Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Dev Cell. Author manuscript; available in PMC 2010 July 1.
Published in final edited form as:
PMCID: PMC2746076

Atg32 is a mitochondrial protein that confers selectivity during mitophagy


Mitochondrial quality control is important to maintain proper cellular homeostasis. Although selective mitochondrial degradation by autophagy (mitophagy) is suggested to have an important role for quality control and there is evidence for a direct relation between mitophagy and neurodegenerative diseases, the molecular mechanism of mitophagy is poorly understood. Using a screen for mitophagy-deficient mutants, we found that YIL146C/ECM37 is essential for mitophagy. This gene is not required for other types of selective autophagy or for nonspecific macroautophagy. We designated this autophagy-related (ATG) gene as ATG32. The Atg32 protein localizes on mitochondria. Following the induction of mitophagy, Atg32 binds Atg11, an adaptor protein for selective types of autophagy, and then Atg32 is recruited to and imported into the vacuole along with mitochondria. Therefore, Atg32 confers selectivity for mitochondrial sequestration as a cargo and is necessary for recruitment of this organelle by the autophagy machinery for mitophagy.

Keywords: autophagy, ECM37, lysosome, mitochondria, stress, vacuole, yeast


One drawback of mitochondrial ATP synthesis is that this process generates reactive oxygen species (ROS) that can damage the organelle. Accordingly, appropriate quality control of this organelle is important to maintain proper cellular homeostasis. The accumulation of damaged protein and DNA in mitochondria is related to aging, cancer and neurodegenerative diseases (Wallace, 2005). Thus, intensive analyses of mitochondrial DNA repair and damaged protein degradation mechanisms have been carried out (Bogenhagen, 1999; Larsson and Clayton, 1995; Rep and Grivell, 1996).

The first report demonstrating the presence of mitochondria within an autophagosome in mammalian cells was made in 1957 (Clark, 1957). Since then it has been assumed that autophagy is one of the pathways for mitochondrial recycling. Recently, it is suggested that engulfment of mitochondria by autophagy occurs through a selective process, termed mitophagy, which may play an important role for mitochondrial quality control (Mijaljica et al., 2007; Nowikovsky et al., 2007; Priault et al., 2005; Twig et al., 2008; Zhang et al., 2007). Direct evidence for mitophagy, however, has been limited. Mitophagy is now attracting considerable attention, because it is related with cellular differentiation and disease. For example, during erythroid cell maturation, mitochondria are eliminated by autophagy when Nix-dependent loss of membrane potential is induced (Sandoval et al., 2008; Schweers et al., 2007). Parkin is an E3 ubiquitin ligase encoded by the PARK gene, and a loss-of-function mutation in PARK2 represents the most common cause of Parkinson disease. Recently, Parkin is shown to be recruited selectively to impaired mitochondria and to promote their degradation via autophagy (Narendra et al., 2008). Despite the important role for mitophagy in development and disease, the molecular mechanism of mitophagy is not well understood.

Autophagy in yeast can be classified into several categories, and the best characterized of these are nonspecific and selective macroautophagy (Cao and Klionsky, 2007). Macroautophagy is the bulk degradation of cytoplasmic components that allows cells to respond to various types of stress and to adapt to changing nutrient conditions (Klionsky, 2005; Yorimitsu and Klionsky, 2007). Selective autophagy includes the cytoplasm to vacuole targeting (Cvt) pathway (Klionsky and Emr, 2000; Yorimitsu and Klionsky, 2005) and pexophagy (Dunn et al., 2005), which have the Cvt complex (aminopeptidase I (Ape1), along with receptor and adaptor proteins) and peroxisomes as specific cargo, respectively.

To date, 31 autophagy-related (ATG) genes have been identified, which function as the molecular machinery for autophagy. Most of the ATG genes are required for both macroautophagy and selective autophagy, but some are required only for specific types of autophagy. For example, Atg19 functions as a receptor protein that binds the precursor form of Ape1 (prApe1). Atg19 also interacts with Atg11, an adaptor protein that is needed along with Atg19 to recruit the Cvt complex to the phagophore assembly site (PAS), where the sequestering cytosolic vesicles are generated (Shintani et al., 2002). Similarly, during pexophagy in Pichia pastoris, Atg30 localizes to peroxisomes. Atg11 binds Atg30 and recruits the peroxisomes to the PAS (Farre et al., 2008). Atg11, Atg19 and Atg30 are not required for nonspecific macroautophagy. In both nonspecific and selective macroautophagy, the morphological hallmark is the sequestration of cargo within double-membrane cytosolic vesicles that deliver the contents to the vacuole. Fusion with the vacuole limiting membrane releases the inner vesicle into the lumen, where it is degraded allowing access to the cargo. During selective types of autophagy, the membrane of the sequestering vesicle is closely apposed to the cargo, resulting in the exclusion of bulk cytoplasm. Thus, specific tags are recognized on the cargo to allow them to be selectively targeted. In the case of the Cvt pathway, this includes the propeptide of prApe1 (Shintani et al., 2002), whereas in pexophagy it is the peroxin Pex14 (Farre et al., 2008).

Most of the ATG genes are required for selective mitochondrial autophagy; however, as these genes are also needed for nonspecific macroautophagy it is unclear as to whether there is a specific mechanism that selects mitochondria as a cargo. Recently, we established a sensitive method to monitor mitophagy that relies on the marker protein Om45-GFP and found that the cargo adaptor protein, Atg11, is essential for mitophagy, suggesting that mitochondria are selectively imported into the vacuole (Kanki and Klionsky, 2008). Using the Om45-GFP marker, we screened a yeast knockout library for strains that are deficient in mitophagy. We found that YIL146C/ECM37 is essential for mitophagy, but is not required for other types of selective autophagy or for nonspecific macroautophagy. We designated this autophagy-related gene as ATG32 and characterized the Atg32 protein in the present study.


Screening and identification of ATG32 as a mitophagy-related gene

To monitor mitophagy, we tagged the C terminus of the mitochondrial outer membrane protein Om45 with the Green Fluorescent Protein (GFP), and induced mitophagy by culturing cells in medium with lactate as the sole carbon source (YPL) for more than two days (Kanki and Klionsky, 2008). When mitochondria are delivered to the vacuole in wild-type cells, GFP fluorescence can be observed within the lumen of this organelle (Fig. 1A, WT). To screen for mutants that affect mitophagy, we transformed a DNA fragment into the yeast knockout library to tag GFP at the chromosomal locus for OM45, induced mitophagy, and monitored the presence or absence of the vacuolar GFP fluorescence. From this screen, we identified a strain with a deletion in the YIL146C/ECM37 gene, which was first characterized as part of a large-scale screen for mutants sensitive to calcofluor white (Lussier et al., 1997); the knockout strain was not previously known to affect mitochondrial degradation. Based on the delivery of Om45-GFP to the vacuole, however, we found that the yil146cΔ strain was completely blocked in mitophagy (Fig. 1A, atg32Δ). Accordingly, we named this gene ATG32. The ATG32 gene encodes a 529 amino acid protein of predicted molecular mass 58,968 Da. It has one predicted transmembrane domain in the C-terminal fifth of the protein; there are no other predicted domains or functional motifs. ATG32 does not have obvious homologs in higher eukaryotes, but does have putative homologs in other fungi including Candida glabrata, Kluyveromyces lactis and Eremothecium gossypii.

Figure 1
Atg32 is required for mitophagy

ATG32 is a mitophagy-specific gene

In addition to post-logarithmic phase growth in lactate medium, mitophagy can be induced when cells are shifted from lactate medium (YPL) to nitrogen starvation medium (SD-N), and the level of mitophagy can be semi-quantitatively monitored by measuring the amount of GFP processed from Om45-GFP in the vacuole (Kanki and Klionsky, 2008). Consistent with the microscopy observation, the atg32Δ mutation blocked mitophagy completely, similar to the result seen with the atg1Δ strain, which is defective in both selective and nonspecific autophagy, and the atg11Δ strain, which is blocked only in selective autophagy pathways (Fig. 1B). The mitophagy deficiency of the atg32Δ strain was complemented by exogenously introducing ATG32, confirming that the defect is the direct result of the absence of the Atg32 protein (Fig. 1C).

Next, we decided to determine whether the ATG32 gene product is required for nonspecific macroautophagy or other types of selective autophagy. The GFP-Atg8 processing assay (Shintani and Klionsky, 2004) is similar to that used for monitoring Om45-GFP delivery to the vacuole. In this case, GFP-Atg8 is a marker for the phagophore, the initial sequestering compartment that forms the autophagosome, and a population of this protein remains associated with the completed vesicle. Wild-type cells showed the accumulation of free GFP from GFP-Atg8 after cells were shifted to medium lacking nitrogen to induce autophagy (Fig. 2A). The atg1Δ mutant maintained the full-length fusion protein and did not generate any of the processed product. In contrast, the atg11Δ mutant was able to proteolytically process GFP-Atg8 essentially the same as the wild type. Similarly, the atg32Δ strain displayed processed GFP-Atg8 and was indistinguishable from wild-type cells, suggesting that nonspecific autophagy was not defective in this strain. To confirm this observation, we relied on a second assay that provides a more quantitative measure of nonspecific autophagy, activation of the precursor form of Pho8Δ60 (Noda et al., 1995). The wild-type strain showed a time-dependent increase in Pho8Δ60 -dependent alkaline phosphatase activity following a shift to medium lacking nitrogen, whereas the atg1Δ strain maintained a background level of activity (Fig. 2B). As with GFP-Atg8 processing, the atg32Δ strain displayed a level of autophagy activity that was similar to the wild type.

Figure 2
Atg32 is a mitophagy-specific gene

The best-characterized example of selective autophagy is the Cvt pathway, which is used for delivery of the resident hydrolase Ape1 to the vacuole (Klionsky and Emr, 2000). We examined processing of prApe1 in wild-type and control strains and compared it to that in the atg32Δ mutant. The wild-type strain showed essentially complete maturation of prApe1 under steady state conditions, whereas the atg1Δ and atg11Δ strains were completely defective for processing (Fig. 2C). The atg32Δ strain showed normal prApe1 maturation, suggesting that Atg32 is not needed for the Cvt pathway. Finally, we used one additional GFP processing assay to monitor another type of selective autophagy, pexophagy. In this case, the marker protein is Pex14 fused to GFP (Reggiori et al., 2005). The wild-type and atg1Δ strains again served as positive and negative controls, respectively (Fig. 2D). The atg32Δ strain displayed processing of Pex14-GFP after shifting cells from medium with oleic acid to glucose minus nitrogen, similar to the wild type. Thus, these results indicate that ATG32 is a mitophagy-specific gene that is not required for either nonspecific autophagy or other types of selective autophagy.

To further examine the phenotype associated with loss of ATG32 we monitored cell growth on a non-fermentable carbon source; we observed no difference from the wild type (Fig. S1A). Similarly, the level of a mitochondrial marker protein, F1-β, and mitochondrial DNA were indistinguishable from the wild type (Fig. S1BC). Next, we examined loss of viability in starvation conditions. Although the autophagy-deficient atg1Δ strain showed a complete loss of viability after 6 days of starvation, both the wild-type and atg32Δ strains grew to similar levels (Fig. S1D), in agreement with our finding that ATG32 is not required for macroautophagy. Finally, we examined cellular production of ROS for cells grown in YPL or shifted to SD-N. The ROS production in the atg32Δ strain was similar to the wild type in both conditions (Fig. S1E). From our present results we can only conclude that Atg32 plays a critical role in eliminating superfluous organelles (e.g., after cells have been shifted to glucose or reach stationary phase) for adaptation to changing environmental conditions.

Mitochondria are eliminated from mammalian cells when they are dysfunctional, for example following treatment with CCCP (Narendra et al., 2008); however, this treatment does not induce mitophagy in yeast (Kanki and Klionsky, 2008). To examine whether mitophagy in yeast in response to damaged mitochondria is Atg32-dependent we relied on doxycycline-regulated depletion of Mdm38 (Kanki and Klionsky, 2008; Nowikovsky et al., 2007). When Mdm38 was depleted we detected a very weak mitophagy response (Fig. S2), similar to our previous results (Kanki and Klionsky, 2008), which may reflect incomplete loss of the Mdm38 protein. Nonetheless, processing of Om45-GFP in response to the shutoff of MDM38 expression was completely blocked in the absence of Atg32, suggesting that the latter protein is involved in the elimination of damaged as well as superfluous mitochondria.

Atg32 localizes to mitochondria and is imported into the vacuole along with mitochondria during starvation

To further characterize Atg32, we chromosomally tagged the N terminus, which did not affect its function (Fig. S3). To examine the localization of Atg32, we expressed GFP-Atg32 under the control of the regulable GAL1 promoter in galactose medium (YPGal). First, we examined the localization of GFP-Atg32 by colabeling the cells with the mitochondrial marker dye MitoFluor Red 589, and observed complete colocalization (Fig. 3A). After nitrogen starvation for 2 h, some of the GFP-Atg32 accumulated in the vacuole (Fig. 3B). To determine whether this accumulation of GFP-Atg32 within the vacuole lumen is a result of mitochondrial vacuolar targeting, we tagged RFP on the C terminus of the mitochondrial outer membrane protein Om45 (Om45-RFP). In growing conditions, GFP-Atg32 showed a punctate and tubular morphology typical of yeast mitochondria and completely overlapped with Om45-RFP (Fig. 3C, upper panel), again confirming the mitochondrial localization of Atg32. After nitrogen starvation, Om45-RFP accumulated in the vacuole along with GFP-Atg32, suggesting that GFP-Atg32 was targeted to the vacuole along with mitochondria (Fig. 3C lower panel).

Figure 3
Atg32 localizes to mitochondria and is imported into the vacuole along with mitochondria during starvation

We further examined the mitochondrial localization of Atg32 using a biochemical approach. A strain expressing protein A-tagged Atg32 (PA-Atg32), 3xHA-tagged Om45 and PA-tagged Tim23 (inner membrane marker) was fractionated by differential centrifugation. Om45-3HA and Tim23-PA were enriched in the mitochondrial (6,500 × g) fraction, along with PA-Atg32 (Fig. 3D), whereas the cytosolic marker Pgk1 was mostly in the supernatant fraction. Next, the isolated mitochondria were treated with proteinase K in the presence or absence of Triton X-100. Although Tim23-PA, was protected from proteinase K in the absence of detergent, PA-Atg32 was degraded, suggesting that Atg32 localizes on the mitochondrial outer membrane (Fig. 3E). Finally, we treated isolated mitochondria with 0.1 M sodium carbonate, pH 11, and separated the membrane (100,000 × g pellet) and supernatant fractions. The peripheral inner membrane β subunit of the F1-ATPase was released into the supernatant, whereas PA-Atg32 was detected only in the membrane fraction along with the majority of Om45 and Tim23 (Fig. 3F), suggesting that Atg32 is a transmembrane protein.

Atg32 is a receptor for recruiting mitochondria to the vacuole

Atg11 serves as an adaptor for the Cvt pathway and pexophagy by linking cargo to the autophagy machinery. Our previous result showing that Atg11 is essential for mitophagy (Kanki and Klionsky, 2008) led us to propose that Atg11 might interact with Atg32 as a receptor on mitochondria. To examine this possibility, a yeast two-hybrid assay was performed between Atg11 fused with the Gal4 DNA binding domain, BD-Atg11, and activation domain-fused Atg32, AD-Atg32. The cells expressing BD-Atg11 and AD-Atg32, but not those expressing either the activation or binding domains as empty vectors, could grow on selective plates (Fig. 4A), suggesting that Atg11 can bind Atg32. To verify the physiological nature of this interaction, we expressed PA-Atg32 and HA-Atg11 in an atg32Δ strain and carried out protein A affinity isolation with IgG-Sepharose (Fig. 4B). Consistent with the yeast two-hybrid result, PA-Atg32 co-precipitated HA-Atg11 in growing conditions (SMD), although the amount of co-precipitated HA-Atg11 was extremely minimal (Fig. 4B, lane 3); however, a one-hour shift to nitrogen starvation (SD-N) conditions dramatically increased the amount of co-precipitated HA-Atg11 (Fig. 4B, lane 6), suggesting that nitrogen starvation induces Atg11-Atg32 binding. This result is consistent with the previous finding that nitrogen starvation can induce mitophagy (Kanki and Klionsky, 2008; Kissova et al., 2007).

Figure 4
Interaction between Atg32 and Atg11

Overexpression of Atg32 enhances the induction of mitophagy

If Atg32 is a mitochondrial tag, we hypothesized that its overexpression would enhance the induction of mitophagy. To test this, we monitored mitophagy in galactose medium with and without the overexpression of Atg32. In wild-type cells overexpressing the vector (pCu-PA), Om45-GFP processing could not be detected by western blot (Fig. 4C). In contrast, when PA-Atg32 was overexpressed (pCu-PA-Atg32), Om45-GFP was processed to generate free GFP. We noted that the overexpressed Atg32 was quickly degraded during starvation (Fig. 4C); however, the majority of this degradation was pep4Δ-independent, indicating that it did not occur in the vacuole (data not shown). Nonetheless, Om45-GFP processing was blocked in atg1Δ and atg11Δ strains, indicating that it was dependent on autophagy (Fig. S4).

A portion of mitochondria with Atg32 can target to the vacuolar surface

During the Cvt pathway, the cargo protein prApe1 forms a complex in the cytosol, and is targeted to the perivacuolar PAS. The Cvt complex is then enwrapped by a Cvt vesicle, that eventually fuses with the vacuole. To see whether mitochondria target to the vacuole through a similar pathway, we monitored GFP-Atg32 localization as a marker of mitochondria using atg1Δ, atg11Δ, and pep4Δ strains. In the atg11Δ strain, GFP-Atg32 showed a typical mitochondrial morphology that localizes largely to the inner side of the plasma membrane; it was difficult to detect any other cytosolic puncta in both growing and starvation conditions (~1% of the cells had puncta on the vacuolar rim in either condition) (Fig. 5A). Presumably, the lack of the adaptor protein Atg11 prevented mitochondria from being recruited to the vacuole. In the atg1Δ strain, most of the Atg proteins and the Cvt complex are recruited to the PAS (Suzuki et al., 2007), but later steps in the autophagic process are blocked. GFP-Atg32 showed cytosolic puncta that localized near the vacuole membrane during starvation in the atg1Δ strain (35.6% of the cells had puncta on the vacuolar rim during starvation, compared to 0.6% in growing conditions; Fig. 5B, arrowheads); this result was similar to the wild-type strain (33.3% and 0% of the cells with vacuolar rim puncta in starvation and growing conditions, respectively; Fig. S5), except that in the wild-type strain a fluorescent signal appeared in the vacuole lumen during starvation. In a pep4Δ strain, which lacks the one of the main vacuolar hydrolases, GFP-Atg32 puncta accumulated within the vacuole during starvation (Fig. 5C). The change in GFP-Atg32 localization could also be seen during a time course analysis (Fig. S6).

Figure 5
Mitochondria target to the vacuole through a Cvt pathway-independent mechanism

In the atg1Δ strain, prApe1 and Atg8 accumulate at a specific point near the vacuole surface that is thought to represent the PAS (Shintani et al., 2002). Most of the GFP-Atg32 puncta did not co-localize with CFP-Ape1 in the atg1Δ strain (Fig. 5D) or a wild-type strain (data not shown), suggesting that mitophagy generally does not occur simultaneously at the same PAS that is being used for the Cvt pathway and nonspecific autophagy. Next, we co-expressed GFP-Atg32 and either CFP-Atg8 or CFP-Atg11, that also form puncta at the PAS with Ape1, and observed their localization in the atg1Δ strain. The majority of the GFP-Atg32 puncta on the vacuole rim overlapped with CFP-Atg8 and CFP-Atg11 puncta, although only approximately half of the CFP-Atg8 and CFP-Atg11 puncta overlapped with GFP-Atg32 (Fig. 5EF). A similar result was seen with the colocalization of these proteins in a wild-type strain (Fig. S7). This finding is consistent with our observation that Atg11 binds Atg32, our previous report that Atg11 accumulates at the PAS for the Cvt pathway and macroautophagy (Yorimitsu and Klionsky, 2005), and with the known role of Atg8 in autophagic sequestration of cargo. Finally, we co-expressed GFP-Atg32 and the mitochondrial matrix protein Idh1-RFP. After starvation for 2 h, the GFP-Atg32 puncta on the vacuole colocalized with Idh1-RFP (Fig. S8), further verifying that Atg32 movement to the vacuole corresponds to mitochondria.

The GFP-Atg32 perivacuolar puncta appeared as a more intense fluorescent signal than other parts of the mitochondria. The intensity of the GFP-Atg32 perivacuolar puncta was 3.3 +/- 1.2 times greater than the intensity of GFP-Atg32 on mitochondria when the GFP signal was standardized relative to Idh1-RFP (Fig. S9). On the other hand, the GFP-Atg32 puncta in the vacuole in the pep4Δ strain was comparable with that at the PAS in the atg1Δ strain (Fig. 5BC), indicating that the GFP-Atg32 puncta, once targeted to the vacuole surface, are eventually delivered into the vacuole.

Electron microscopy analysis of mitophagy

To further examine the localization of Atg32 and its targeting to the vacuole associated with mitochondria, we used immunoelectron microscopy. The pep4Δ strain expressing GFP-Atg32 under the GAL1 promoter was cultured in YPGal medium and then shifted to SD-N for 4 h. Cells were immunostained with anti-YFP antibody and observed by electron microscopy. In YPGal, mitochondria were detected in the cytosol and Atg32 localized on their surface (Fig. 6AB), further confirming our biochemical observation that Atg32 localizes on the mitochondrial outer membrane. After starvation, mitochondria stained with Atg32 were delivered into the vacuole (Fig. 6CD).

Figure 6
Atg32 localizes on mitochondria and targets to the vacuole during mitophagy

In SD-N, the mitochondria were detected primarily within double-membrane vesicles. These results suggest that under the conditions we used to induce mitophagy, the majority of sequestration occurred by a macroautophagic process. In addition, the vesicles containing the mitochondria, mitophagosomes, appeared to be devoid of bulk cytosol, further indicating the selective nature of the process; during specific types of autophagy such as the Cvt pathway and pexophagy, the targeted cargo is closely apposed to the sequestering vesicle membrane and bulk cytosol is largely excluded.


The presence of mitochondria in autophagosomes was first reported in 1957 in mammalian cells (Clark, 1957). Because of connections with damaged mitochondria, aging and disease, a major question has been whether mitochondrial engulfment by autophagosomes can occur in a selective manner. Recent studies suggest that mitochondrial autophagy can occur by a selective process (Elmore et al., 2001; Nowikovsky et al., 2007; Priault et al., 2005; Rodriguez-Enriquez et al., 2004; Twig et al., 2008). Because some ATG genes essential for both macroautophagy and the Cvt pathway are required for mitophagy (Kissova et al., 2004; Kissova et al., 2007; Tal et al., 2007; Zhang et al., 2007), the fundamental mechanism of mitophagy is believed to be similar to other types of autophagy. However, the mechanism of mitochondrial selection, which is essential for mitochondrial quality control, has not been known. This information may provide important insight into certain neurodegenerative diseases, such as Parkinson disease, as well as mechanisms involved in cellular differentiation.

Based on the examples of other types of selective autophagy, including the Cvt pathway and pexophagy, we hypothesized that an organelle tag would be used to mark mitochondria and that this tag would be recognized by a component of the autophagy machinery. To identify such a tag, we performed a mitophagy screen and found that the atg32Δ strain is completely blocked in mitophagy (Fig. 1). Atg32 is not required for nonspecific macroautophagy, the Cvt pathway, or pexophagy, suggesting that Atg32 is a mitophagy-specific protein (Fig. 2). Two genes have been reported to be required for mitophagy, AUP1 and UTH1 (Kissova et al., 2004; Tal et al., 2007). Aup1 is a mitochondrial protein phosphatase, and strains with null mutations in this gene show rapamycin sensitivity (Ruan et al., 2007). On the other hand, Uth1 is SUN family protein, and in this case null strains show rapamycin resistance (Kissova et al., 2004) and display a partial inhibition of mitophagy; these strains are blocked at an early stage of mitophagy (Kissova et al., 2007). The functions of these proteins with regard to mitophagy have not been clarified. We examined these strains for their affects on mitophagy; however, in our hands the corresponding knockout strains showed completely normal levels of mitophagy (Fig. S10). Furthermore, the aup1Δ strain purchased from Research Genetics/Invitrogen (BY4742, aup1Δ::KanR) blocked mitophagy, whereas an aup1Δ strain that we constructed using the same (data not shown) and a second different genetic background did not block mitophagy. The difference between our results and a previous report showing a 40% reduction of mitophagy in the uth1Δ strain (Kissova et al., 2007) may be due to the difference in strain background and the detection method. Thus, atg32Δ is the first identified mutant, to our knowledge, that shows a complete block only in specific mitophagy and not in the Cvt pathway or nonspecific macroautophagy. In addition, our data indicate that Atg32 can connect the Atg proteins and mitochondria directly, providing an explanation for the mechanism of selectivity during mitophagy.

We found that Atg32 localized on mitochondria and could interact with the adaptor protein Atg11, an interaction that was enhanced under mitophagy conditions (Fig. 4 and Fig. 5). By interacting with Atg11, Atg32 is delivered to the vacuole surface and then delivered into the lumen. The last step of mitochondrial delivery into the vacuolar lumen is still unclear. Kissova et al., (2007) reported that mitochondria are directly sequestered via microautophagy; however, the majority of our electron microscopy data suggest that uptake occurs via macroautophagy (Fig. 6C, arrow), although we cannot rule out a microautophagic mechanism

In yeast, both the Cvt pathway and nonspecific macroautophagy use the phagophore assembly site (PAS), which is thought to be the site of Cvt vesicle and autophagosome formation, respectively. Most of the Atg proteins and prApe1, the principal cargo of the Cvt pathway, accumulate at the PAS. The autophagosome or Cvt vesicle is then formed, and eventually this compartment fuses with the vacuole. Because these Atg proteins and prApe1 accumulate at the same point on the vacuolar surface, it is thought that macroautophagy and the Cvt pathway are using a common PAS, although there is no experimental evidence to support this hypothesis. To see whether the Cvt pathway and mitophagy use the same PAS or a different pathway, we used the atg1Δ strain. This strain accumulates Atg proteins and prApe1 at the PAS, but blocks further steps in the sequestration process. We then monitored the localization of GFP-Atg32 as a marker of mitochondria, and CFP-Ape1 as a marker of the Cvt-dependent PAS. GFP-Atg32 formed strong puncta near the vacuole surface, but usually did not colocalize with the PAS that is marked by CFP-Ape1 puncta during starvation (Fig. 5D). This may be expected based on the observation that the membranes that form during selective autophagy closely appose the particular cargo being enwrapped. Thus, engulfment of prApe1 likely occurs at a distinct site relative to sequestration of mitochondria. During starvation conditions we infrequently observed CFP-Ape1 puncta that colocalized with GFP-Atg32 (data not shown); in this case both cargos were presumably present at a larger phagophore that was a precursor to an autophagosome.

We also monitored the localization of GFP-Atg32 and CFP-Atg8 or CFP-Atg11 in atg1Δ and wild-type strains. Atg8 determines the size of nonselective autophagosomes (Xie et al., 2008) and accumulates at the PAS during autophagy. In agreement with the requirement of Atg8 for mitophagy (Kanki and Klionsky, 2008; Zhang et al., 2007), CFP-Atg8 partially colocalized with GFP-Atg32 (Fig. 5E, Fig. S7). Similarly, CFP-Atg11, which is required for the Cvt pathway and binds prApe1, formed multiple puncta on the vacuole surface and approximately half of them displayed colocalization with the GFP-Atg32 puncta (Fig. 5F, Fig. S7). This finding is consistent with our data that Atg32 binds Atg11. Presumably, the remaining (i.e., non-overlapping) CFP-Atg8 and CFP-Atg11 puncta localize at the PAS used for the Cvt pathway and nonspecific autophagy. Finally, we noticed that the GFP-Atg32 puncta observed near the vacuolar surface of the atg1Δ strain or in the vacuolar lumen of the pep4Δ strain generated a more intense fluorescent GFP signal compared to that on mitochondria localized just inside of the plasma membrane (Fig. 5D-F, Fig. S4, S9).

Our current model for mitophagy relative to the Cvt pathway and pexophagy is summarized in Figure 7. All three processes require Atg11 as an adaptor protein. Atg11 interacts with the Cvt pathway receptor Atg19 to recruit the prApe1-Ams1-Atg19 complex to the PAS (Shintani et al., 2002), and a slightly modified PAS is used in common with nonspecific macroautophagy (Cheong et al., 2008). Following the induction of pexophagy in Pichia pastoris, PpAtg30 binds the peroxisomal protein PpPex14 and is phosphorylated. PpAtg30 further interacts with PpAtg11, allowing recruitment of the peroxisome to the PAS for both macropexophagy and micropexophagy (Farre et al., 2008). In Saccharomyces cerevisiae, the requirement of Atg11 for pexophagy has already been reported (Kim et al., 2001), but other steps similar to those found in Pichia pastoris have not been demonstrated. When mitophagy is induced, Atg11 binds the mitochondrial resident protein Atg32. Atg11 recruits mitochondria to the vacuole surface, where uptake may occur by a microautophagic process or by a type of macroautophagy that is separate from the PAS that is used for the Cvt pathway. Clearly additional experiments will be needed to elucidate the details of this selective method of organelle targeting.

Figure 7
Model of selective autophagy


Strains, Media, and Enzymes

The yeast strains used in this study are listed in Table SI. Yeast cells were grown in the following media: rich (YPD; 1% yeast extract, 2% peptone, 2% glucose), lactate (YPL; 1% yeast extract, 2% peptone, 2% lactate), galactose (YPGal; 1% yeast extract, 2% peptone, 2% galactose), synthetic minimal with glucose (SMD; 0.67% yeast nitrogen base, 2% glucose, amino acids, and vitamins), synthetic minimal with lactate (SML; 0.67% yeast nitrogen base, 2% lactate, amino acids, and vitamins) or synthetic minimal with galactose (SMGal; 0.67% yeast nitrogen base, 2% galactose, amino acids, and vitamins). Starvation experiments were performed in synthetic minimal medium lacking nitrogen (SD-N; 0.17% yeast nitrogen base without amino acids, 2% glucose).

Plasmids and Antibodies

The plasmids to express hemagglutinin (HA)-tagged Atg11, CFP-Ape1, CFP-Atg8, HA-CFP-Atg11, and the yeast two-hybrid plasmid pGBDU-Atg11 have been described previously (Shintani et al., 2002; Yorimitsu and Klionsky, 2005). To generate an Atg32-expressing palsmid (ATG32(416)), a DNA fragment encoding the ATG32 gene including sequences from the promoter and terminator regions was PCR-amplified from yeast genomic DNA with primers that included SpeI and EcoRI restriction sites, and ligated into the corresponding sites of the pRS416 vector. For the yeast two-hybrid vector pGAD-Atg32, the protein A (PA)-tagged (pCuPA-ATG32(416)), and GFP-tagged (pCuGFP-ATG32(416)) Atg32 expression vectors, a DNA fragment encoding ATG32 was PCR-amplified with BglII and SalI, or EcoRI and SalI restriction sites from yeast genomic DNA and ligated into the BamHI and SalI sites of pGAD-C1 (James et al., 1996), the EcoRI and SalI sites of pRS416-CuProtA (Kim et al., 2001), and the EcoRI and SalI sites of pCuGFP(416) (Kim et al., 2001), respectively. Monoclonal anti-YFP antibody clone JL8 (Clontech/Takara Bio Group, Mountain View, CA), monoclonal anti-HA antibody clone-HA7 (Sigma-Aldrich, St. Louis, MO), anti-Ape1 serum (Kim et al., 2001), anti-ATPase F1-β antiserum (Dr. Michael Douglas, Washington University, Saint Louis) and anti-Pgk1 antiserum (Dr. Jeremy Thorner, University of California, Berkeley) were used for immunoblotting.

Mitophagy Screening

A yeast knockout strain library (BY4742 background) was purchased form Research Genetics/Invitrogen. To express Om45-GFP, a DNA fragment encoding GFP was integrated at the 3′ end of OM45 by a PCR-based integration method (Longtine et al., 1998). Cells grown on SMD plates were shifted to YPL medium and cultured for 3 days (60 +/- 5 h), then observed by fluorescence microscopy.

Assays for Nonspecific Autophagy, Mitophagy and Pexophagy

For monitoring nonspecific autophagy, the alkaline phosphatase activity of Pho8Δ60, and processing of the GFP-Atg8 chimera were carried out as described previously (Noda et al., 1995; Shintani and Klionsky, 2004). Om45-GFP processing to monitor mitophagy (Kanki and Klionsky, 2008) and Pex14-GFP processing to monitor pexophagy (Reggiori et al., 2005), have been described previously.

Fluorescence Microscopy

Yeast cells expressing fluorescent protein-fused chimeras were grown to mid-log phase or starved in the indicated media. To label the vacuolar membrane or mitochondria, we incubated cells in medium containing 20 μg/ml N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl) pyridinium dibromide (FM 4-64; Molecular Probes, Eugene, OR) or 1 μM MitoFluor Red 589 (Molecular Probes) at 30°C for 30 min, respectively. After being washed with medium, the cells were incubated in medium at 30°C for 30 to 60 min. Fluorescence microscopy observation was carried out as described previously (Monastyrska et al., 2008). To quantify the number of cells that have GFP-Atg32 puncta on the vacuole surface, we counted 80 to 100 cells for each strain.

Protein A Affinity Isolation

Protein A-affinity isolation was carried out essentially as described previously (Shintani et al., 2002).

Subcellular Localization

We followed the protocol for cellular fractionation as described previously (Graham, 2001). Cells expressing Om45-3HA, Tim23-PA and PA-Atg32 were converted to spheroplasts with Zymolyase, suspended in homogenization buffer (0.6 M mannitol, 20 mM HEPES, pH 7.4 and proteinase inhibitors) and homogenized in a Dounce homogenizer. The cell homogenate was centrifuged at 600 × g for 10 min to remove the nucleus and unbroken cells. The supernatant fraction was centrifuged at 6,500 × g for 10 min. The pellet was collected as the mitochondrial fraction. Isolated mitochondria was suspended in ice-cold suspension medium (0.6 M mannitol, 20 mM HEPES, pH 7.4) and treated with proteinase K (200 μg/ml) for 30 min on ice with/without 0.5% Triton X-100. The proteinase K reaction was stopped by adding 10% trichloroacetic acid (TCA). Isolated mitochondria were separately suspended in ice-cold 0.1 M sodium carbonate, pH 11.0, incubated for 30 min on ice and then centrifuged at 100,000 × g for 30 min at 4°C. Proteins in the pellet and supernatant fractions were precipitated by adding 10% TCA. TCA precipitated proteins were washed with acetone and subjected to immunoblotting.

Electron Microscopy

The pep4Δ strain expressing GFP-Atg32 under the GAL1 promoter (TKYM151) was cultured in YPGal medium, then shifted to SD-N and cultured for 4 h. Cells were frozen in a Balzer’s HPM010 high-pressure freezer (Bal-tec, Lichtenstein) or KF80 freezing device (Lica, Austria). Immunoelectron microscopy (IEM) was performed according to the procedures described previously (Baba, 2008). Ultrathin sections were stained with anti-YFP antibody, followed by 0.8-nm colloidal gold-conjugated goat anti-mouse IgG (Aurion, Wageningen, The Netherlands), and then visualized with silver enhancement.

Supplementary Material



This work was supported by National Institutes of Health Grant GM53396 (to D. J. K) and Japan Society for the Promotion of Science Postdoctoral Fellowships for Research Abroad (to T. K.).


aminopeptidase I
cytoplasm to vacuole targeting
green fluorescent protein
protein A
phagophore assembly site
reactive oxygen species


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Abeliovich H, Zhang C, Dunn WA, Jr., Shokat KM, Klionsky DJ. Chemical genetic analysis of Apg1 reveals a non-kinase role in the induction of autophagy. Mol Biol Cell. 2003;14:477–490. [PMC free article] [PubMed]
  • Baba M. Electron microscopy in yeast. Methods Enzymol. 2008;451:133–149. [PubMed]
  • Bogenhagen DF. Repair of mtDNA in vertebrates. Am J Hum Genet. 1999;64:1276–1281. [PubMed]
  • Cao Y, Klionsky DJ. Physiological functions of Atg6/Beclin 1: a unique autophagy-related protein. Cell Res. 2007;17:839–849. [PubMed]
  • Cheong H, Nair U, Geng J, Klionsky DJ. The Atg1 kinase complex is involved in the regulation of protein recruitment to initiate sequestering vesicle formation for nonspecific autophagy in Saccharomyces cerevisiae. Mol Biol Cell. 2008;19:668–681. [PMC free article] [PubMed]
  • Clark SL., Jr. Cellular differentiation in the kidneys of newborn mice studies with the electron microscope. J Biophys Biochem Cytol. 1957;3:349–362. [PMC free article] [PubMed]
  • Dunn WA, Jr., Cregg JM, Kiel JAKW, van der Klei IJ, Oku M, Sakai Y, Sibirny AA, Stasyk OV, Veenhuis M. Pexophagy: the selective autophagy of peroxisomes. Autophagy. 2005;1:75–83. Epub 2005 Jul 2013. [PubMed]
  • Elmore SP, Qian T, Grissom SF, Lemasters JJ. The mitochondrial permeability transition initiates autophagy in rat hepatocytes. FASEB J. 2001;15:2286–2287. Epub 2001 Aug 2217. [PubMed]
  • Farre JC, Manjithaya R, Mathewson RD, Subramani S. PpAtg30 tags peroxisomes for turnover by selective autophagy. Dev Cell. 2008;14:365–376. [PMC free article] [PubMed]
  • Graham JM. Isolation of mitochondria from tissues and cells by differential centrifugation. Curr Protoc Cell Biol. 2001 Chapter 3, Unit 3 3. [PubMed]
  • Jakubowski W, Bilinski T, Bartosz G. Oxidative stress during aging of stationary cultures of the yeast Saccharomyces cerevisiae. Free Radic Biol Med. 2000;28:659–664. [PubMed]
  • James P, Halladay J, Craig EA. Genomic libraries and a host strain designed for highly efficient two-hybrid selection in yeast. Genetics. 1996;144:1425–1436. [PubMed]
  • Kanki T, Klionsky DJ. Mitophagy in yeast occurs through a selective mechanism. J Biol Chem. 2008;283:32386–32393. [PMC free article] [PubMed]
  • Kim J, Kamada Y, Stromhaug PE, Guan J, Hefner-Gravink A, Baba M, Scott SV, Ohsumi Y, Dunn WA, Jr., Klionsky DJ. Cvt9/Gsa9 functions in sequestering selective cytosolic cargo destined for the vacuole. J Cell Biol. 2001;153:381–396. [PMC free article] [PubMed]
  • Kissova I, Deffieu M, Manon S, Camougrand N. Uth1p is involved in the autophagic degradation of mitochondria. J Biol Chem. 2004;279:39068–39074. Epub 32004 Jul 39069. [PubMed]
  • Kissova I, Salin B, Schaeffer J, Bhatia S, Manon S, Camougrand N. Selective and non-selective autophagic degradation of mitochondria in yeast. Autophagy. 2007;3:329–336. Epub 2007 Jul 2021. [PubMed]
  • Klionsky DJ. The molecular machinery of autophagy: unanswered questions. J Cell Sci. 2005;118:7–18. [PMC free article] [PubMed]
  • Klionsky DJ, Emr SD. Autophagy as a regulated pathway of cellular degradation. Science. 2000;290:1717–1721. [PMC free article] [PubMed]
  • Larsson NG, Clayton DA. Molecular genetic aspects of human mitochondrial disorders. Annu Rev Genet. 1995;29:151–178. [PubMed]
  • Longtine MS, McKenzie A, III, Demarini DJ, Shah NG, Wach A, Brachat A, Philippsen P, Pringle JR. Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast. 1998;14:953–961. [PubMed]
  • Lussier M, White AM, Sheraton J, di Paolo T, Treadwell J, Southard SB, Horenstein CI, Chen-Weiner J, Ram AF, Kapteyn JC, et al. Large scale identification of genes involved in cell surface biosynthesis and architecture in Saccharomyces cerevisiae. Genetics. 1997;147:435–450. [PubMed]
  • Mijaljica D, Prescott M, Devenish RJ. Different fates of mitochondria: alternative ways for degradation? Autophagy. 2007;3:4–9. Epub 2007 Jan 2008. [PubMed]
  • Monastyrska I, He C, Geng J, Hoppe AD, Li Z, Klionsky DJ. Arp2 links autophagic machinery with the actin cytoskeleton. Mol Biol Cell. 2008;19:1962–1975. [PMC free article] [PubMed]
  • Narendra D, Tanaka A, Suen DF, Youle RJ. Parkin is recruited selectively to impaired mitochondria and promotes their autophagy. J Cell Biol. 2008;183:795–803. [PMC free article] [PubMed]
  • Noda T, Matsuura A, Wada Y, Ohsumi Y. Novel system for monitoring autophagy in the yeast Saccharomyces cerevisiae. Biochem Biophys Res Commun. 1995;210:126–132. [PubMed]
  • Nowikovsky K, Reipert S, Devenish RJ, Schweyen RJ. Mdm38 protein depletion causes loss of mitochondrial K+/H+ exchange activity, osmotic swelling and mitophagy. Cell Death Differ. 2007;14:1647–1656. Epub 2007 Jun 1641. [PubMed]
  • Priault M, Salin B, Schaeffer J, Vallette FM, di Rago JP, Martinou JC. Impairing the bioenergetic status and the biogenesis of mitochondria triggers mitophagy in yeast. Cell Death Differ. 2005;12:1613–1621. Epub 2005 Jun 1610. [PubMed]
  • Reggiori F, Monastyrska I, Shintani T, Klionsky DJ. The actin cytoskeleton is required for selective types of autophagy, but not nonspecific autophagy, in the yeast Saccharomyces cerevisiae. Mol Biol Cell. 2005;16:5843–5856. Epub 2005 Oct 5812. [PMC free article] [PubMed]
  • Rep M, Grivell LA. The role of protein degradation in mitochondrial function and biogenesis. Curr Genet. 1996;30:367–380. [PubMed]
  • Robinson JS, Klionsky DJ, Banta LM, Emr SD. Protein sorting in Saccharomyces cerevisiae: isolation of mutants defective in the delivery and processing of multiple vacuolar hydrolases. Mol Cell Biol. 1988;8:4936–4948. [PMC free article] [PubMed]
  • Rodriguez-Enriquez S, He L, Lemasters JJ. Role of mitochondrial permeability transition pores in mitochondrial autophagy. Int J Biochem Cell Biol. 2004;36:2463–2472. [PubMed]
  • Ruan H, Yan Z, Sun H, Jiang L. The YCR079w gene confers a rapamycin-resistant function and encodes the sixth type 2C protein phosphatase in Saccharomyces cerevisiae. FEMS Yeast Res. 2007;7:209–215. [PubMed]
  • Sandoval H, Thiagarajan P, Dasgupta SK, Schumacher A, Prchal JT, Chen M, Wang J. Essential role for Nix in autophagic maturation of erythroid cells. Nature. 2008;454:232–235. [PMC free article] [PubMed]
  • Schweers RL, Zhang J, Randall MS, Loyd MR, Li W, Dorsey FC, Kundu M, Opferman JT, Cleveland JL, Miller JL, et al. NIX is required for programmed mitochondrial clearance during reticulocyte maturation. Proc Natl Acad Sci U S A. 2007;104:19500–19505. [PubMed]
  • Shintani T, Huang W-P, Stromhaug PE, Klionsky DJ. Mechanism of cargo selection in the cytoplasm to vacuole targeting pathway. Dev Cell. 2002;3:825–837. [PMC free article] [PubMed]
  • Shintani T, Klionsky DJ. Cargo proteins facilitate the formation of transport vesicles in the cytoplasm to vacuole targeting pathway. J Biol Chem. 2004;279:29889–29894. Epub 22004 May 29811. [PMC free article] [PubMed]
  • Suzuki K, Kubota Y, Sekito T, Ohsumi Y. Hierarchy of Atg proteins in pre-autophagosomal structure organization. Genes Cells. 2007;12:209–218. [PubMed]
  • Tal R, Winter G, Ecker N, Klionsky DJ, Abeliovich H. Aup1p, a yeast mitochondrial protein phosphatase homolog, is required for efficient stationary phase mitophagy and cell survival. J Biol Chem. 2007;282:5617–5624. Epub 2006 Dec 5613. [PubMed]
  • Twig G, Elorza A, Molina AJ, Mohamed H, Wikstrom JD, Walzer G, Stiles L, Haigh SE, Katz S, Las G, et al. Fission and selective fusion govern mitochondrial segregation and elimination by autophagy. EMBO J. 2008;27:433–446. Epub 2008 Jan 2017. [PubMed]
  • Wallace DC. A mitochondrial paradigm of metabolic and degenerative diseases, aging, and cancer: a dawn for evolutionary medicine. Annu Rev Genet. 2005;39:359–407. [PMC free article] [PubMed]
  • Xie Z, Nair U, Klionsky DJ. Atg8 controls phagophore expansion during autophagosome formation. Mol Biol Cell. 2008;19:3290–3298. [PMC free article] [PubMed]
  • Yorimitsu T, Klionsky DJ. Atg11 links cargo to the vesicle-forming machinery in the cytoplasm to vacuole targeting pathway. Mol Biol Cell. 2005;16:1593–1605. Epub 2005 Jan 1519. [PMC free article] [PubMed]
  • Yorimitsu T, Klionsky DJ. Eating the endoplasmic reticulum: quality control by autophagy. Trends Cell Biol. 2007;17:279–285. Epub 2007 May 2003. [PubMed]
  • Zhang Y, Qi H, Taylor R, Xu W, Liu LF, Jin S. The role of autophagy in mitochondria maintenance: characterization of mitochondrial functions in autophagy-deficient S. cerevisiae strains. Autophagy. 2007;3:337–346. Epub 2007 Jul 2009. [PubMed]