Localization of Atg3 to autophagy-related membranes and its enhancement by the Atg8-family interacting motif to promote expansion of the membranes

Authors

  • Machiko Sakoh-Nakatogawa,

    1. Frontier Research Center, Tokyo Institute of Technology, Yokohama 226-8503, Japan
    Search for more papers by this author
  • Hiromi Kirisako,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama 226-8503, Japan
    2. CREST, Japan Science and Technology Agency, Saitama 332-0012, Japan
    Search for more papers by this author
  • Hitoshi Nakatogawa,

    Corresponding author
    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama 226-8503, Japan
    2. CREST, Japan Science and Technology Agency, Saitama 332-0012, Japan
    • Corresponding authors at: Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama 226-8503, Japan. Fax: +81 45 924 5121 (H. Nakatogawa). Frontier Research Center, Tokyo Institute of Technology, Yokohama 226-8503, Japan. Fax: +81 45 924 5121 (Y. Ohsumi).

    Search for more papers by this author
  • Yoshinori Ohsumi

    Corresponding author
    1. Frontier Research Center, Tokyo Institute of Technology, Yokohama 226-8503, Japan
    • Corresponding authors at: Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama 226-8503, Japan. Fax: +81 45 924 5121 (H. Nakatogawa). Frontier Research Center, Tokyo Institute of Technology, Yokohama 226-8503, Japan. Fax: +81 45 924 5121 (Y. Ohsumi).

    Search for more papers by this author

Abstract

The E2 enzyme Atg3 conjugates the ubiquitin-like protein Atg8 to phosphatidylethanolamine (PE) to drive autophagosome formation in Saccharomyces cerevisiae. In this study, we show that Atg3 localizes to the pre-autophagosomal structure (PAS) and the isolation membrane (IM), providing crucial evidence that Atg8-PE conjugates are produced on these structures. We also find that mutations in the Atg8-family interacting motif (AIM) of Atg3 significantly impairs the PAS/IM localization of Atg3, resulting in inefficient IM expansion. It is suggested that the AIM-mediated PAS/IM localization of Atg3 facilitates membrane expansion in these structures probably by ensuring active production of Atg8-PE on the membranes.

1 Introduction

Macroautophagy (hereafter autophagy) is a lysosome/vacuole-mediated degradation system in eukaryotic cells, in which various intracellular materials such as proteins and organelles are sequestered into a double-membrane vesicle called the autophagosome and transported to the lysosome or vacuole [1–3]. Autophagy is classified into two types based on selectivity in sequestration of degradation targets (or cargoes) by the autophagosome: in non-selective autophagy, the autophagosome randomly encloses a portion of the cytoplasm, whereas in selective autophagy, it preferentially or exclusively incorporates specific cargoes [4,5]. In both of these cases, the biogenesis of the autophagosome requires autophagy-related (Atg) proteins [1,2]. In the budding yeast Saccharomyces cerevisiae, in response to autophagy-inducing signals, the Atg proteins form an assembly called the pre-autophagosomal structure (PAS) in the vicinity of the vacuole, and then these proteins are also likely to expand a membrane component in the PAS into a cup-shaped membrane called the isolation membrane (IM) [6–8]. The IM further expands, curves, and ultimately closes to become the autophagosome.

Atg8 is a ubiquitin-like protein that localizes to autophagy-related membranes, including the PAS, IM, and autophagosome, as a conjugate with the lipid phosphatidylethanolamine (PE) [6,9,10]. The Atg8-PE conjugation reaction is catalyzed by the E1 enzyme Atg7, the E2 enzyme Atg3, and the E3 enzyme complex Atg12-Atg5-Atg16, which is composed of the ubiquitin-like protein conjugate Atg12-Atg5 and Atg16 [10–14]. Previous studies suggest that Atg8-PE conjugates play multiple roles in both the formation of the autophagosomal membrane and cargo selection [4,5,15–24]. Therefore, determining the sites of Atg8-PE production, which may differ between membrane formation and selective cargo incorporation, is crucial in order to understand the mechanisms underlying these key events during autophagy. The localization of Atg8 to the PAS and IM suggests that Atg8 is conjugated to PE on membranes in these structures. However, it is also possible that Atg8-PE is first produced in other membranes, and then transferred to these membranes. Determining the localization of the Atg8-PE-producing enzymes would help to resolve this issue. However, the localization of Atg3, which directly catalyzes Atg8 conjugation to PE, remains unknown, whereas Atg12-Atg5-Atg16 localizes to the PAS and IM [7,8].

In this study, we succeed in visualizing the localization of Atg3 by inserting GFP into a flexible region in the protein without impairing its function. It is revealed that Atg3 localizes to the PAS and IM, and that its PAS localization depends on the Atg12-Atg5-Atg16 complex. Moreover, we show that Atg3 localization to the PAS and IM is enhanced by its Atg8-family interacting motif (AIM), and that this enhancement is important for membrane expansion in these structures.

2 Materials and methods

2.1 Yeast strains and plasmids

The yeast strains used in this study, all derived from SEY6210 [25], are listed in Table 1 ; oligo DNA sequences are listed in Table 2 . Gene disruption and tagging were performed by a standard method [26,27]. pFA6a-2×mCherry-natNT2 used for constructing ATG17-2×mCherry strains was gifted by Dr. Hayashi Yamamoto (Tokyo Institute of Technology). To construct the strains expressing Atg3-GFP, DNA fragments were amplified by PCR using the plasmids pFA6a-ATG3-EGFP-kanMX4 or pFA6a-ATG3-EGFP-natNT2 as templates, and the oligo DNA ATG3-kan-integrate-Fw/Rv or ATG3-kan-integrate-Fw/ATG3-nat-integrate-Rv as primers, respectively. These fragments were integrated into the chromosomal ATG3 locus by homologous recombination. The resulting strains express Atg3-GFP with the original promoter and terminator. Strains expressing the AIM mutant version of Atg3-GFP were constructed similarly using pFA6a-atg3W270A L273A -EGFP-kanMX4 or pFA6a-atg3W270A L273A -EGFP-natNT2 as the template for PCR. pFA6a-ATG3-EGFP-kanMX4 and pFA6a-ATG3-EGFP-natNT2 were constructed as follows. First, BamHI and SalI sites were introduced between Gln117 and Ser118 of Atg3 in pRS315-ATG3 [14] using the QuikChange site-directed mutagenesis kit (Agilent Technologies) with the oligo DNA BamHI-SalI-Fw and -Rv. Into these sites was ligated a DNA fragment encoding EGFP, which was obtained by PCR using pFA6a-EGFP-kanMX6 and the oligo DNA BamHI-EGFP-Fw and SalI-EGFP-Rv, resulting in pRS315-ATG3-EGFP. Using this plasmid and the oligo DNA ATG3-EGFP-Fw and -Rv, PCR was performed to obtain DNA fragments encompassing the ATG3-EGFP gene and the ATG3 terminator region, which were then used as pairs of long primers for site-directed mutagenesis to insert them into pFA6a-kanMX4 or pFA6a-natNT2, yielding pFA6a-ATG3-EGFP-kanMX4 and pFA6a-ATG3-EGFP-natNT2. The plasmid pRS424[PCUP1-prApe1] was gifted by Dr. Kuninori Suzuki (The University of Tokyo) [8].

Table Table 1. Yeast strains used in this study
NameGenotypeFiguresReference
SEY6210 MATα leu2-3, 112 ura3-52 his-Δ200 trp1-Δ901 suc2-Δ9 lys2-801   [25]
MAN752SEY6210 atg3::ATG3-natNT2 1BThis study
MAN653SEY6210 atg3::ATG3-EGFP-kanMX4 1B, S1This study
MAN717SEY6210 atg3::atg3AIM mut-EGFP-kanMX4 S1This study
MAN755SEY6210 pho8::PGPD-pho8Δ60-kanMX4 atg3::ATG3-natNT2 1CThis study
MAN756SEY6210 pho8::PGPD-pho8Δ60-kanMX4 atg3::ATG3-EGFP-natNT2 1CThis study
MAN707SEY6210 atg3::ATG3-EGFP-kanMX4 atg11Δ::hphNT1 atg17::ATG17-2×mCherry-natNT2 1D, S2A, S2BThis study
MAN709SEY6210 atg3::ATG3-EGFP-kanMX4 atg11Δ::hphNT1 atg17::ATG17-2×mCherry-natNT2 atg12Δ::zeoNT3 1DThis study
MAN743SEY6210 atg3::ATG3-EGFP-kanMX4 atg11Δ::hphNT1 atg17::ATG17-2×mCherry-natNT2 atg14Δ::zeoNT3 1DThis study
MAN711SEY6210 atg3::ATG3-EGFP-kanMX4 atg11Δ::hphNT1 atg17::ATG17-2×mCherry-natNT2 atg8Δ::zeoNT3 S2AThis study
MAN700SEY6210 atg3::ATG3-EGFP-kanMX4 leu2::2×mCherry-ATG8-hphNT1 2AThis study
MAN719SEY6210 atg3::atg3AIM mut-EGFP-kanMX4 leu2::2×mCherry-ATG8-hphNT1 2AThis study
MAN727SEY6210 atg3::atg3C234A-EGFP-kanMX4 atg11Δ::hphNT1 atg17::ATG17-2×mCherry-natNT2 S2BThis study
MAN748SEY6210 atg5::ATG5-EGFP-kanMX4 atg3::ATG3-natNT2 2BThis study
MAN749SEY6210 atg5::ATG5-EGFP-kanMX4 atg3::atg3AIM mut-natNT2 2BThis study
MAN758SEY6210 atg3::ATG3-natNT2 pep4Δ::kanMX4 2C-EThis study
MAN759SEY6210 atg3::atg3AIM mut-natNT2 pep4Δ::kanMX4 2C-EThis study
Table Table 2. Oligonucleotides used in this study
NameSequence
ATG3-kan-integrate-Fw5′-ATGATTAGATCTACAC-3′
ATG3-kan-integrate-Rv5′-CACAAGAATCCCGCGAAAGGCCGTATCCAGTAGAAACAGAGAATGGTGAAAGTTTTCGACACTGGATGGCGGCGTTAGTA-3′
ATG3-nat-integrate-Rv5′-CACAAGAATCCCGCGAAAGGCCGTATCCAGTAGAAACAGAGAATGGTGAAAGTCCTCTGAGGACATAAAATACACACCGA-3′
BamHI-SalI-Fw5′-GAAACTGAACATGTGCAAGGATCCCCCCCCGTCGACGGTGCATCTGGAGCTAGTACGCCTGCGGGGGGG-3′
BamHI-SalI-Rv5′-CCCCCCCGCAGGCGTACTAGCTCCAGATGCACCGTCGACGGGGGGGGATCCTTGCACATGTTCAGTTTC-3′
BamHI-GFP-Fw5′-CCCGGATCCAGCGCATCGGGTGCA-3′
SalI-GFP-Rv5′-GGGGGGGTCGACATGGATCTTTAGGATCC-3′
ATG3-EGFP-Fw5′-CCGCCAGCTGAAGCTTCGTACGCTGCAGGTCGACGGATCCATGATTAGATCTACAC-3′
ATG3-EGFP-Rv5′-GAGGCAAGCTAAACAGATCTGGCGCGCCTTAATTAACCCGGGAAACAGAGAATGGTG-3′

2.2 Media

Except as noted otherwise, yeast cells were cultured in SD + CA + ATU medium (0.17% yeast nitrogen base without amino acids and ammonium sulfate, 0.5% ammonium sulfate, 0.5% casamino acids, 0.002% adenine sulfate, 0.002% tryptophan, 0.002% uracil, and 2% glucose) at 30 °C. Cells carrying pRS424[PCUP1-prApe1] were cultured in SD + CA + ATU without tryptophan and supplemented with 250 μM CuSO4, which induces Ape1 overexpression from the plasmid. To induce autophagy, cells were treated with 0.2 μg/ml rapamycin or incubated in nitrogen starvation medium SD-N (0.17% yeast nitrogen base without amino acids and ammonium sulfate, and 2% glucose).

2.3 Fluorescence microscopy

Fluorescence microscopy was performed as described previously [28]. Fluorescence intensities of Atg5-GFP were measured by the ‘linescan’ function of the MetaMorph software, and using Microsoft Excel, the fluorescence profiles along IMs were drawn, and the full widths at half maximum were calculated, which were regarded as IM lengths [8].

2.4 ALP assay

To quantitate autophagic activity of yeast cells, the ALP assay was performed as described previously [29].

2.5 Immunoblotting

Sample preparation, SDS–PAGE, Urea-SDS–PAGE (for separation of Atg8 and Atg8-PE), and immunoblotting analysis were performed as described previously [20,30]. Rabbit polyclonal antibodies against Atg3 (anti-Atg3 IN-11) [10] and Atg8 (anti-Atg8-2) [27] were used to detect these proteins. Rabbit polyclonal antibodies against Ape1 (anti-Ape1-2) were gifted by Dr. Maho Hamasaki (Osaka University).

2.6 Electron microscopy

Electron microscopy of autophagic bodies was performed by Tokai EMA, Inc. based on the methods described previously [9]. The size of autophagosomes was measured using Photoshop CS5 (Adobe).

3 Results and discussion

3.1 Construction of functional Atg3 labeled with GFP

To date, it had not been able to investigate the intracellular localization of Atg3, because N-terminal or C-terminal tagging of this protein abolishes its function in autophagy. To overcome this technical obstacle, we inserted GFP into a number of disordered regions or loops in Atg3 chosen on the basis of its crystal structure (Fig. 1 A) [31], and expressed these variants from the chromosomal gene locus. We found that GFP insertion into a disordered region of Atg3 (Fig. 1A) did not significantly affect the protein level and its function in the production of Atg8-PE (Fig. 1B). We also confirmed that both the cytoplasm-to-vacuole targeting (Cvt) pathway (a selective autophagy-related pathway), assessed by vacuolar processing of the cargo protein Ape1 (vacuolar aminopeptidase) [32], and non-selective autophagy induced by nitrogen starvation, assessed by the alkaline phosphatase (ALP) assay [29], occurred normally in cells expressing GFP-inserted Atg3 (Atg3-GFP) (Fig. 1B and C). These data demonstrate that we successfully constructed functional Atg3 labeled with GFP.

Figure 1.

Function and subcellular localization of GFP-inserted Atg3. (A) Crystal structure of Atg3. GFP was inserted between Gln117 and Ser118 in the flexible region (FR). The position of the catalytic cysteine residue Cys234 is also shown. (B) Yeast cells were grown to mid-log phase, treated with rapamycin for 2 h, and examined by immunoblotting using antibodies against Atg3, Atg8, and Ape1. (C) Cells grown to mid-log phase were incubated in SD-N for 4 h and subjected to the ALP assay. The graph shows average values with standard deviations (n = 3). (D) Cells expressing Atg3-GFP and Atg17-2×mCherry were treated with rapamycin for 1 h, followed by fluorescence microscopy. Arrowheads indicate colocalization of Atg3 and Atg17. Bars, 5 μm.

3.2 Atg3 localizes to the PAS and IM

We first observed the localization of Atg3-GFP in cells treated with rapamycin, which induces non-selective autophagy. When Atg17 (Atg17-2×mCherry), a subunit of the PAS-scaffolding complex, is used as a marker, the PAS is observed as one to a few dots per cell [6] (Fig. 1D). Although a large proportion of Atg3-GFP appeared to disperse throughout the cytoplasm, it was also observed as dots. We found that almost all of these dots colocalizes with Atg17, suggesting that Atg3 localizes to the PAS. Previous studies suggested that the Atg12-Atg5-Atg16 complex, which also localizes to the PAS, interacts with Atg3 and enhances its conjugase activity, thereby specifying the site for Atg8-PE production to be autophagy-related structures, including the PAS [7,33]. Consistent with this model, the dots of Atg3 were not observed in cells lacking Atg12, which directly interacts with Atg3 [33,34] (Fig. 1D). Similarly, Atg3 dot formation was abolished by deletion of ATG14, which encodes a subunit of the phosphatidylinositol 3-kinase complex that is required for the PAS localization of Atg12-Atg5-Atg16 [7] (Fig. 1D).

Ape1 self-assembles into a higher-order complex that is selectively sequestered by the autophagosomal membrane and transported into the vacuole [32]. When cells overexpressing Ape1 are treated with rapamycin, the IM associated with the enlarged Ape1 complex can be observed under the fluorescence microscope as elongated or arched Atg8-positive structures [8]. We found that Atg3-GFP also localized to the IM labeled with 2×mCherry-Atg8 (Fig. 2 A). It appears that Atg3 is not evenly distributed on the IM; it is sometimes enriched at the vacuole-IM contact site, similar to Atg1 (see the next paragraph) [8].

Figure 2.

Atg3 binding to Atg8-PE via the AIM enhances the PAS and IM localization of Atg3. (A) 2×mCherry-ATG8 cells carrying an Ape1-overexpression plasmid and expressing Atg3-GFP or the AIM mutant (AIM mut) were treated with rapamycin for 1 h and observed under a fluorescence microscope. Bars, 5 μm. (B) Cells expressing Atg3WT-GFP or Atg3AIM mut-GFP were grown to mid-log phase and treated with rapamycin for 1 h, and then examined by fluorescence microscopy. (C) Wild-type and atg8Δ cells expressing Atg3-GFP and Atg17-2×mCherry were treated with rapamycin for 1 h, and then examined by fluorescence microscopy. Bars, 5 μm. (D) ATG3WT-GFP or atg3C234A-GFP cells carrying pRS316-ATG3 or the empty vector were grown in SD + CA + ATU without uracil to mid-log phase and treated with rapamycin for 1 h, followed by fluorescence microscopy. Bars, 5 μm.

3.3 Atg3 localization to the PAS and IM is enhanced by its binding to Atg8 via the AIM

Next, we investigated how Atg3 associates with the IM. It is possible that the Atg12-Atg5-Atg16 complex is involved in the IM localization of Atg3. On the other hand, recent studies reported that Atg1, a protein kinase responsible for initiating autophagosome formation, localizes to the IM by binding to Atg8-PE on the membrane via its Atg8 family interacting motif (AIM) [8,20,23]. In addition, Atg3 also contains an AIM, which is not essential for conjugation of Atg8 to PE but mediates the high-affinity interaction between Atg3 and Atg8 [35]. Therefore, we asked whether the Atg3-Atg8 interaction via the AIM is involved in Atg3 localization to the IM. The IM localization of Atg3 was strongly impaired in the AIM mutant Atg3W270A L273A [35] (Fig. 2A). A faint Atg3-GFP signal on the IM was still observed in the AIM mutant. These results suggest that the IM localization of Atg3 is enhanced by its AIM-mediated interaction with Atg8.

The PAS localization of Atg3 (in cells not overexpressing Ape1) was also severely reduced by AIM mutations (Fig. 2B), suggesting that AIM-dependent Atg8 binding is important for Atg3 accumulation in the PAS as well as the IM. PAS accumulation of Atg3 was also impaired by deletion of ATG8 (Fig. 2C). In addition, whereas Atg3-GFP containing the Cys234Ala mutation, which abolishes the conjugase activity of Atg3 [10], did not accumulate in the PAS when it was expressed alone (in atg3Δ cells), clear PAS localization of this mutant was observed in the presence of wild-type Atg3, which could produce Atg8-PE at the PAS (Fig. 2D). Taken together, these results suggest that Atg3 accumulates in the PAS via AIM-dependent binding to Atg8-PE produced by Atg3 itself. This mechanism is likely to serve as a positive feedback loop to accelerate the production of Atg8-PE at the PAS and IM.

3.4 AIM-mediated Atg3 localization to the PAS and IM facilitates membrane expansion

Finally, we analyzed the impact of the AIM-mediated PAS/IM localization of Atg3 on autophagosome formation. We first compared the lengths of IMs (visualized by Atg5-GFP [8]) associated with the enlarged Ape1 complex between cells expressing non-tagged wild-type Atg3 and those expressing an AIM mutant. The IM was expanded less efficiently in AIM mutant cells than wild-type cells (Fig. 3 A), suggesting that the AIM-dependent enrichment of Atg3 in the PAS/IM promotes expansion of the IM.

Figure 3.

Mutations in the AIM of Atg3 impair IM expansion. (A) ATG5-GFP cells carrying an Ape1 overexpression plasmid were treated with rapamycin for 1 h and examined by fluorescence microscopy. Bars, 5 μm. The lengths of IMs (n = 31 and 29 for wild-type Atg3 and the AIM mutant, respectively) were measured as described in ‘Experimental procedures’, and the average values are shown with standard deviations. The P value was derived from an unpaired, two-tailed Student's t-test (∗∗∗ P < 0.001). (B–D) pep4Δcells expressing wild-type Atg3 or the AIM mutant were treated with rapamycin for 3 h, and autophagic bodies accumulated in these cells were examined by electron microscopy (C). Bars, 1 μm. The average numbers of autophagic bodies per vacuole section are shown (50 and 49 sections were examined for wild-type and Atg3 AIM mutant cells, respectively) (D). The size (diameter) of cross-sections of 607 and 511 autophagic bodies were measured in wild-type and Atg3 AIM mutant cells, respectively. The numbers of autophagic bodies within indicated sizes were counted, and their percentages of total autophagic body numbers are shown (E).

A previous study reported that the AIM of Atg3 is important for the Cvt pathway, but not for non-selective autophagy [35]. These results suggested that the AIM-dependent enrichment of Atg3 in the PAS and IM might be specifically required for membrane expansion in the Cvt pathway. However, it also remained possible that a defect in autophagosome formation in Atg3 AIM mutant cells, which was assessed by the ALP assay and light microscopy, might not have been detected in the previous study [35]. Yeast cells lacking the vacuolar protease Pep4 accumulate autophagic bodies (inner autophagosomal membrane vesicles released into the vacuolar lumen) [36], whose number and size correlate with those of autophagosomes formed in the cytoplasm. Therefore, we closely examined autophagic bodies in pep4Δ atg3AIM mut cells by electron microscopy (Fig. 3B). The results showed that autophagic bodies reduced not severely but significantly both in number and size in pep4Δ atg3AIM mut cells compared with pep4Δ ATG3WT cells (Fig. 3C and D). It is likely that the AIM-mediated enhancement of the PAS/IM localization of Atg3 also facilitates membrane expansion during non-selective autophagy.

In this study, we revealed that the Atg8-PE producing enzyme Atg3 localizes to the PAS and IM. Together with the previous observation that the Atg3-activating complex Atg12-Atg5-Atg16 exhibits a similar localization [7,8], this finding strongly suggests that Atg8-PE is produced on membranes in these autophagy-related structures. In addition, we found that the PAS/IM localization of Atg3 is significantly enhanced by its AIM-dependent interaction with Atg8. This enhancement was suggested to promote membrane expansion during autophagosome formation. On the other hand, the residual localization of Atg3 to the PAS/IM in the AIM mutant, which was barely detectable due to a relatively high cytoplasmic (background) population, may depend on Atg12-Atg5-Atg16 and produce Atg8-PE essential for autophagosome formation.

How does Atg3 concentrated onto autophagy-related membranes via the AIM-Atg8 interaction promotes membrane expansion? The enzymatic product Atg8-PE plays multiple roles on these membranes, including membrane expansion and selective cargo sequestration into the autophagosome [4,5,15–24]. Moreover, the protein level of Atg8 correlates with the size of the autophagosome [17]. It is possible that concentrating Atg3 onto the PAS/IM leads to generation of more Atg8-PE on membranes in these structures, thereby facilitating their expansion. In addition, elevated levels of Atg8-PE on the PAS/IM may tighten the association of these structures with Ape1 complexes, because Atg8 binds to the Ape1 receptor Atg19 in the complex [32]; in the Cvt pathway, this tight association may be important to expand the membranes along the surface of the Ape1 complex.

A recent study reported that acetylation of Atg3 strengthens the interaction between Atg3 and Atg8 [37]. As with the AIM, this modification may also be involved in Atg8-dependent Atg3 recruitment to autophagy-related membranes. In this manner, many as-yet-unknown interactions and modifications may regulate or fine-tune the autophagic activity of cells.

Acknowledgments

We thank Dr. Kuninori Suzuki, Dr. Hayashi Yamamoto, and Dr. Maho Hamasaki for providing plasmids and antibodies, and the members of our laboratory for materials, discussions, and technical and secretarial support. This work was supported in part by Grants-in-Aid for Scientific Research 26840017 (to M.S.-N.), 25111003 (to H.N.), 25711005 (to H.N.), and 23000015 (to Y.O.) from the .