 did not show such impairment . Induction of mitophagy as observed on loss of Mdm38 was prevented by deletion of DNM1, further supporting the idea of the requirement for fission . However, a second screening did not identify a role of Dnm1 or the other fission factors in mitophagy . Recently, we investigated the effect of altered mitochondrial dynamics on mitophagy in detail, applying enzymatic, biochemical and fluorescence-based assays . This study demonstrated that fragmentation of mitochondria alone was not sufficient to trigger mitophagy. Furthermore, drug-induced inhibition of oxidative phosphorylation also did not induce mitophagy. When mitochondrial fission was blocked by expressing a dominant-negative variant of Dnm1, the level of mitophagy induced by rapamycin was not altered compared with the wild-type control. In addition, strains lacking one of the fission factors Dnm1, Mdv1 or Caf4 did not show an impairment of mitophagy on rapamycin treatment, suggesting that fission is not essential for mitochondrial degradation. However, mitophagy and (to a lesser extent) also autophagy were reduced in the strain deleted for the fission factor Fis1. How can one explain this observation? Does Fis1 have a second, fission-independent role in mitophagy or does the Δfis1 strain contain a secondary suppressor mutation in a gene required for mitophagy? Indeed, a secondary point mutation in the WHI2 locus introducing a premature stop codon could be identified in the Δfis1 strain used in this study, consistent with an earlier report . Moreover, rapamycin-induced mitophagy was reduced in a Δwhi2 strain resembling the effect in the Δfis1 strain containing the secondary whi2 mutation. Expression of Whi2 but not Fis1 could complement the mitophagy- and autophagy-deficient phenotype of Δfis1, showing that the reduced level of mitophagy in the Δfis1 strain was actually caused by the secondary loss-of-function mutation in WHI2. Taken together, several lines of evidence show that mitophagy is truly independent of mitochondrial fission in yeast. Furthermore, we identified Whi2 as a novel mitophagy-promoting factor .
These results differ from findings in mammalian cells and throw into question the hypothesis of mitochondrial dynamics as quality control checkpoints (Figure 1). However, it cannot be ruled out that basal levels of mitophagy depend on mitochondrial fission and might be too little to be detected by the assays. Additional triggers might be needed in addition to fragmentation for inducing fission-dependent mitophagy. Furthermore, an alternative, Dnm1-independent fission mechanism might exist that is able to mediate the spatial separation of individual mitochondria for degradation. Thus it cannot be fully excluded that mitochondrial morphology is important for the selection of dysfunctional mitochondria or that it is involved in signalling of mitochondrial dysfunction to induce mitophagy in yeast. But how can the degradation of mitochondria occur without a preceding fission event? One possibility is that mitochondrial fission is mediated by an unknown mechanism or the autophagy machinery itself. It is also conceivable that the mitochondrial network in yeast creates tubules that are small enough to be entirely surrounded by the autophagosomes in contrast with the larger mammalian network. The different possibilities are summarized in Figure 1(B).
Whi2 links mitophagy to the general stress response
The level of mitophagy was shown to be reduced by ~40% in cells deleted for WHI2 compared with the wild-type control after treatment with rapamycin . In contrast, autophagy was only slightly inhibited in Δwhi2 cells under the same conditions. These observations point to a specific role for Whi2 in mitophagy. Furthermore, the Cvt pathway seems to be unaffected in strains lacking Whi2, suggesting no general function of this factor in other selective types of autophagy. Interestingly, overexpression of Whi2 seems to be able to positively modulate autophagic flux.
To date, not much is known about this novel mitophagy-promoting protein. Still, a few studies carried out over the last 30 years indicate that the fungi-specific protein Whi2 links proliferation and stress response to environmental sensing mechanisms. Sudbery et al.  isolated the whi2 mutant and found that this gene is involved in cell cycle control. The Δwhi2 cells continue proliferation during stationary phase and fail to arrest in the G1-phase of the cell cycle. They retain the properties of exponentially growing cells and their arrest is randomly distributed in the cell cycle. Consequently, the cells become abnormally small and are predominantly budded compared with wild-type cells in stationary phase. Moreover, these cells exhibited a lower viability, failed to accumulate storage carbohydrates and became less resistant to environmental stresses such as heat shock [35,36]. WHI2 transcriptional levels were shown to be increased in the fermentative phase of growth (glucose) and to be drastically reduced in non-fermentative, aerobic growth conditions (ethanol) similar to the transcription levels of genes required for glycolysis . However, the expression of Whi2 is sensitive to the growth rate rather than to catabolite repression. Moreover, Whi2 seems to be a negative regulator of catabolite repressible functions. Thus cells deleted for Whi2 respired more actively in the presence of glucose and grew more rapidly on glycerol . The mRNA levels of the G1 cyclins CLN1 and CLN2 were found to be increased in the Δwhi2 strain during exponential phase and to persist longer at high levels during stationary phase than in the wild-type cells. Otherwise the overexpression of CLN1 in stationary phase wild-type cells resulted in a Δwhi2-like phenotype, confirming that Whi2 negatively regulates G1 cyclin expression . Interestingly, Radcliffe et al.  revealed that overexpression of Whi2 leads to filamentous growth of yeast cells, suspecting also a function in budding and cytokinesis.
The identification of a regulatory function of Whi2 in the general stress response provided novel insights into the cellular role of Whi2 . It was shown that Whi2 was necessary for full activation of gene expression controlled by STREs (stress-responsive elements). Furthermore, Whi2 can interact with the transcription factor Msn2 and the plasma-membrane-associated phosphatase Psr1, both necessary for the activation of stress response genes. Msn2 was found to be hyperphosphorylated in strains deleted for Whi2 or Psr1/2, suggesting a role for Whi2 in mediating the dephosphorylation of transcription factors by specific phosphatases.
Another study points to a regulatory role of Whi2 in the Ras/PKA (protein kinase A) pathway . In Δwhi2 cells, actin aggregation, mitochondrial fragmentation, dissipation of membrane potential, ROS accumulation and loss of viability were observed during diauxic shift. This was shown to result from the hyperactivation of the PKA pathway. Moreover, Ras2 was found to be abnormally activated and located to mitochondria, indicating the requirement for Whi2 for deactivating and degrading Ras2 during nutrient depletion.
Taken together, Whi2 seems to link cell cycle regulation, general stress response and the PKA pathway and thereby specifically modulates the regulation of mitophagy. Indeed, autophagy was shown to be regulated by highly conserved signalling pathways such as the TOR (target of rapamycin) pathway or the PKA pathway [6,43]. These two kinases negatively regulate autophagy either by directly phosphorylating components of the initial Atg1–kinase complex or by inactivation of transcription factors preventing the expression of autophagic genes. In addition, inhibition of TOR via rapamycin promotes the nuclear localization of Msn2 and Msn4, resulting in the activation of the general stress response . It has also been shown that Msn2 subcellular localization was regulated by PKA activity and that TOR seems to signal through the PKA pathway to modulate stress response and induction of autophagy [45,46].
Altogether, Whi2 may be a key player in adapting the complex network of signalling pathways in response to the nutritional status of the cell (Figure 2). We propose a model according to which Whi2 is responsible for inactivation of the PKA pathway and activation of the general stress response during nutrient depletion, thereby inducing autophagy and mitophagy. Mutants lacking Whi2 are not able to inactivate PKA and/or to activate Psr1/2 when cells enter stationary phase. Thus Msn2 becomes hyperphosphorylated and is not capable of inducing the general stress response. Mitophagy seems to be selectively inhibited under these conditions, whereas the level of autophagy remains affected only modestly.
The finding that Whi2 selectively influences mitophagy has shed new light on the regulation of this process. The decreased mitophagic response in cells lacking Whi2 after inhibition of the TOR kinase with rapamycin  resembles the previously observed failed response of these cells to certain stresses such as nutrient depletion in stationary phase [35,36]. In light of all these findings, mitochondrial dysfunction and fragmentation observed in Δwhi2 during diauxic shift  could well be explained by reduced elimination of damaged mitochondria. Also, the observed activation of Ras2 might be a direct consequence of this, consistent with a recent report by Graef and Nunnari  showing that mitochondrial dysfunction negatively regulates autophagy via activation of PKA. Interestingly, earlier studies have shown that hyperactivated PKA inhibits early steps of autophagosome formation . In mammalian cells, PKA was reported to phosphorylate and inactivate the mitochondrial fission factor DRP1 . This resulted in hyperfused mitochondria that are less prone to be degraded by autophagy, demonstrating that PKA may inhibit mitophagy at various levels. Overall, several lines of evidence indicate an important role of Whi2 in regulating the PKA signalling and the general stress response pathway. It remains an open question how this is exerted mechanistically and why mitophagy is specifically promoted by Whi2.
The molecular mechanisms and the regulation of mitophagy are far from being understood yet. Furthermore, the molecular factors are only partly conserved between yeast and mammals. In the latter, mitochondrial dynamics seem to play an important role in mitophagy, ensuring quality control of mitochondria. However, in yeast, fission of mitochondria does not appear to be required for mitophagy. Possibly, here fission is only required under certain conditions yet to be identified, fission is exerted by unknown mechanisms, or fission is indeed generally dispensable for mitophagy. However, the fact that mitophagy is controlled by a complex network of signalling pathways is conserved between yeast and mammals. In particular, the role of the Ras/PKA signalling pathway has become more evident from recent studies. Future studies have to concentrate on deciphering the complex interplay between the pathways involved in regulating mitophagy, which will finally allow us to gain a better molecular understanding of mitochondrial quality control.
We acknowledge financial support from the Deutsche Forschungsgemeinschaft through the Cluster of Excellence Frankfurt ‘Macromolecular Complexes’ at the Goethe University Frankfurt [grant number EXC 115]; the Deutsche Forschungsgemeinschaft [grant number RE-1575/1-1]; and the Bundesministerium für Bildung und Forschung project GerontoMitoSys [grant number 0315584A].
We apologize for not having cited the work of many colleagues because of space limitations.
8th International Meeting on Yeast Apoptosis: An Independent Meeting held at Keynes College, University of Kent, Canterbury, U.K., 2–6 May 2011. Organized and Edited by Paula Ludovico (University of Minho, Braga, Portugal).
Abbreviations: Cvt, cytoplasm-to-vacuole targeting; PKA, protein kinase A; STRE, stress-responsive element; TOR, target of rapamycin
- © The Authors Journal compilation © 2011 Biochemical Society