Yeast Mitophagy: Unanswered questions

Abstract

Superfluous and damaged mitochondria need to be efficiently repaired or removed. Mitophagy is a selective type of autophagy that can engulf a portion of mitochondria into a double-membrane structure, called a mitophagosome, and deliver it to the vacuole for degradation. Mitophagy has significant physiological functions from yeast to human, and recent advances in yeast mitophagy shed light on the molecular mechanisms of mitophagy, especially the regulation of mitophagy induction. This review summarizes our current knowledge about yeast mitophagy and considers several unsolved questions, with a particular focus on Saccharomyces cerevisiae.

I. Introduction

Mitochondria, the “energy plant” of the cell, sit at the hub of metabolism and signaling pathways [1]. In the process of aerobic ATP synthesis, reactive oxygen species (ROS) generated from the electron transport chain could damage mitochondrial DNA and proteins, and ultimately components in the rest of the cell [2–4]. Dysfunctional mitochondria will disrupt cellular homeostasis, and if the defect is not repaired or the damaged mitochondria properly cleared, pro-apoptotic CYCS (cytochrome c, somatic) can be released into the cytosol and ultimately lead to cell death in mammals [5–8]. Therefore, abnormal mitochondrial quality control has been linked with many diseases, such as Parkinson disease, Alzheimer disease, diabetes, and cancer [9–11]. Furthermore, mitochondria are energetically costly to maintain; in organisms such as yeast, superfluous mitochondria are removed by mitophagy following a shift to fermentative conditions [12–14]. Cells have developed several sophisticated mechanisms to ensure the healthy state of mitochondria, including the control of fission-fusion dynamics and the clearance of dysfunctional mitochondria [15–17].

Macroautophagy (hereafter called autophagy) is an evolutionarily conserved catabolic process, in which transient double-membrane compartments engulf cytoplasmic macromolecules and damaged or superfluous organelles, then deliver them to the lysosome (in mammals) or vacuole (yeast and plants) for degradation [18]. Mitophagy, a selective type of autophagy, is essential for maintaining mitochondria homeostasis [12–14]. Although many core autophagy proteins (i.e., those that are needed for formation of the double-membrane autophagosome) are required for mitophagy, cells must utilize some unique molecular strategies to overcome the additional challenges imposed when mitochondria are the specific cargos: First, mitophagy machinery needs to recognize dysfunctional, but not healthy mitochondria. This ability to differentially sequester a subset of the organellar population requires the presence of specific molecular markers on the surface of damaged mitochondria; these receptors bridge the damaged mitochondria with the core autophagy machinery. Furthermore, unlike cytosolic macromolecules, mitochondria are relatively large double-membrane organelles. Thus, how can the mitophagy machinery incorporate such cargo into phagophores, the sequestering precursor compartments that mature into autophagosomes? Finally, as with any membranous organelle, let alone the double-membrane mitochondria, there is the problem of breaking down the cargo after vacuolar delivery—how do vacuolar proteases and lipases function to degrade mitochondria? On a related note, what mechanisms are used to clear accumulated ROS (Fig 1A, 1B)?

Fig 1.

Overview of yeast mitophagy.

(A) Stationary phase-induced mitophagy and nitrogen starvation-induced mitophagy are two common protocols used to trigger yeast mitophagy.

(B) The proper induction of mitophagy requires selective labels on mitochondria destined for clearance, followed by recruitment of autophagy machinery. The phagophore engulfs part of a mitochondrion and the resulting mitophagosome delivers it into the vacuole via vacuole-mitophagosome fusion. Finally, the superfluous or damaged mitochondria will be degraded by vacuolar hydrolases, and the breakdown products will be released into the cytosol.

Yeast provides a convenient system to dissect the molecular mechanism of mitophagy [19–25]. Mitophagy’s physiological functions in yeast include the maintenance of redox balance, maintenance of mitochondrial morphology and preservation of genome stability, in addition to the regulation of cellular lipid homeostasis [26–30]. Extensive efforts have been made to study the mechanism of mitophagy in budding yeast (Saccharomyces cerevisiae); however, recent studies showed that similar regulatory mechanisms are also conserved in other fungi such as the fission yeast Schizosaccharomyces pombe, the methylotrophic budding yeast Komagataella phaffii/Pichia pastoris, the entomopathogenic yeast Beauveria bassiana, the human pathogenic yeast Candida glabrata, and the diary yeast Kluyveromyces marxianus [31–38]. In this review, we summarize recent progress in the yeast mitophagy field, with a particular focus on Saccharomyces cerevisiae, and discuss some fundamental unanswered questions.

II. How is mitophagy induced in yeast?

To trigger mitophagy in budding yeast, two different types of protocols are widely used: In the first scenario, the yeasts are pre-cultured in a medium with a non-fermentable carbon source (e.g., glycerol or lactic acid), then switched to glucose-rich, nitrogen-starvation medium. Because budding yeast mainly obtain energy from glycolysis in glucose-rich conditions (even under aerobic conditions), such a switch can drive the degradation of superfluous mitochondria to replenish the depleted amino acid pool resulting from nitrogen starvation [39–42]. Mitophagy can also be induced by culturing yeast in medium with a non-fermentable carbon source for 3-5 days to ensure that they reach stationary phase [43, 44]. In this case, constitutive high mitochondrial activity requires upregulated mitophagy to remove dysfunctional mitochondrial fragments (Fig 1A). Of note, the above protocols can induce robust mitophagy but only moderate nonselective autophagy, which is generally induced by directly switching yeast from nutrient-rich medium to nitrogen-starvation medium [45]. Alternatively, Benjamin Tu’s lab found that switching a prototrophic strain of S. cerevisiae from nutrient-rich medium to minimal medium could also induce autophagy in the absence of nitrogen starvation. Similarly, treatment with rapamycin, a TOR kinase inhibitor, can induce mitophagy. These protocols, however, trigger both nonselective autophagy and mitophagy, making the dissection of mitophagy-specific pathways challenging [42, 46–48]. Oxidative stress was suggested to be a key factor that induces mitophagy because the antioxidant N-acetyl-l-cysteine (NAC) can completely prevent mitophagy induction in both conditions [30]. However, NAC inhibits Atg32 expression, and other antioxidants such as ascorbic acid, resveratrol and tiron have no effect on mitophagy, suggesting a different interpretation of the effect of NAC treatment—enzymes that protect against ROS are very active in yeast; thus NAC may prevent mitophagy by fueling the glutathione pool rather than directly scavenging the ROS [30, 43]. A less common method for mitophagy induction involves the use of mutants such as fmc1Δ, mip1 ts or mdm38Δ that are defective in mitochondrial homeostasis [49].

A useful scheme for viewing selective autophagy is the ligand-receptor-scaffold model [50]. In this general model, the receptor contains an Atg8-family interacting motif/AIM (similar to the LC3-interacting region/LIR present in mammalian selective autophagy receptors). This short amino acid sequence binds Atg8–PE present on the phagophore membrane, to link the cargo with the forming autophagosome. In yeast, the Atg11 scaffold participates to facilitate this interactional[51, 52]. The first yeast mitophagy receptor was revealed from two independent genetic screens. Kanki et al. and Okamoto et al. used similar strategies, tagging mitochondrial proteins with GFP to generate a chimeric reporter, and monitoring the occurrence of mitophagy based on the release of free GFP within the vacuole; GFP is relatively stable in the presence of vacuolar proteases and is generally released intact when the chimeric protein is degraded. Both groups identified mitochondrial-anchored Atg32 as being essential for mitophagy, but not nonselective autophagy. Further analysis showed that Atg32 is localized on the mitochondrial outer membrane and can interact with the autophagy-related proteins Atg8 and Atg11, indicating that Atg32 is a mitophagy receptor that directly recruits autophagy machinery around superfluous or damaged mitochondria (Fig. 2A) [41, 43]. Atg32 is conserved in P. pastoris and C. glabrata, but not in S. pombe [34, 36]. Recently, Atg43 was identified as a mitophagy receptor in fission yeast [33]. Moreover, the mammalian homolog of Atg32, BCL2L13 (BCL2 like 13), has been linked with mitophagy, and can rescue the mitophagy defects in an atg32Δ yeast strain [53].

Fig 2.

The regulation of mitophagy induction in yeast.

(A) ScAtg32 is a single-transmembrane protein that contains 529 amino acids. Its binding affinity with Atg11 can be regulated by phosphorylation at serine 114 and Yme1-dependent C-terminal proteolysis.

(B) Transcriptional control of ATG32. The Ume6-Sin3-Rpd3 and Paf1 complexex directly inhibit ATG32 expression while several other factors can also influence the induction of ATG32. The transcriptional activators responsible for ATG32 induction remain unknown.

URS, upstream repression sequence.

How is Atg32 expression induced, and how is the protein transported on to the outer membrane of mitochondria? At the transcriptional level, nitrogen starvation leads to an approximately 10-fold increase in the ATG32 mRNA level [34]. Currently two transcriptional repressors have been reported to regulate ATG32 expression: the Ume6–Sin3–Rpd3 complex, a downstream effector of TOR signaling, and the Paf1 complex (polymerase-associated factor 1 complex, Paf1C) [34, 54]. However, neither of these two complexes affect Atg32 in a mitophagy-specific manner, and no direct Atg32 activator has yet been found. The N-terminal acetyltransferase A (NatA) complex has been positively linked with ATG32 mRNA induction during nitrogen starvation, but considering the fact that that ^~30% of all yeast proteins undergo N-terminal modifications from NatA, the precise molecular mechanism behind NatA’s positive regulatory role on ATG32 remains unclear [55, 56]. In addition, Opi3, an endoplasmic reticulum (ER)-localized phospholipid methyltransferase responsible for biosynthesis of phosphatidylcholine (PC) from phosphatidylethanolamine (PE), is partially required for induction of ATG32, but the molecular mechanism again remains to be elucidated (Fig. 2B) [57].

S. cerevisiae Atg32 is a 529 amino acid, single-transmembrane protein. The transmembrane domain is located between residues 389 and 411, and truncating the C terminus at residue 389 leads to a cytosolic distribution of Atg32, completely blocking mitophagy induction. Under mitophagy-inducing conditions, Atg32 forms complexes with Atg11, and such puncta are called mitophagy initiation sites [58]. The link between upstream mitophagy stimuli and the dynamic localization of Atg32 is an interesting topic to explore. Along these lines, it is incompletely understood what drives Atg32-Atg11 complex formation in dysfunctional mitochondria. Recent findings suggest that the phosphorylation of mitochondrial matrix proteins by Pkp1 and Pkp2 might regulate the mitochondria selectivity [59]. It is now well established that mitochondrial proteins are degraded at different rates [60], suggesting a mechanism for intramitochondrial selectivity involving mitochondrially localized kinases and phosphatases [59].

Another mechanism to govern mitophagy induction is phosphorylation. Atg32 can interact with Atg11 and Atg8 through its N terminus, but only the interaction with Atg11 is indispensable for mitophagy induction. Intriguingly, Atg32 is phosphorylated at serines 114 and 119 under conditions that induce mitophagy, and serine 114 phosphorylation significantly increases the binding affinity between Atg32 and Atg11 (Fig 2A) [61]. Casein kinase 2 (CK2) is responsible for serine 114 phosphorylation; however, the pro-mitophagy CK2 complex is ubiquitously expressed and constitutively active, making it unclear how this enzyme functions to modify Atg32 in a temporally appropriate manner [62]. Ppg1, a PP2A-like protein phosphatase, appears to antagonize CK2 activity when mitophagy is unnecessary. Together with the Far complex, Ppg1 can reverse Atg32 phosphorylation. Nonetheless, it is still not clear how could the Ppg1-Far complex together with CK2 maintain the necessary temporal and spatial control of mitophagy [63].

In addition to phosphorylation, Atg32 also undergoes other types of post-translational modifications [64–66]. Under mitophagy stimuli, the C terminus of Atg32 can be cleaved by the i-AAA protease Yme1. The cleavage of the Atg32 C terminus results in increased binding affinity between Atg32 and Atg11, thus promoting mitophagy induction; this finding indicates that the C terminus of Atg32 exhibits an inhibitory role. The specific Yme1-recognizing and cleavage sites and what factors trigger this proteolytic processing remain unknown [64]. Recent structural analysis hints that between residue 200 and residue 341 of Atg32 exists a pseudo-receiver (PsR) domain that is required for efficient C-terminal proteolysis, which would thus promote mitophagy induction [67]. Furthermore, ubiquitination at the lysine 282 residue in Atg32 has been reported. Although it is tempting to speculate that this modification might regulate Atg32’s stability, further analysis is required. For example, the ubiquitin E3 ligase and deubiquitination complex responsible for Atg32 ubiquitination have not been identified yet [66].

In addition, genetic studies have identified many mitophagy regulators that may act upstream of Atg32, includes MAPK-signaling pathway components Hog1 and Slt2, retrograde signaling pathway molecules Rtg3 and Ptc6/Aup1, the deubiquitination complex Ubp3-Bre5, the vacuolar K⁺/H⁺ transporter Mdm38, and the mitochondrial inner membrane protein Uth1 [42, 44, 68–72]. In all cases, however, their mechanistic connections with Atg32 regulation are poorly defined. Considering all of the above data, Atg32 is the only known mitophagy receptor in S. cerevisiae, and its expression and activity are subject to complex controls. In the next section, we discuss how active Atg32 can promote phagophore formation to engulf dysfunctional mitochondrial fragments.

III. How are mitochondria engulfed and delivered to the vacuole during mitophagy?

Although ultrastructural analysis shows that mitochondria may be degraded through microautophagy (i.e., direct uptake by protrusion or invagination of the vacuole limiting membrane) in yeast, the requirement of the core autophagy machinery for yeast mitophagy suggests that macroautophagy is involved, namely, the damaged mitochondria fragments would be engulfed by a phagophore that expands to generate a double-membrane autophagosome/mitophagosome; subsequent fusion of the autophagosome with the vacuole releases the mitochondrial fragments into the vacuole lumen for degradation and recycling [19–25].

Atg11 is a large cytosolic protein, containing 1178 amino acids, which forms a homo-dimer, acting as a crucial scaffold for various types of selective autophagy [73]. Atg11’s role in selective autophagy was first discovered in the cytoplasm-to-vacuole targeting (Cvt) pathway, a biosynthetic transport process that utilizes the autophagy machinery. In the Cvt pathway, the primary cargo protein, precursor aminopeptidase I (prApe1) binds to the Atg19 receptor; Atg11 bridges prApe1-Atg19 with Atg8, a ubiquitin-like protein that lines both sides of the phagophore and plays a critical role in cargo recognition, as well as with the Atg1-Atg13 complex [74, 75]. Atg11 contains 4 coiled-coil (CC) domains, and can physically interact with the Atg1 kinase via the second (CC2) and third (CC3) coiled-coil domains, thus activating Atg1’s kinase activity [76, 77]. Atg1 then activates the phosphatidylinositol-3-phosphate (PtdIns3P)-kinase complex, generating PtdIns3P in the phagophore membrane [78]. PtdIns3P allows the recruitment of downstream effectors such as Atg18 to promote autophagosome formation [79]. With regard to mitophagy, the CC4 domain mediates the interaction between Atg11 and Atg32’s N-terminal domain [61]. As mentioned above, Atg32 forms complexes at mitochondrial initiation sites, and this occurs in an Atg11-dependent manner; a recent analysis attributes this Atg32 concentrating effect to the CC3 domain [58].

The mitochondrial initiation sites are proximal to the vacuole membrane, presumably mediated by the interaction between the Atg1 complex and the vacuolar membrane protein Vac8 [80–82]. The Rab GTPase Ypt1 can interact with both Atg11 and Atg1, and may also contribute to the formation of mitochondrial initiation sites around the perivacuolar region [83–85]. The mechanism that allows the degradation of the distal damaged mitochondrial fragments via mitophagy is still unclear. The expansion of the phagophore requires a lipid supply, and 30- to 60-nm Atg9-containing vesicles can feed into the expanding phagophore [86]. This membrane incorporation process is facilitated by the interaction between Atg9 and Atg11 [87, 88]. For yeast mitophagy, the addition of Atg9-containing vesicles is likely regulated by a special structure called the ER-mitochondria contact site [89, 90]. Loss of subunits of the ER-mitochondria encounter structure (ERMES) complex results in the accumulation of immature mitophagosomes and defective mitophagy [91]. The ubiquitination on ERMES complex subunits Mdm12 and Mdm34 by the E3 ligase Rsp5 is required for mitophagy [92]. Interestingly, the defective mitophagy phenotype seen with mdm12 or mdm34 mutants can be rescued by overexpression of Vps13, a non-ERMES complex protein present in multiple membrane contact sites, indicating that the ERMES function in mitophagosome formation can be bypassed by other contact sites (Fig 3A) [93, 94]. Indeed, together with the mitochondrial protein Mcp1, Vps13 forms vacuole-mitochondria contact sites (vCLAMP) [95, 96]. Considering the close distance between the nascent mitophagosome and the vacuole, it is tantalizing to speculate that the latter could act as an additional lipid supply for mitophagosome biogenesis [97]. Because the Vps39-Tom40 complex establishes a functionally distinct vCLAMP, investigating the function of Vps39 and Tom40 in mitophagosome formation may shed light on the interaction between the vCLAMP and mitophagy [98].

Fig 3.

The formation, maturation of mitophagosome in yeast.

(A) Mitophagosome biogenesis occurs near the vacuolar membrane. Mitochondrial fission factors separate portions of mitochondria that can be engulfed by the phagophore. Mitochondrial contact sites contribute to phagophore expansion. Mitochondrial fragments can also be generated through the MDC pathway.

(B) The fusion between autophagosome and vacuole is achieved by specific SNAREs (Ykt6, Vam7, Vam3, and Vti1), the Rab-GTPase Ypt7 and the HOPS complex. The autophagic body’s membrane is degraded by the vacuolar lipase Atg15, whereas the lipase(s) responsible for mitochondrial membrane digestion are not known. The reducing equivalents that clear accumulated ROS also remain to be identified.

The shape and number of yeast mitochondria can vary dramatically depending on the growth conditions [99]. To remove superfluous or damaged mitochondria, fission machinery is proposed to be utilized to separate the damaged fragments from the remainder of the organelle to allow sequestration. The yeast fission factor Dnm1, a dynamin-related protein, is recruited to mitophagy initiation sites by Atg11, thus allowing the phagophore to engulf the isolated smaller mitochondrial fragments [100]. However, in both the nitrogen-starvation protocol and stationary-phase protocol, Dnm1 loss of function only causes a partial defect in yeast mitophagy, suggesting either the existence of alternative fission factors responsible for isolating the mitochondrial fragments, or the possibility that fission is not essential for mitophagy, but that it enhances the efficiency of the process [60, 101].

Recent studies imply that rather than degrade large portions of mitochondria through canonical mitophagy, selective mitochondria degradation can also be achieved by packing abnormal mitochondrial components into vesicle-like structures called the mitochondrial-derived compartment (MDC) (Fig. 3A). In aged yeast, the impaired vacuolar acidity triggers a substantial cleavage of Tom70-GFP, a mitophagy reporter, in a Dnm1-dependent and Atg32-independent manner [102]. In the future, it would be crucial to examine this MDC pathway using a standard mitophagy induction protocol in a strain with normal vacuolar function.

In summary, mitochondria fragments, rather than the entire organelle, are engulfed by a growing phagophore. The scaffold protein Atg11 bridges the mitophagy receptor Atg32 and the mitophagy machinery at the phagophore to initiate mitophagosome formation, with the lipid molecules required for membrane expansion being supplied in part by Atg9-containing vesicles. Mitochondria contact sites, including ER-mitochondria and vacuole-mitochondria contacts, are also involved in mitophagosome biogenesis, but the specific mechanisms remain to be explored. Fission machinery contributes to the isolation of selective portions of mitochondria destined for degradation, but fission may not be essential and alternative mechanisms might also exist.

IV. How is mitophagy completed in yeast?

The mitophagosome need to fuse with the vacuole to deliver the mitochondrial cargo into the vacuole lumen. Proteins such as Atg8 and lipid molecules including PtdIns3P on the outer membrane of the autophagosome inhibit the fusion between the autophagosome and the vacuole; thus, both need to be removed in the process of autophagosome/mitophagosome maturation. In yeast, this removal is facilitated by the cysteine protease Atg4 and the PtdIns3P phosphatase Ymr1 [103–107].

The fusion between the mature autophagosome and the vacuole is mediated by specific soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes. In yeast, one R-SNARE (Ykt6) and three Q-SNARE proteins (Vam3, Vti1 and Vam7) form parallel four-helix bundles, and together with the homotypic fusion and protein sorting (HOPS) complex and Rab GTPase Ypt7, fuse the outer membrane of the autophagosome with the vacuole membrane, leaving the autophagosome inner membrane (now called an autophagic body) within the acidic vacuolar lumen [108–110]. The vacuolar lipase Atg15 degrades the autophagic body membrane and exposes the cargo to various vacuolar hydrolases (Fig 3B) [111].

However, once exposed to the vacuolar lumen, the degradation of mitochondria is still not an easy task. The mitochondrial double membrane can protect against proteolytic activity. Thus, to efficiently clear mitochondrial proteins, vacuolar hydrolases need to first digest the membrane [112]. Atg15 is the only vacuolar lipase so far identified, and it is not known if Atg15 is responsible for degradation of the mitochondrial membrane; atg15Δ mutant strains cannot digest the autophagic body membrane resulting in an epistatic phenotype. Alternative vacuolar lipases responsible for degrading the mitochondria membrane may therefore exist. In addition, some of the mitochondrial proteins, such as the extremely hydrophobic Oli1/Atp9 F_o ATPase subunit, pose particular challenges in terms of proteolysis.

Damage to mitochondria often originate from excessive oxidative stress; thus, vacuoles need to have a mechanism to clear mitochondrially generated oxidants in the final step of mitophagy. Glutathione (L-Q-glutamyl-L-cysteinylglycine) is an important cellular reducing equivalent and its level can reach as high as 10 mM in yeast [113]. However, glutathione is unstable in the vacuole because it is degraded by Ecm38/Cis2, a γ-glutamyltranspeptidase. Thus, it is unclear how the vacuole can manage the ROS in damaged mitochondrial fragments [113–116]. Because mitochondria are responsible for Fe-S cluster and heme biogenesis in yeast, the fates of these metal ions in mitophagy is an intriguing question [117]. Delineating how the metal ions that result from mitophagy get recycled and transported out of the vacuole is yet another interesting research topic for the future [118].

Conclusions

This review depicts the current understanding about mitophagy in yeast. The current model establishes that superfluous or damaged mitochondria will be marked by the mitophagy receptor Atg32 on their surface, although how the cell is able to distinguish problematic from healthy mitochondria is still unknown. Once activated by phosphorylation at serine 114, Atg32 can recruit the scaffold protein Atg11 and form complexes around the mitophagy initiation sites. The Atg11 homodimer would then recruit additional autophagy proteins to the stage, which stimulate the expansion of the phagophore membrane. Atg9-containing vesicles provide some of the lipids for phagophore expansion, a process that also involves mitochondrial contact sites. The closed mitophagosome will undergo maturation in preparation for the fusion with the vacuole. The resulting autophagic body membrane is degraded by the vacuolar lipase Atg15 but how the mitochondrial fragments are digested is poorly understood. The clearance of mitochondrially generated oxidants and the recycling of breakdown products including metal ions are additional topics for future researchers to explore.

Highlights.

• Mitophagy can be induced by changing the culture medium in S. cerevisiae.

• Expression of the Atg32 mitophagy receptor of S. cerevisiae is tightly regulated.

• The scaffold Atg11 bridges Atg32 and the mitophagy machinery at the phagophore.

• Lipids for phagophore expansion are supplied in part by Atg9-containing vesicles.

• Similar regulatory mechanisms of mitophagy are conserved in other fungi.