Macromitophagy is a longevity assurance process that in chronologically aging yeast limited in calorie supply sustains functional mitochondria and maintains cellular lipid homeostasis

Macromitophagy is a longevity assurance process

As a first step towards addressing a role of selective macroautophagic mitochondrial removal in sustaining essential biological processes in yeast, we evaluated the importance of macromitophagy in longevity assurance. We compared the chronological lifespan (CLS) of wild-type (WT) strain to that of the single-gene-deletion mutant strain atg32Δ, which is impaired only in the mitophagic pathway of selective macroautophagy but not in other pathways of selective or non-selective macroautophagy [15, 16]. Both yeast strains were cultured in the nutrient-rich YP (1% yeast extract and 2% peptone) medium initially containing low (0.2%) concentration of glucose, a fermentable carbon source. Importantly, yeast cells grown under these longevity-extending caloric restriction (CR) conditions are not starving [18]. We found that the atg32Δ mutation substantially shortens both the mean and maximum CLS of yeast limited in calorie supply (Figure 1). Thus, macromitophagy is an essential longevity assurance process in chronologically aging yeast grown under CR on 0.2% glucose.

The essential role of macromitophagy in regulating yeast longevity under CR conditions is limited to this form of selective macroautophagy. Indeed, the single-gene-deletion mutation atg36Δ, known to impair only the pexophagic pathway of selective macroautophagy but not mitophagic or any other pathways of selective or non-selective macroautophagy [28], did not influence the mean or maximum CLS of yeast limited in calorie supply (Figure 2).

Macromitophagy is required for longevity extension by an anti-aging compound

Our recent study revealed that lithocholic acid (LCA), a bile acid, extends the CLS of yeast grown under CR on 0.2% glucose [19]. To assess a role of selective macroautophagic mitochondrial removal in longevity extension by LCA, we tested if this bile acid added to growth medium at the time of cell inoculation is able to extend longevity of the macromitophagy-deficient atg32Δ mutant. As we found, the atg32Δ mutation abolishes the ability of LCA to increase both the mean and maximum CLS of yeast limited in calorie supply (Figure 1). Hence, macromitophagy is required for longevity extension by LCA in chronologically aging yeast grown under CR on 0.2% glucose.

Unlike macromitophagy, the pexophagic form of selective macroautophagy is not obligatory for the ability of LCA to extend longevity of chronologically aging yeast under CR conditions. In fact, the exclusive impairment of macropexophagy by the single-gene-deletion mutation atg36Δ did not compromise the longevity-extending efficacy of LCA in yeast limited in calorie supply (Figure 2).

Macromitophagy defines the size and number of mitochondria, their shape and morphology, and their ability to exist as a network

To provide a mechanistic insight into the demonstrated essential role of macromitophagy in defining longevity of chronologically aging yeast under CR conditions, we used electron microscopy (EM) to compare the age-related dynamics of changes in the size, number and morphology of mitochondria in WT cells to that in macromitophagy-deficient atg32Δ cells. We found that in cells grown under CR on 0.2% glucose and recovered at different periods of CLS, the atg32Δ mutation (i) results in accumulation of greatly enlarged mitochondria (Figures 3A and 3B); (ii) substantially increases the number of mitochondria (Figures 3A and 3C); (iii) alters mitochondrial morphology by elevating the proportion of round-shaped mitochondria, especially in cells entered post-diauxic (PD) growth phase and recovered on day 4 of cell culturing (Figure 3A); and (iv) reduces the length of mitochondrial cristae and alters their morphological appearance (Figure 3A). Furthermore, our indirect immunofluorescence microscopy analysis of cells grown under CR on 0.2% glucose and recovered at different periods of CLS revealed that the atg32Δ mutation also causes massive fragmentation of the elaborate mito-chondrial network seen in WT cells (Figures 3B and 3E).

In sum, our microscopical analyses suggest that under CR conditions macromitophagy is required to maintain a network of mitochondria, perhaps by selectively eliminating individual mitochondria segregated from this network due to certain changes in the morphology of their cristae. One could envisage that the inability of macromitophagy-deficient atg32Δ cells to eliminate these morphologically distinct (and possibly dysfunctional) mitochondria following their segregation from the mitochondrial network may result in their progressive accumulation with age, thereby leading to the establishment of a pro-aging cellular pattern and ultimately shortening yeast longevity.

Macromitophagy sustains a healthy population of functional mitochondria

Based on our finding that macromitophagy plays an essential role in preserving the mitochondrial network in CR yeast, we hypothesized that this form of selective macroautophagy is required for sustaining a healthy population of functional mitochondria in chronologically aging cells. To test the validity of our hypothesis, we assessed the age-related dynamics of changes in several vital mitochondrial functions during chronological aging of WT and atg32Δ yeast cells. We found that in cells cultured under CR on 0.2% glucose and recovered at different periods of CLS, the atg32Δ mutation (i) leads to accumulation of dysfunctional mitochondria exhibiting reduced respiration (Figure 4A); (ii) causes a build-up of dysfunctional mitochondria having a reduced electrochemical potential across the inner mitochondrial membrane (IMM) (Figure 4B); (iii) triggers a release of cytochrome c from the intermediate space of mitochondria into the cytosol (Figure 4C), known to be a hallmark event of mitochondrial outer membrane permeabilization (MOMP) [29]; and (iv) reduces the cellular level of ATP (Figure 4D), which under CR conditions is produced mainly in mitochondria [30]. All these changes in vital mitochondrial functions were observed in macromitophagy-deficient atg32Δ cells as early as after 2 days of culturing (Figure 4).

These findings suggest that under CR conditions macromitophagy is required to sustain a healthy network of functional mitochondria that actively respire, exhibit high electrochemical potential across the IMM, maintain outer mitochondrial membrane impermeability to proteins and efficiently synthesize ATP. Thus, it is conceivable that in chronologically aging yeast limited in calorie supply macromitophagy selectively eliminates dysfunctional mitochondria impaired in vital mitochondrial functions that define longevity.

Macromitophagy maintains a population of mitochondria whose inner membrane exhibits abundant respiratory and non-respiratory protein supercomplexes of distinct compositions

A body of recent evidence supports the view that, instead of existing as separate units freely moving within the IMM, individual respiratory protein complexes are assembled into supramolecular structures known as respiratory supercomplexes or respirasomes [31, 32]. Some changes in such supramolecular organization of respiratory protein complexes may cause detrimental alterations of the mitochondrial respiratory chain, thereby eliciting certain age-related pathologies [33]. We therefore hypothesized that the observed inability of prematurely aging under CR conditions atg32Δ cells to sustain a healthy population of functional mitochondria (Figure 4) may be caused in part by some changes in the abundance and/or composition of certain respiratory protein super-complexes. To validate this hypothesis, we purified mitochondria from WT and atg32Δ cells, used digitonin to solubilize protein complexes from the inner membrane of these mitochondria, and separated the solubilized protein complexes on a linear 4-13% acrylamide gradient gel for first-dimension blue native PAGE (1-D BN-PAGE). This gradient gel is known to resolve protein complexes within the mass range of 100 kDa - 3 MDa [34]. Our 1-D BN-PAGE analysis revealed that in cells cultured under CR on 0.2% glucose and recovered on day 4 of culturing, the atg32Δ mutation reduces the levels of several protein complexes in the IMM, including complexes 2, 3, 4 and 5 (Figure 5A). Because the molecular masses of all these and other protein complexes resolved by 1-D BN-PAGE exceeded 545 kDa, a molecular mass of the largest commercially available urease tetramer marker protein (Figure 5A), they represent very large protein supercomplexes assembled from several protein complexes [35]. Noteworthy, our separation of denatured monomeric membrane proteins from purified mitochondria by SDS-PAGE and their subsequent quantitative analysis by immunoblotting revealed that the atg32Δ mutation does not cause significant changes to the total levels of the respiratory complexes III, IV or V (Figures 5B and C). Hence, the observed reduction of the levels of protein supercomplexes 2, 3, 4 and 5 in the IMM of atg32Δ cells (Figure 5A) was not due to a decrease in the total amounts of the respiratory complexes III, IV or V. It is conceivable therefore that in the IMM of atg32Δ cells a significant portion of these protein supercomplexes undergo a remodeling caused by their complete or partial dissociation into individual proteins or small protein subcomplexes.

To define the composition of each of the protein supercomplexes recovered by 1-D BN-PAGE (Figure 5A), we resolved their individual protein components by second-dimension tricine-SDS-PAGE (2-D SDS-PAGE), visualized these components by silver staining and identified them by mass spectrometry. We found that, in addition to one or more kinds of individual respiratory complexes, many protein supercomplexes in the IMM of WT cells contain one or more proteins that have not been traditionally viewed as components associated (even transiently) with the mitochondrial respiratory chain. These novel protein components of respiratory protein supercomplexes in the IMM of WT cells include: (i) the ADP-ATP carrier protein Aac2p, a component of supercomplexes 1, 2, 4 and 6; (ii) the mitochondrial porin Por1p, a component of supercomplexes 1, 2 and 8; (iii) the Pda1p, Pdb1p, Lat1p and Lpd1p subunits of the mitochondrial pyruvate dehydrogenase complex, components of the supercomplex 1; (iv) the mitochondrial external NADH dehydrogenase Nde1p, a component of the supercomplex 1; (v) the Kgd1p subunit of the mitochondrial alpha-ketoglutarate dehydrogenase complex, a component of the supercomplex 2; (vi) the Ald4p isoform of acetaldehyde dehydrogenase, a component of the supercomplex 8; and (vii) the Idh1p and Idh2p isoforms of isocitrate dehydrogenase, a component of the supercomplex 8 (Figure 6A; Table S1). Moreover, the protein supercomplex 5 in the IMM of WT cells is a non-respiratory supercomplex composed of many molecules of Yme2p (Figure 6A; Table S1), an integral IMM protein with a role in maintaining mitochondrial nucleoid structure and number. It should be stressed that the atg32Δ mutation not only eliminates this non-respiratory supercomplex 5 but also causes a dissociation of several proteins or protein complexes from supercomplexes 1, 2, 3, 4, 6, 7 and 8. Some of these dissociated in atg32Δ cells proteins or protein complexes are known as components of the mitochondrial respiratory chain, including (i) Ndi1p, a dissociated from the supercomplex 1 oxidoreductase that (akin to the mitochondrial respiratory complex I in higher eukaryotes) transfers electrons from NADH to ubiquinone in the respiratory chain but (unlike respiratory complex I in higher eukaryotes) does not pump protons; (ii) the respiratory complex III, which dissociates from supercomplexes 3, 7 and 8; (iii) the respiratory complex IV, which dissociates from the supercomplex 7; and (iv) the respiratory complex V, which dissociates from supercomplexes 4, 6 and 8 (Figures 6A and 6B; Table S1). Other of the dissociated in atg32Δ cells proteins or protein complexes have not been traditionally viewed as components associated (even transiently) with the mitochondrial respiratory chain; they include (i) dissociated from the supercomplex 1 subunits Pda1p, Pdb1p, Lat1p and Lpd1p of the mitochondrial pyruvate dehydrogenase complex; (ii) a dissociated from the supercomplex 1 mitochondrial external NADH dehydrogenase Nde1p; and (iii) a dissociated from the supercomplex 2 subunit Kgd1p of the mitochondrial alpha-ketoglutarate dehydrogenase complex (Figures 6A and 6B; Table S1).

In sum, these data suggest that under CR conditions macromitophagy is required to sustain a population of mitochondria whose inner membrane exhibits abundant protein supercomplexes, each composed of a distinct set of respiratory and non-respiratory protein complexes. One could therefore foresee that in chronologically aging yeast limited in calorie supply macromitophagy selectively eliminates mitochondria whose inner membrane (i) has reduced levels of respiratory and non-respiratory protein supercomplexes 2, 3, 4 and 5 due to their complete dissociation into individual proteins or small protein subcomplexes; (ii) accumulates respiratory and non-respiratory protein supercomplexes 1, 3, 4, 6, 7 and 8 that lack one or more respiratory complexes; (iii) amasses respiratory and non-respiratory protein supercomplexes 1, 2 and 8 that lack proteins not traditionally viewed as components associated with the mitochondrial respiratory chain; and/or (iv) has diminished level of the non-respiratory protein supercomplex 5 composed of many molecules of Yme2p, an integral IMM protein involved in maintaining mitochondrial nucleoid structure and number.

Macromitophagy preserves a population of mitochondria that efficiently couple electron transport to ATP synthesis and exhibit high and balanced activities of all five oxidative phosphorylation (OXPHOS) complexes

We sought to investigate if the observed in macromitophagy-deficient atg32Δ cells reduced abundance and altered composition of the major respiratory and non-respiratory protein supercomplexes (Figures 5 and 6) have an impact on the functional state of the electron transport and OXPHOS system in the IMM of these prematurely aging mutant cells.

To attain this objective, we used polarography of mitochondria purified from WT and atg32Δ cells for measuring the following three kinds of respiratory rate: (i) the rate of state III (also called ADP-stimulated state) respiration [36, 37], which was assessed immediately following addition of 50 μM ADP to mitochondria supplemented with a respiratory substrate; (ii) the rate of state IV (also known as resting, basal or ADP-independent state) respiration [37, 38], which was monitored after the exogenously added ADP was entirely converted to ATP; and (iii) the rate of uncoupled (UC) respiration [37, 39], which was determined after addition of the uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP) to mitochondria incubated with a respiratory substrate in the absence of exogenous ADP. In these polarographic assays, the rates of state III, state IV and UC respiration were measured using NADH as a respiratory substrate of external NADH:quinone oxidoreductase or succinate as a respiratory substrate of complex II.

We found that the atg32Δ mutation significantly reduces the rates of state IV (Figures 7D and 7E) and UC (Figures 7G and 7H) respiration in the presence of NADH or succinate as a respiratory substrate. Both these respiratory rates are known to reflect the capacity of the electron transport chain (ETC) in the IMM [36, 37]. Thus, in cells grown under CR conditions on 0.2% glucose macromitophagy is required to preserve a population of mitochondria that exhibit high capacity of electron transport along the respiratory chain.

We also found that the atg32Δ mutation considerably decreases (i) the rate of state III respiration (Figures 7A and 7B); (ii) the state III rate/state IV rate ratio (known as the respiratory control ratio, RCR) (Figures 7J and 7K); (iii) the state III rate/UC rate ratio (Figures 7M and 7N); and (iv) the ADP/O ratio (i.e., the number ATP molecules synthesized per oxygen atom reduced) (Figures 7P and 7Q). All these parameters are known to reflect the extent of the functional and physical integrity of the IMM; the ADP/O ratio is also a direct measure of the efficacy of coupling between ADP phosphorylation and electron transport along the respiratory chain [36, 37, 39]. We therefore concluded that in yeast cells limited in calorie supply macromitophagy is obligatory for sustaining a population of mitochondria that (i) preserve the functional and physical integrity of the IMM; and (ii) efficiently couple electron transport to ATP synthesis.

To get insight into the effect of the observed in macromitophagy-deficient atg32Δ cells reduced abundance and altered composition of the major respiratory and non-respiratory protein supercomplexes on the functional state of each of the five OXPHOS protein complexes, we measured their enzymatic activities in mitochondria purified from WT and atg32Δ cells grown under CR conditions on 0.2% glucose.

As we found, the atg32Δ mutation causes a significant reduction in enzymatic activities of NADH: decylubi- quinone oxidoreductase (NQR; an impaired in proton pumping across the IMM analog of the respiratory complex I in higher eukaryotes), succinate:decylubiquinone-2,6-dichlorophenolindo-phenol oxidoreductase (the respiratory complex II), ubiquinol:cytochrome c oxidoreductase (the respiratory complex III), cytochrome c oxidase (the respiratory complex IV) and F1F0-ATP synthase (the respiratory complex V) (Figures 7C, 7F, 7I, 7L and 7O). Noteworthy, the extent of such reduction for the respiratory complexes IV and V substantially exceeded that for NQR as well as for the respiratory complexes II and III (Figure 7R). Hence, in yeast cells limited in calorie supply macromitophagy is needed for maintaining a population of mitochondria that display high and balanced activities of all five OXPHOS complexes.

Altogether, our polarographic and enzymatic assays in mitochondria purified from WT and atg32Δ cells suggest that in chronologically aging yeast under CR conditions macromitophagy may selectively eliminate mitochondria whose inner membrane exhibits (i) a compromised functional and physical integrity of the IMM; (ii) disproportionally reduced activities of the five OXPHOS complexes; (iii) a low capacity of electron transport along the respiratory chain; and (iv) a reduced efficacy of coupling between ADP phosphorylation and electron transport.

Macromitophagy sustains a population of mitochondria that generate low levels of ROS and exhibit only a minor oxidative damage to mitochondrial macromolecules

A body of evidence supports the view that the multifactorial functional state of the mitochondrial ETC and OXPHOS system in the inner membrane of yeast mitochondria (rather than just the total number of electrons transported along the respiratory chain) determines the rate of mitochondrial ROS production, thereby modulating the extent of oxidative cellular damage and other vital processes defining yeast longevity [40-44]. It is well established that the functional state of the mitochondrial ETC and coupled OXPHOS system in yeast is influenced in part by the rate of mitochondrial respiration, capacity of electron flow through the respiratory chain, number and activity of individual OXPHOS enzymes, and relative levels of the five OXPHOS complexes [40-42, 44-46]. We demonstrated that in yeast limited in calorie supply the atg32Δ-dependent mutational block of macromitophagy reduces mitochondrial respiration, decreases capacity of the mitochondrial ETC, lowers the efficacy of coupling between ADP phosphorylation and electron transport, and disproportionally lessens activities of all OXPHOS enzymes (Figures 4A and 7). This observation prompted us to investigate if the longevity-shortening atg32Δ mutation alters the age-related dynamics of changes in mitochondrially generated ROS and/or impacts the extent of oxidative damage to mitochondrial macromolecules. We found that in chronologically aging yeast cultured under CR on 0.2% glucose and recovered at different periods of CLS, the atg32Δ mutation elevates the level of mitochondrially produced ROS (Figure 8A). Furthermore, our comparative analysis of mitochondria purified from WT and atg32Δ mutant cells limited in calorie supply revealed that this mutation increases the extent of oxidative carbonylation of mitochondrial proteins (Figures 8B and S1) and elevates the levels of oxidatively damaged mitochondrial membrane lipids (Figure 8C). We also observed that atg32Δ mutant cells grown under CR on 0.2% glucose exhibited augmented, as compared to WT cells, frequencies of deletion (rho⁻ and rho⁰) and point (rib2 and rib3) mutations in mitochondrial DNA (mtDNA) (Figures 8D and 8E) - likely due to an elevated extent of oxidative damage to mtDNA in these mutant cells.

Taken together, these findings suggest that under CR conditions macromitophagy is required to sustain a population of mitochondria that generate low levels of ROS and exhibit only a minor oxidative damage to mitochondrial proteins, membrane lipids and DNA. Thus, it is conceivable that in chronologically aging yeast limited in calorie supply macromitophagy selectively eliminates mitochondria producing excessively high concentrations of ROS and accumulating elevated levels of macromolecules oxidatively damaged due to their exposure to mitochondrial ROS. One could predict that the inability of macromitophagy-deficient atg32Δ cells to eliminate these oxidatively impaired (and therefore dysfunctional) mitochondria may result in their progressive accumulation with age, thereby shortening yeast longevity.

Macromitophagy is required for maintaining the homeostasis of membrane lipids in mitochondria, the ER and the PM

Mitochondria are known to be the only site within a yeast cell where cardiolipin (CL), the signature glycerophospholipid of these organelles, is synthesized [25, 26]. Furthermore, yeast mitochondria house the synthesis of the glycerophospholipid phosphatidy-lethanolamine (PE), minor quantities of which are also formed in the Golgi apparatus [26]. It should be emphasized that both CL and PE are synthesized in the the inner membrane of yeast mitochondria from their precursor lipid species formed exclusively in the ER; cytidine diphosphate-diacylglycerol (CDP-DAG) serves as a precursor for CL, whereas the glycerophospholipid phosphatidylserine (PS) is a precursor for PE (Figure S2) [25]. Thus, the confined to mitochondria synthesis of CL and PE in yeast cells relies on a delivery of their CDP-DAG and PS precursors from the ER to the IMM [21, 25, 47]. Moreover, the mitochondrially synthesized PE serves as a precursor for the glycerophospholipid phosphatidylcholine (PC), which is synthesized only in the ER (Figure S2) [26, 27]. Hence, following its synthesis in the ER, PC must be transported to mitochondria incapable of synthesizing this essential glycerophospholipid constituent of both mitochondrial membranes (Figure S2) [25, 47]. In addition, the glycerophospholipids phosphatidic acid (PA) and phosphatedylinositol (PI) are synthesized in the ER but not in mitochondria of yeast cells; therefore, these two lipid species playing essential roles in mitochondrial membranes need to be translocated to these membranes from the site of their synthesis in the ER (Figure S2) [25, 26]. Altogether, these findings imply that the normal functioning of yeast mitochondria requires a bidirectional lipid exchange between the mitochondrial and ER membranes. Such an exchange occurs via zones of close apposition, known as membrane contact sites or junctions, between the outer mitochondrial membrane and the mitochondria-associated membrane (MAM) domain of the ER; from 80 to 110 of these mitochondria-ER junctions are believed to exist per yeast cell (Figure S2) [20, 21, 27]. Of note, following their synthesis within the ER and inner mitochondrial membranes, lipids can be translocated to the PM; as many as 1,100 PM-ER junctions per yeast cell are involved in lipid transport from the PM-associated membrane (PAM) domain of the ER to the PM (Figure S2) [21-24].

Considering such intricate integration of mitochondria into a network governing lipid dynamics not only within these organelles but also within the ER and the PM, we hypothesized that the observed inability of prematurely aging under CR conditions atg32Δ cells to eliminate morphologically distinct, dysfunctional and oxidatively damaged mitochondria may compromise lipid metabolism and movement within the entire network. One could therefore envisage that the atg32Δ-dependent mutational block of macromitophagy may alter membrane lipid composition not only in mitochondria but also in the ER and the PM.

To test the validity of our hypothesis, we used mass spectrometry to compare the membrane lipidomes of mitochondria, ER and PM purified from WT and atg32Δ cells that were cultured under CR on 0.2% glucose and recovered on day 1, 2 or 4 of culturing. We found that (i) the atg32Δ mutation considerably alters levels of several membrane lipid species in mitochondria, ER and PM; and (ii) the effect of atg32Δ on the membrane lipidomes of mitochondria, the ER and the PM is different in cells recovered on day 1, 2 or 4 of culturing - and, thus, is age-related (Figure 9).

Indeed, in cells recovered at logarithmic growth phase on day 1, the atg32Δ mutation (i) increased the levels of PE (a glycerophospholipid delivered to the ER after its synthesis in the IMM) in mitochondria and the PM, and reduced its level in the ER; (ii) lowered the level of CL (a mitochondrially synthesized and accumulated glycerophospholipid) in mitochondria; (iii) decreased the level of PC (a glycerophospholipid synthesized in the ER membrane and then delivered to other membranes) in the PM, but did not alter its abundance in mitochondria or the ER; (iv) elevated the levels of PI (a glycerophospholipid synthesized in the ER membrane and then delivered to other membranes) in mitochondria and the ER, but did not alter its abundance in the PM; (v) lowered the level of TG (neutral lipid species synthesized in the ER but not in mitochondria or the PM) in the ER; and (vi) did not affect the levels of PA and PS (two glycerophospholipids synthesized in the ER membrane and then delivered to other membranes) in mitochondria, the ER and the PM (Figures 9A, 9D and 9G).

Furthermore, in cells recovered at diauxic growth phase on day 2, the atg32Δ mutation (i) reduced the level of PE in the ER membrane but not in mitochondria or the PM; (ii) elevated the level of CL in mitochondria; (iii) decreased the levels of PC in mitochondria and the PM, but did not alter its abundance in the ER; (iv) reduced the level of PI in mitochondria and elevated the level of this glycerophospholipid in the ER, but did not alter its abundance in the PM; (v) lowered the level of TG in the ER; (vi) increased the level of PA in the ER membrane, but did not alter its abundance in mitochondria or the PM; and (vii) reduced the level of PS in the ER and elevated the level of this glycerophospholipid in the PM, but did not alter its abundance in mitochondria (Figures 9B, 9E and 9H).

Moreover, in cells recovered at post-diauxic growth phase on day 4, the atg32Δ mutation (i) elevated the levels of PE in mitochondria and the PM, and reduced its abundance in the ER; (ii) decreased the level of CL in mitochondria; (iii) reduced the level of PC in the ER and increased the level of this glycerophospholipid in the PM, but did not alter its abundance in mitochondria; (iv) lowered the levels of PI in mitochondria and the ER, but did not alter its abundance in the PM; (v) did not alter the level of TG in the ER; (vi) elevated the levels of PA in mitochondria and the ER, and reduced its abundance in the PM; and (vii) decreased the level of PS in mitochondria, but increased the levels of this glycerophospholipid in the ER and the PM (Figures 9C, 9F and 9I).

Altogether, these findings imply that under CR conditions macromitophagy is essential for sustaining the homeostasis of membrane lipids in mitochondria, the ER and the PM. It is plausible that the observed in macromitophagy-deficient atg32Δ cells age-related changes in the membrane lipidomes of mitochondria, the ER and the PM may shorten yeast longevity by establishing a pro-aging cellular pattern.

Yeast strains, media and growth conditions

The wild-type strain Saccharomyces cerevisiae BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) as well as the single-gene-deletion mutant strains atg32Δ (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 atg32Δ::kanMX4) and atg36Δ (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 atg36Δ::kanMX4) in the BY4742 genetic background [all from Thermo Scientific/Open Biosystems; #YSC1054, #YSC1021-550234 and #YSC1021-548143, respectively] were grown in YP medium (1% yeast extract, 2% peptone) [both from Fisher Scientific; #BP1422-2 and #BP1420-2, respectively] containing 0.2% glucose [#D16-10; Fisher Scientific] as carbon source. Cells were cultured at 30°C with rotational shaking at 200 rpm in Erlenmeyer flasks at a “flask volume/medium volume” ratio of 5:1.

Identification of proteins by mass spectrometry

Solubilization of purified mitochondria for first-dimension blue native PAGE (1-D BN-PAGE), separation of the solubilized mitochondrial membrane protein supercomplexes by 1-D BN-PAGE, subsequent resolution of individual protein components of these supercomplexes by second-dimension tricine-SDS-PAGE (2-D SDS-PAGE) and following silver staining were carried out as previously described [34]. Protein bands were excised from the gel, reduced, alkylated and in-gel digested with trypsin [77]. The resulting peptides were separated by reverse phase high performance liquid chromatography coupled to mass spectrometry (RP-HPLC/MS) using an LTQ Orbitrap (#10145339; Thermo Scientific) [78]. The raw MS data file obtained by Xcalibur was analyzed using the Thermo Scientific Xcalibur Proteome Discoverer application (version 1.3) hereafter referred to as the Proteome Discoverer. The Proteome Discoverer was used to identify individual protein components of the isolated mitochondrial respiratory supercomplexes by comparing the raw data of mass spectra of digested fragments to the mass spectra of peptides within the Uniprot FASTA database. The analysis by the Proteome Discoverer coupled to the FASTA database was enabled by using the peak-finding search engine SEQUEST, which processes MS data using a peak-finding algorithm to search the raw data for generating a peak probability list with relative protein abundances.

Analysis of lipids by mass spectrometry

Extraction of lipids from purified mitochondria, ER and PM and following mass spectrometric identification and quantitation of various lipid species were carried out as previously described [79]. Mass spectrometric analyses were performed with a Thermo OrbitrapVelos mass spectrometer equipped with a HESI-II ion source (#10145339; Thermo Scientific) operating at a flow rate of 5 μl/min. Numerous lipid species were separated by Fourier transform tandem mass spectrometry. Mass spectra were converted to an open format (mzXML, mzML) using the ProteoWizard MSConvert software freely available from http://proteowizard.sourceforge.net/. The raw data were then imported into the open source software LipidXplorer (freely available from https://wiki.mpi-cbg.de/lipidx/Main_Page) for the automated detection and quantitation of lipid species.

Miscellaneous procedures

Chronological lifespan assay and pharmacological manipulation of chronological lifespan by addition of lithocholic acid (LCA) [Sigma; #L6250] were carried out as previously described [19]. The stock solution of LCA in water was made on the day of adding this compound to cell cultures. LCA was added to growth medium at the final concentration of 50 μM immediately following cell inoculation into the medium.

Fluorescence [18], immunofluorescence [18] and electron [80] microscopies followed by morphometric analyses of the resulting images were performed according to established procedures.

Cellular respiration assay [18], monitoring of the mitochondrial membrane potential [18], ROS measurement 81], immunodetection of carbonyl groups in oxidatively damaged mitochondrial proteins [81] and measurement of the frequencies of spontaneous mutations in mitochondrial DNA [56] were performed as previously described. Oxidatively damaged proteins and lipids in samples of purified mitochondria were determined with a Protein Carbonyl Assay Kit assay kit (#10005020; Cayman Chemical) and a PeroXOquant Quantitative Peroxide Assay Kit assay kit (#23285; Thermo Scientific Pierce), respectively, following the manufacturer's instructions.

Preparation of cellular extracts and a microanalytic biochemical assay for measuring ATP have been described elsewhere [82].

Subcellular fractionation of yeast [83] followed by isolation of mitochondria [84], ER [83] and cytosolic fraction [80] were performed according to established procedures. Yeast plasma membrane was isolated as previously described [85].

Protein concentration in samples of purified mitochondria was determined with an RC DC protein assay kit (#500-0122; Bio-Rad) following the manufacturer's instructions. SDS-PAGE and immunoblotting using a Trans-Blot SD semi-dry electrophoretic transfer system (#170-3940; Bio-Rad) were performed as previously described [86]. Blots were decorated with monoclonal antibodies raised against porin (#459500; Invitrogen) or polyclonal antisera raised against cytochrome c (kind gift of Dr. Roland Lill, Philipps Universität Marburg), cytochrome c1, subunit II of cytochrome c oxidase or alpha and beta subunits of the mitochondrial F1ATPase (all three antisera were kind gifts of Dr. Carla Koehler, University of California, Los Angeles). Antigen-antibody complexes were detected by enhanced chemilumines-cence using an Amersham ECL Plus Western Blotting Detection Reagents(#RPN2132; GE Healthcare).

State III and IV respiratory rates measurement using NADH as a respiratory substrate of external NADH:quinone oxidoreductase or succinate as a respiratory substrate of complex II, as well as uncoupled respiratory rates (UC) measurement in the presence of the uncoupler CCCP and a respiratory substrate (NADH or succinate) were performed by polarography of purified mitochondria according to established procedures [36]. The ratios of state III rate/state IV rate (i.e., RCR), state III rate/UC rate and ADP/O (i.e., the stoichiometry of coupled oxidation and phosphorylation) were calculated as previously described [37]. Enzymatic activities of NADH:decylu-biquinone oxidoreductase (NQR; an impaired in proton pumping across the IMM analog of the respiratory complex I in higher eukaryotes) [35], succinate:decy-lubiquinone-2,6-dichlorophenolindo-phenol oxido-reductase (the respiratory complex II) [87], ubiquinol:cytochrome c oxidoreductase (the respiratory complex III) [88], cytochrome c oxidase (the respiratory complex IV) [38] and F₁F₀-ATP synthase (the respiratory complex V) [39] were determined by established methods.