Tor and the Sin3–Rpd3 complex regulate expression of the mitophagy receptor protein Atg32 in yeast

Autophagy (bulk autophagy) is the random degradation of cytoplasmic components in response to several types of cellular stress and changes in environmental status. After certain cellular stresses, cytosolic double-membrane vesicles, called isolation membranes, emerge and sequester cytoplasmic proteins and organelles as cargos. This is followed by delivery of these cargos into vacuoles in yeast or lysosomes in mammalian cells (Nakatogawa et al., 2009). In contrast, selective autophagy degrades specific proteins or organelles, such as peroxisomes, endoplasmic reticulum, ribosomes and mitochondria (Suzuki, 2012).

The mitochondrion is an organelle that carries out several important metabolic processes, such as oxidative phosphorylation, fatty acid oxidation and the Krebs cycle. Mitochondrial oxidative phosphorylation produces a large amount of energy, which contributes to a range of cellular activities. However, this organelle is also the major source of cellular reactive oxygen species, which cause damage to mitochondrial DNA, lipids and proteins. Moreover, accumulation of this damage is related to aging, cancer, and neurodegenerative disease (Wallace, 2005). Therefore, quality control of mitochondria is important for maintaining cellular homeostasis. Recently, growing evidence has suggested that autophagy selectively degrades damaged mitochondria, and as a result, maintains mitochondrial quality (Narendra et al., 2008; Twig et al., 2008; Narendra et al., 2010). This selective degradation of mitochondria by autophagy is called mitophagy (Lemasters, 2005).

Although many questions remain to be answered, mitophagy is relatively well understood in Saccharomyces cerevisiae (Sc), a fermenting yeast, compared with other organisms. In this organism, autophagy-related protein 32 (ScAtg32) has been identified as a mitophagy-indispensable protein by a yeast genome-wide screen (Kanki et al., 2009c; Kanki et al., 2009b; Okamoto et al., 2009). ScAtg32 is a single-spanning mitochondrial outer membrane protein and plays a role as a mitophagy receptor. When mitophagy is induced, ScAtg32 is phosphorylated, and then ScAtg11, a cytosolic adaptor protein for selective autophagy, interacts with phosphorylated ScAtg32. ScAtg11 then delivers the residential mitochondria to the phagophore assembly site or pre-autophagosomal structure where the isolation membrane is generated for degradation (Aoki et al., 2011). Therefore, phosphorylation of ScAtg32 is considered as the initial step of selective autophagic degradation of mitochondria.

The physiological importance of mitophagy in S. cerevisiae has been recently determined (Kurihara et al., 2012; Richard et al., 2013). When this yeast is grown in glucose-rich conditions, the cells preferentially ferment glucose to obtain ATP and release ethanol. Under this condition, limited numbers of mitochondria are present within the cell. After the cells consumed glucose, they switch metabolism from fermentation to respiration, which aerobically uses ethanol. Mitochondria then proliferate in the cell for efficient respiration. In this way, this yeast converts metabolism between fermentation and respiration, and depending on the type of metabolism, the number of cellular mitochondria increases or decreases. Mitophagy is important for reducing the amount of mitochondria when cells switch metabolism from respiration to fermentation (Kurihara et al., 2012). Accordingly, mitophagy is efficient when this yeast is pre-cultured in a non-fermentable medium and then is transferred to nitrogen-starvation medium supplemented with glucose (Kanki and Klionsky, 2008). Mitophagy is also known as a longevity assurance process, which sustains functional mitochondria and maintains cellular lipid homeostasis in chronologically aging yeast (Richard et al., 2013)

In S. cerevisiae (also called top-fermenting yeast) the amount of mitochondria changes depending on the metabolism. Most organisms from yeast to mammals are low-fermenting (after glycolysis, most of the pyruvic acid is used for oxidative phosphorylation within mitochondria, whereas a small amount is converted to alcohol or acids) or non-fermenting (after glycolysis, almost all of the pyruvic acid is used for oxidative phosphorylation within mitochondria), respectively – and the amount of mitochondria within them does not change much. Therefore, it is unclear whether the regulation and role of mitophagy in non-fermenting or low-fermenting organisms are the same as for S. cerevisiae. Pichia pastoris (Pp) is a methylotrophic yeast that has genetic similarity with S. cerevisiae and is low-fermenting (Cregg, 2007).

In this study, we investigated mitophagy in P. pastoris. We identified the ScAtg32 homolog in P. pastoris (PpAtg32), and characterized this protein. We found that PpAtg32 is barely expressed before induction of mitophagy and is rapidly expressed after induction of mitophagy by starvation. We further studied the expression of PpAtg32 and found that PpAtg32 expression is regulated by the kinase target of rapamycin (Tor) and the PpSin3–PpRpd3 complex. Regulation of PpAtg32 expression is important for efficiently inducing mitophagy. Although PpAtg32 is phosphorylated when mitophagy is induced, this phosphorylation is not regulated by Tor signaling. Finally, we showed that, in S. cerevisiae, the expression of ScAtg32 is also regulated by a similar mechanism to that of PpAtg32.

For observing mitophagy in S. cerevisiae, the method of tagging GFP to mitochondrial protein is widely used (Kissova et al., 2004; Tal et al., 2007; Kanki et al., 2009a; Okamoto et al., 2009). We decided to apply this method to examine mitophagy in P. pastoris. Isocitrate dehydrogenase 1 (Idh1) is a mitochondrial matrix protein. We tagged GFP to the C-terminus of PpIdh1 and expressed PpIdh1–GFP to observe mitochondria by fluorescence microscopy. PpIdh1–GFP was localized in the mitochondria (supplementary material Fig. S1A). After nitrogen starvation, this GFP signal was observed within the vacuole, marked by FM4-64, as a result of mitophagy (Fig. 1A). PpIdh1–GFP was delivered into vacuoles by mitophagy and was hydrolyzed. Because GFP has a typical β-barrel structure and is relatively stable in vacuoles, PpIdh1 or the PpIdh1–GFP linker region was initially degraded, and GFP was detected as an intact free form by immunoblotting. Therefore, the amount of free GFP is semi-quantitative evidence of mitophagy (this method is termed the Idh1–GFP processing assay). We found that the amount of free GFP increased depending on the duration of starvation (Fig. 1B). In strains deleted for PpATG1, which is an essential gene for isolation membrane extension, accumulation of vacuolar GFP, as shown by microscopy, and PpIdh1–GFP processing, as shown by immunoblotting, were completely blocked (Fig. 1A,B). Similarly, deletion of an adaptor protein for selective autophagy, PpAtg11, also completely blocked mitophagy in P. pastoris (Fig. 1A,B). These findings suggest that mitochondria are selectively degraded by autophagy in P. pastoris and that the ScAtg32 homolog or counterpart is present in this organism.

We performed a computational homology search to identify the ScAtg32 homolog or counterpart of P. pastoris. We initially found a candidate ScAtg32 homolog in Ogataea parapolymorpha (Pichia angusta) (EMBL: EFW94927.1), and then found a candidate homolog in P. pastoris (EMBL: CAY71556.1, designated PpAtg32 hereafter). Recently, Farre et al. (Farre et al., 2013) also identified this protein as the ScAtg32 homolog of P. pastoris. Similar to ScAtg32, PpAtg32 has a predicted transmembrane domain with a long N-terminus and short C-terminus regions. We found that although the sequence similarity between ScAtg32 and PpAtg32 was not high, the N-terminus Atg11-binding region, including the phosphorylation sites in Atg32, was well conserved (Fig. 1C). GFP-tagged PpAtg32 colocalized with the mitochondrial marker MitoTracker Red (Fig. 1D), which suggests that this protein is localized in mitochondria. To determine whether PpAtg32 is a mitophagy-related protein, we assessed mitophagy in Ppatg32Δ cells. As expected, Ppatg32Δ cells did not show mitophagy (Fig. 1A,B).

We then examined whether PpAtg32 is necessary for other types of autophagy. To observe macroautophagy, including bulk autophagy and selective autophagy, we used the yellow fluorescence protein (YFP)–PpAtg8 processing assay, which can be used to examine macroautophagy semi-quantitatively in a similar manner to the PpIdh1–GFP processing assay. Cells expressing YFP–PpAtg8 were cultured in synthetic glucose medium and transferred to synthetic methanol (SM) medium to induce lag-phase autophagy (Yamashita et al., 2009). The wild-type strain showed accumulation of free YFP when transferred to SM medium, while the Ppatg1Δ strain did not show accumulation of free YFP (Fig. 2A). In the Ppatg11Δ and Ppatg32Δ strains, however, accumulation of free YFP was observed, but it was weaker compared with that in the wild-type strain. Because mitophagy and other types of selective autophagy are a type of macroautophagy, Atg8 is required for these processes. Accordingly, processed YFP from YFP–Atg8 in Ppatg32Δ or Ppatg11Δ cells might be decreased as a result of deficiency of mitophagy or selective autophagy. The Cvt pathway is a type of selective autophagy characterized only in yeast. To observe the Cvt pathway, we examined the processing of precursor-formed aminopeptidase1 (PpprApe1) fused with cyan fluorescence protein (CFP). When PpprApe1 is processed in vacuoles, a small fragment fused with CFP is detected by immunoblotting (Yamashita et al., 2009). We observed that Ppatg32Δ cells showed similar levels of free CFP compared with those in wild-type cells, whereas in Ppatg1Δ and Ppatg11Δ cells the Cvt pathway was almost completely blocked (Fig. 2B). Finally, we studied pexophagy. To monitor pexophagy, we assessed autophagic degradation of peroxisomal alcohol oxidase (PpAox) by immunoblotting. There was proliferation of peroxisomes and PpAox was strongly expressed when cells were cultured in methanol-containing medium (Fig. 2C). Cells were then transferred to starvation medium to induce macropexophagy (Till et al., 2012). We observed that PpAox was quickly degraded in wild-type and Ppatg32Δ cells, whereas in autophagy-deficient Ppatg1Δ cells and selective autophagy-deficient Ppatg11Δ cells, degradation of PpAox was blocked (Fig. 2C). Based on these findings, we conclude that PpAtg32 is essential for mitophagy, but not for other types of autophagy.

To determine the role of PpAtg32 in mitophagy, we developed an anti-PpAtg32 antibody and observed PpAtg32 by immunoblotting. Surprisingly, PpAtg32 was barely expressed before induction of mitophagy by starvation in wild-type cells, whereas another mitochondrial protein, PpPor1, was highly expressed (Fig. 3A,B). After starvation, PpAtg32 was rapidly expressed to detectable levels (2 h), and then gradually decreased in wild-type cells (Fig. 3A). Because ScAtg32 is degraded by mitophagy (Aoki et al., 2011), we speculated that PpAtg32 might be degraded by mitophagy. Therefore, we assessed the expression of PpAtg32 in mitophagy-deficient Ppatg11Δ cells. As expected, the amount of PpAtg32 continuously increased following the period of starvation, presumably because mitophagic degradation was prevented (Fig. 3A). This PpAtg32 expression pattern was not affected if cells were pre-cultured in non-fermentable medium before starvation (Fig. 3B). Interestingly, the molecular mass of PpAtg32 was increased after starvation (Fig. 3A,B). Because ScAtg32 is phosphorylated when mitophagy is induced (Aoki et al., 2011), we speculated that the molecular mass gain of PpAtg32 was also due to phosphorylation. Therefore, we treated the cell lysates with λ protein phosphatase (λ PPase). The molecular mass of PpAtg32 was decreased by λ PPase treatment, which indicated that a large part of the molecular mass shift of PpAtg32 was caused by its phosphorylation (Fig. 3C).

In S. cerevisiae, the phosphorylation of ScAtg32 is indispensable for mitophagy, especially phosphorylation of serine 114 on ScAtg32. We found that serine 114 on ScAtg32 and the surrounding sequences were almost completely conserved on PpAtg32 (corresponding to serine 159 on PpAtg32, Fig. 1C). To study the importance of phosphorylation of serine 159 on PpAtg32 for mitophagy, we expressed HA-tagged wild-type PpAtg32 (PpAtg32^WT) and a mutant in which this conserved serine residue was replaced by an alanine residue (PpAtg32^S159A) in Ppatg32Δ cells, and assessed mitophagy. As expected, mitophagy was rescued by expressing HA–PpAtg32^WT, but not HA–PpAtg32^S159A (Fig. 3D). We then mutated the conserved serine residue to an aspartic acid or glutamic acid residue, which can mimic the phosphorylated residue (HA–PpAtg32^S159D and HA–PpAtg32^S159E, respectively). Although HA–PpAtg32^S159E did not rescue mitophagy, HA–PpAtg32^S159D partially rescued mitophagy (Fig. 3E). We also mutated the conserved serine residue to a threonine or tyrosine residue, which can be phosphorylated by kinases (HA–PpAtg32^S159T and HA–PpAtg32^S159Y, respectively). In this case, HA–PpAtg32^S159T rescued mitophagy (Fig. 3F). Because both serine and threonine are phosphorylatable amino acids, and aspartic acid is a phosphomimetic residue, these findings suggest that phosphorylation of amino acid residue 159 on PpAtg32 is important for inducing mitophagy. We then speculated that overexpression of phosphomimetic PpAtg32^S159D might induce mitophagy, even under conditions of growth. Therefore, we overexpressed the PpAtg32^S159D mutant in PpAtg32Δ cells and assessed mitophagy. However, overexpression of PpAtg32^S159D did not induce mitophagy without starvation (supplementary material Fig. S1B). Presumably, the macroautophagic process, which is essential for mitophagy, was not activated under conditions of growth.

Despite its importance, serine 114 is not efficiently phosphorylated in S. cerevisiae (Aoki et al., 2011). To test how efficiently serine 159 on PpAtg32 is phosphorylated during mitophagy, we compared the phosphorylation of HA–PpAtg32^WT and HA–PpAtg32^S159A in Ppatg11Δ cells. We observed that HA–PpAtg32^WT and HA–PpAtg32^S159A were similarly phosphorylated, suggesting that serine 159 on PpAtg32 is not predominantly phosphorylated (Fig. 3G). Taken together, these results suggest that because serine 114 on ScAtg32 is not efficiently phosphorylated, but is important for mitophagy, the corresponding serine 159 on PpAtg32 is also partially phosphorylated and plays an important role in mitophagy.

As shown above, we found two phenomena regarding PpAtg32 under mitophagy-inducing conditions: an increase in PpAtg32 expression and phosphorylation of PpAtg32. We then focused on PpAtg32 expression. We first measured the amount of PpAtg32 mRNA by a quantitative PCR (qPCR). We observed that PpATG32 mRNA expression was strongly stimulated by starvation (Fig. 4A). Starvation inhibits the protein kinase Tor and inhibition of Tor induces autophagy (Noda and Ohsumi, 1998). Because PpAtg32 expression was induced by starvation, we speculated that PpAtg32 expression might be regulated by Tor. To examine this possibility, we used rapamycin, which efficiently inhibits Tor (supplementary material Fig. S2A), and assessed PpAtg32 mRNA and protein expression. As expected, PpAtg32 mRNA and protein expression levels were increased by rapamycin treatment, although the level of expression was weaker compared with that induced by starvation (Fig. 4B; supplementary material Fig. S2B). Interestingly, PpAtg32 induced by rapamycin was not phosphorylated (Fig. 4B). Therefore, we decided to examine whether rapamycin, which cannot phosphorylate PpAtg32, can induce mitophagy by using the PpIdh1–GFP processing assay. As shown in Fig. 4C, the starvation stimulus led to efficient phosphorylation of PpAtg32 and induced mitophagy, whereas rapamycin did not lead to PpAtg32 phosphorylation and did not induce mitophagy. This lack of mitophagy upon rapamycin treatment might be due to the lower expression of PpAtg32 compared with that induced by starvation. To examine this possibility, we exogenously expressed PpAtg32 under the CUP1 promoter (which is induced by Cu²⁺) and assessed mitophagy. Although the expression level of PpAtg32 was similar for the starvation stimulus and for rapamycin treatment, rapamycin treatment did not induce mitophagy (Fig. 4D). Finally, we expressed PpAtg32 by rapamycin treatment and then shifted to starvation in Ppatg11Δ cells. In this case, expressed PpAtg32 was phosphorylated only after starvation, which suggested that starvation activated a signaling pathway of PpAtg32 phosphorylation, but this signaling pathway was not related to Tor (Fig. 4E). Our results suggest that Tor predominantly regulates PpAtg32 expression. However, this pathway is not sufficient to trigger mitophagy.

We showed above that PpAtg32 expression was regulated by Tor. However, the expression of PpAtg32 itself could not trigger mitophagy. These findings raised the issue of whether the amount of PpAtg32 expression is important for mitophagy or whether a small amount of PpAtg32 is sufficient for inducing mitophagy. To investigate this issue, we constructed PpAtg32 under-expressing cells (exogenous PpAtg32 under the CUP1 promoter, induced by a low concentration of Cu²⁺) and PpAtg32 overexpressing cells (endogenous PpAtg32 plus exogenous PpAtg32 with the CUP1 promoter, induced by a high concentration of Cu²⁺) and assessed mitophagy by using the PpIdh1–GFP assay. Fig. 4F shows that PpAtg32 under-expression resulted in weak mitophagy compared with the wild type, whereas PpAtg32 overexpression did not affect mitophagy. This finding suggests that sufficient expression of PpAtg32 is required for efficient mitophagy.

To gain more insight into PpAtg32 expression, we attempted to identify the promoter region of PpAtg32. We constructed an HA-tagged PpATG32 expression vector, which has several lengths upstream of the start codon region as a promoter. We then transformed the vector into Ppatg32Δ cells and assessed PpAtg32 expression (Fig. 5A). PpAtg32 was normally expressed after starvation only if there were more than 400 bp upstream of the start codon (Fig. 5A). Surprisingly, if there were only 200 bp upstream of the start codon, PpAtg32 was expressed, even before starvation (Fig. 5A). This finding suggests that the upstream 200-bp region contains the promoter and the upstream 201- to 400-bp region contains the upstream repression sequence (URS) of PpATG32. To study the promoter of PpATG32 in more detail, we further shortened the sequence upstream of the start codon of PpATG32 and assessed PpAtg32 expression (Fig. 5B). We found that a promoter with 200 bp, 150 bp and 100 bp upstream of the start codon expressed PpAtg32, but not 50 bp upstream of the start codon. This finding suggests that the promoter of PpATG32 is between 51 and 100 bp upstream of the start codon. We then searched for the URS in detail. We deleted the 201–250, 251–300, 301–350, and 351–400-bp regions upstream of the start codon and assessed PpAtg32 expression. Only upon deletion of 301–350 bp upstream of the start codon, was PpAtg32 expressed without starvation (Fig. 5C). This finding suggests that the 301–350-bp region upstream of the start codon contains the URS of PpATG32. Notably, mitophagy was not induced just because PpAtg32 was expressed before starvation (Fig. 5A).

To identify the transcription regulator of PpATG32, we examined several transcription regulators that are thought to be involved in autophagy in P. pastoris (Mukaiyama et al., 2002). However, none of them were affected by changes in PpAtg32 expression (supplementary material Fig. S2C).

Atg8 is an ubiquitin-like protein that is important for autophagosome formation, and it is regulated by Tor signaling. Recently, it has been reported that the Ume6–Sin3–Rpd3 complex functions as a transcription repressor, and directly interacts with the ATG8 promoter and suppresses Atg8 expression in S. cerevisiae (Bartholomew et al., 2012). Because the Atg8 expression pattern is similar to the expression pattern of PpAtg32, we speculated that this complex might be involved in PpAtg32 expression. Therefore, we attempted to identify Ume6, Sin3 and Rpd3 homologs in P. pastoris, and we succeeded in identifying PpSin3 and PpRpd3. We then generated Ppsin3Δ and Pprpd3Δ strains, and assessed PpAtg32 protein and PpATG32 mRNA expression. As expected, PpAtg32 was strongly expressed without starvation in Pprpd3Δ and Ppsin3Δ strains, and the phenotype was suppressed by exogenous expression of PpRpd3–HA and PpSin3–HA, respectively (Fig. 5D,E). This finding suggests that PpRpd3 and PpSin3 function as transcription repressors and bind to the promoter of PpATG32, which is presumably localized 301–350 bp upstream of the start codon. To confirm whether the PpSin3–PpRpd3 complex binds to the promoter of PpATG32, we performed a chromatin immunoprecipitation (ChIP) assay (Fig. 5F). HA–PpSin3 bound to the promoter of PpINO1 (positive control) ∼4 times higher, and to the promoter of PpATG32 ∼2.5 times higher, than the control (promoter of PpCKA1). Based on these findings, we conclude that the PpSin3–PpRpd3 complex binds to the promoter of PpATG32 and represses PpAtg32 expression, but the binding efficiency is lower compared with that to the promoter of PpINO1. Finally, we assessed mitophagy in Ppsin3Δ and Pprpd3Δ strains (Fig. 5G). Although PpAtg32 was overexpressed in Ppsin3Δ and Pprpd3Δ cells before starvation, it did not increase mitophagy, but rather decreased it instead.

To induce mitophagy in S. cerevisiae, cells are generally pre-cultured in non-fermentable medium to induce proliferation of mitochondria and are then transferred to starvation conditions. In a previous study, ScAtg32 was fully expressed during pre-culture in non-fermentable medium and its levels gradually decreased in starvation conditions (Aoki et al., 2011). In contrast, ScAtg32 was barely expressed in fermentable medium. Therefore, we believe that expression of ScAtg32 is determined by the carbon source of the medium and the amount of mitochondria that have proliferated. However, in the current study, we found that expression of PpAtg32 was regulated by Tor and the PpSin3–PpRpd3 complex in P. pastoris. Therefore, we decided to investigate whether Tor and the ScUme6–ScSin3–ScRpd3 complex are involved in ScAtg32 expression in S. cerevisiae. We observed that although ScAtg32 was barely expressed when cultured in fermentable medium, it was expressed by starvation and rapamycin (Fig. 6A,B). When a component of the Ume6 complex was deleted, ScAtg32 was expressed and ScATG32 mRNA was increased, even in fermentable medium before starvation (Fig. 6C,D). We then assessed whether HA–ScUme6 bound to the promoter of ScATG32 by using a ChIP assay. As shown in Fig. 6E, HA–ScUme6 bound to the promoter of ScATG32 and ScATG8 ∼2.5 times higher than the control (promoter of ScTFC1). Therefore, we conclude that expression of ScAtg32 is regulated by Tor and the Ume6 complex. Presumably, the Ume6 complex is regulated downstream of Tor and functions as a repressor of ATG32 expression in P. pastoris and S. cerevisiae.

Protein kinase Rim15 phosphorylates Ume6, disrupts the Ume6–Sin3–Rpd3 complex and induces Atg8 expression during starvation in S. cerevisiae (Bartholomew et al., 2012). Because the expression of Atg32 is suppressed by the Ume6 complex, we speculated that Rim15 is also involved in Atg32 expression. Therefore, we first assessed ScUme6 phosphorylation by immunoblotting during starvation. Consistent with a previous report (Bartholomew et al., 2012), ScUme6 was phosphorylated during starvation and this phosphorylation was not observed in Scrim15Δ cells (supplementary material Fig. S2D). Because ScAtg32 was efficiently expressed under non-fermenting conditions (Fig. 6A), we speculated that ScUme6 might be phosphorylated in this condition. As shown in supplementary material Fig. S2E, ScUme6 was phosphorylated when wild-type cells were cultured in non-fermentable YPL medium. Interestingly, the phosphorylation of ScUme6 under non-fermenting conditions was observed in Scrim15Δ cells, suggesting that this phosphorylation of ScUme6 is independent of Rim15 in these conditions. We then assessed ScAtg32 protein and ScATG32 mRNA expression in Scrim15Δ cells. Although ScUme6 was phosphorylated by ScRim15 during starvation (supplementary material Fig. S2D), ScAtg32 protein and ScATG32 mRNA expression levels were the same or only slightly decreased in Scrim15Δ cells compared with wild-type cells before and after starvation (Fig. 7A,B). Similarly in P. pastoris, PpAtg32 protein and PpATG32 mRNA expression levels were not affected by deletion of PpRIM15 (Fig. 7C,D). To test whether Rim15 is involved in Atg8 expression, we assessed ATG8 mRNA expression when RIM15 was deleted in S. cerevisiae and P. pastoris. We found that the expression of ATG8 mRNA was not affected by deletion of RIM15 in S. cerevisiae and P. pastoris (Fig. 7E,F). This result is different from a previous report (Bartholomew et al., 2012), probably because of different experimental conditions; for example, in this report Atg8 protein was measured after a shorter starvation period.

Because protein kinase Rim11 and Mck1 are involved in Ume6 phosphorylation (Xiao and Mitchell, 2000), we speculated that Rim11 or Mck1 might be involved in Atg32 expression. However, in the current study, ScATG32 mRNA expression was only slightly affected in Scrim11Δ and Scmck1Δ strains (Fig. 7G,H). Similarly, ScATG8 mRNA expression was only slightly affected in Scrim11Δ and Scmck1Δ strains (Fig. 7I). Based on these findings, we conclude that the protein kinases Rim15, Rim11 and Mck1 are only marginally involved in Atg32 and Atg8 expression.

During the last few years, there has been great progress in the understanding of mitophagy in S. cerevisiae. Progress has been made in identification and characterization of ScAtg32. S. cerevisiae is a fermenting yeast, and it is required for pre-culture in non-fermentable medium to efficiently induce mitophagy (supplementary material Fig. S3A). This characteristic might make it difficult to understand the signaling pathways that regulate mitophagy in yeast. In the current study, we used P. pastoris as a model organism and found that PpAtg32 was expressed when mitophagy was induced by starvation. Further experiments showed that the expression of PpAtg32 was regulated by Tor and the PpSin3–PpRpd3 complex. Although Tor signaling regulated PpAtg32 expression, Tor was not involved in PpAtg32 phosphorylation. The regulation of Atg32 expression by Tor and the Ume6–Sin3–Rpd3 complex was conserved in S. cerevisiae.

Notably, mitophagy was easily induced in P. pastoris compared with in S. cerevisiae. In S. cerevisiae, mitophagy was slightly induced when cells were pre-cultured in fermentable medium, and it was efficiently induced when cells were pre-cultured in non-fermentable medium (supplementary material Fig. S3A,B). In contrast, mitophagy was efficiently induced independently of pre-culture medium in P. pastoris (supplementary material Fig. S3C). One of the reasons for this finding is that S. cerevisiae is a fermenting yeast and has less mitochondria available for degradation during fermentation (e.g. when cultured in glucose-containing medium; supplementary material Fig. S3A, ScIdh1–GFP and ScPor1). However, P. pastoris is a low-fermenting yeast that always has a certain minimal amount of mitochondria, independently of the type of carbon source available in the medium (supplementary material Fig. S3C). Expression of ScAtg32 was higher in non-fermenting medium than in fermenting medium in S. cerevisiae (Fig. 6A). This weak ScAtg32 expression might be another reason why mitophagy was induced less when cells were pre-cultured in fermentable medium than in non-fermentable medium.

Sin3 and Rpd3 form a histone deacetylase complex. The Sin3–Rpd3 complex binds the promoter region, deacetylates histones of the region and suppresses gene expression (Kadosh and Struhl, 1998). To localize to the promoter region, Sin3–Rpd3 needs to bind with a DNA-binding transcription factor. We showed that in S. cerevisiae and P. pastoris, Sin3 and Rpd3 suppressed Atg32 expression (Fig. 5D; Fig. 6C). In S. cerevisiae, the transcription factor Ume6 is required to localize the Sin3–Rpd3 complex to the promoter region (Rundlett et al., 1998). Although we could not identify the Ume6 homolog in P. pastoris by using homology search programs, there should be a counterpart transcription factor in this organism.

Because PpAtg32 expression was increased in Pprpd3D or Ppsin3Δ cells (Fig. 5D), we expected that mitophagy might be increased in these cells. However, as shown in Fig. 5G, mitophagy was decreased in these cells. One of the reasons for this finding is that overexpression of PpAtg32 did not always increase mitophagy (Fig. 4F). Another reason for this finding might be that Pprpd3Δ or Ppsin3Δ strains were, to some degree, unhealthy compared with the wild-type strain. As shown in supplementary material Fig. S4A, the cellular growth of Pprpd3Δ or Ppsin3Δ strains in YPD medium was slower than that of the wild-type strain. Because PpRpd3 and PpSin3 are involved in diverse transcriptional repression and activation processes, deletion of PpRPD3 or PpSIN3 affected cellular growth and activity, which might have indirectly affected mitophagy.

P. pastoris is a methylotrophic yeast and is often used as a model organism to study pexophagy. As an organelle-specific selective autophagy types, mitophagy and pexophagy have been extensively studied, and both require an adaptor protein, Atg11. However, it is unclear how cells distinguish between mitochondria and peroxisome degradation, or if there is any cross talk between mitophagy and pexophagy. Because mitophagy is easily induced in P. pastoris, this organism might be useful for understanding the relationship between mitophagy and pexophagy.

The yeast strains and plasmids used in this study are listed in supplementary material Tables S1 and S2, respectively. Yeast cells were grown in rich medium (YPD, 1% yeast extract, 2% peptone and 2% glucose), lactate medium (YPL, 1% yeast extract, 2% peptone and 2% lactate), glycerol medium (YPGly, 1% yeast extract, 2% peptone and 3% glycerol), synthetic minimal medium with glucose (SD, 0.67% yeast nitrogen base without amino acids, 2% glucose, 100 µg/ml arginine and 100 µg/ml histidine; SD without copper, 0.67% yeast nitrogen base without copper and amino acids, 2% glucose, 100 µg/ml arginine and 100 µg/ml histidine) and synthetic minimal medium with methanol (SM, 0.67% yeast nitrogen base, 0.5% methanol, 100 µg/ml arginine and 100 µg/ml histidine). Nitrogen starvation experiments were performed in synthetic minimal medium lacking nitrogen (SD-N, 0.17% yeast nitrogen base without amino acids and ammonium sulfate, and 2% glucose; SD-N without copper, 0.17% yeast nitrogen base without copper, amino acids and ammonium sulfate, and 2% glucose).

Anti-PpAtg32 antiserum was produced by immunizing rabbits with the recombinant His-tagged N-terminus (300 amino acids) of PpAtg32. The serum was affinity purified using the recombinant GST-tagged N-terminus (300 amino acids) of PpAtg32-conjugated Sepharose. Anti-PpPor1 antibody was produced by immunizing rabbits with recombinant GST-tagged PpPor1 and affinity purifying the serum using recombinant GST-tagged PpPor1 bound to a polyvinylidene fluoride membrane. The specificity of these antibodies is shown in supplementary material Fig. S4B,C. Anti-HA antibody (Sigma-Aldrich, St Louis, MO), anti-beta actin antibody (Abcam, Cambridge, MA), anti-GFP antibody (Takara Bio Inc., Shiga, Japan), anti-Ume6 antibody (GeneTex Inc., Irvine, CA), anti-Rsp6 antibody (Abcam), anti-phospho-Rsp6 antibody (Cell Signaling Technology, Beverly, MA) and anti-Pgk1 antibody (Nordic Immunological Laboratories, Tilburg, The Netherlands) were used for immunoblotting. Anti-ScAtg32 and anti-PpAox antibodies were as described previously (Sakai et al., 1998; Aoki et al., 2011).

To observe mitophagy in S. cerevisiae, the ScIdh1–GFP processing assay was carried out as described previously (Kanki et al., 2009a). To observe mitophagy in P. pastoris, PpIdh1–GFP-expressing cells were grown in YPD medium to mid-log phase. For starvation, the cells were washed in sterilized water two times and cultured in SD-N medium. The cells were harvested at 0, 2 and 4 hours, and the cell lysates equivalent to A₆₀₀ = 0.1 unit of cells were subjected to immunoblot analysis. To observe macroautophagy and the Cvt pathway, the YFP–PpAtg8 and CFP–PpprApe1 processing assays were carried out, respectively (Yamashita et al., 2009). YFP–PpAtg8- or CFP–PpprApe1-expressing cells were cultured in SD medium until the mid-log phase and transferred into SM medium. The cells were collected, and cell lysates were subjected to the immunoblot assay. To observe pexophagy, cells were cultured in SD medium until mid-log phase and transferred to SM medium for 10 hours. The cells were washed in sterilized water two times and cultured in SD-N medium. Degradation of PpAox was monitored by immunoblot assay using anti-PpAox antibody.

The Ppatg11Δ strain was cultured in YPD medium until the mid-log phase, and then transferred to SD-N medium for 6 hours. Cells were collected and lysed with glass beads in phosphatase buffer supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) and protease inhibitors. After centrifugation at 10,000 g for 10 minutes, the supernatant was incubated with λ PPase (New England Biolabs, Ipswich, MA) for 1 hour at 30°C.

Cells expressing GFP-tagged PpAtg32 under the control of the CUP1 promoter were grown in YPD medium with 10 µM Cu²⁺ until the mid-log growth phase. To label mitochondria, cells were incubated with 1 µM of MitoTracker Red CMXRos (Molecular Probes, Carlsbad, CA) at 30°C for 30 minutes. To observe mitophagy, PpIdh1–GFP-expressing cells were grown in YPD medium until the mid-log phase, and were then incubated with 100 µM PMSF and 1.6 µg/ml N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl) pyridinium dibromide (FM 4-64; Biotium Inc., Hayward, CA) at 30°C for 30 minutes. After cells were washed with sterilized water two times and cultured with SD-N medium for 3 hours, fluorescence signals were visualized by using a fluorescence microscope (ECLIPSE TE2000-U; Nikon, Tokyo, Japan) with a 100× 1.45 NA oil immersion objective at room temperature. Images were captured with a charged-coupled device camera (RetigaEXi blue; Qimaging, Surrey, Canada).

Cells were cultured in YPD medium until the mid-log phase. For starvation, the cells were washed with sterilized water two times and transferred to SD-N medium. Total RNA was extracted using the RNeasy Mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. The concentration and quality of total RNA were determined using a spectrophotometer (Nanodrop-1000; Thermo Fisher Scientific Inc., Waltham, MA). To obtain cDNA, total RNA was reverse transcribed using the PrimeScript RT Reagent kit (Takara Bio Inc.). To measure mRNA expression, quantitative real-time PCR was performed with the following primers: PpATG32, 5′-CAAATATGGAGCTGGATTTCTGTGG-3′ and 5′-CTTTTCACGAGACTGGCTCATTTGT-3′; PpACT1, 5′-TTGGCCGGTAGAGATTTGAC-3′ and 5′-ACAGATGGGTGGAACAAAGC-3′; PpPGK1, 5′-GCCACGGTTGAGTCTGGTAT-3′ and 5′-GCACCACCGAACTTCTTAGC-3′; ScATG32: 5′-GCCACTGCATCTCCTTCTTC-3′ and 5′-TTCATTTTGCCCCAAGTCTC-3′; ScATG8, 5′-TGTGATTTGCGAAAAAGCTG-3′ and 5′-CCTTCTCAGGGGGTAGCATA-3′; and ScACT1, 5′-TCCGGTGATGGTGTTACTCA-3′ and 5′-TCCGGTGATGGTGTTACTCA-3′.

HA–PpSin3-expressing and HA–PpSin3-nonexpressing (for control) P. pastoris cells or HA–ScUme6-expressing and HA–ScUme6-nonexpressing (for control) S. cerevisiae cells were used. Immunoprecipitation was carried out using anti-HA antibody conjugated beads by a method described previously (Aparicio et al., 2005). To quantify the immunoprecipitated DNA, quantitative real-time PCR was performed with the following primers: PpATG32, 5′-GGCCAGCCTTATCATGGGTG-3′ and 5′-GGTTCTGTGATGGATGACTCA-3′; PpINO1, 5′-GCTCCTTGCACGCTCTATTG-3′ and 5′-GTAACCAGGGCTGTAACAACG-3′; PpCKA1, 5′-GCGGAAGCTGAGGTTGATAG-3′ and 5′-GGAAAAACGGCACCTTCATA-3′; ScATG32: 5′-GTTTGACGCACCCCTTTTAC3′ and 5′-AAAACGAGAAATGGGCTTTG-3′; ScATG8, 5′-ACCCGTGAAATCATAGCACA-3′ and 5′-ATCAATCCCCTCCTCAACCT-3′; ScINO1, 5′-GATGCGGAATCGAAAGTGTT-3′ and 5′-GCTTTCTCTGCTCCATGTGAA-3′; and ScTFC1, 5′-AGGCGAAACATTGCAAGACT-3′ and 5′-TCTCCCAAGCAGTTGATCCT-3′.

We thank Daniel J. Klionsky (University of Michigan, Ann Arbor, MI) for providing the yeast strains and plasmids. We appreciate support from colleagues at the Department of Clinical Chemistry and Laboratory Medicine, Kyushu University Hospital, especially Kazue Hayashida.