Atg4 plays an important role in efficient expansion of autophagic isolation membranes by cleaving lipidated Atg8 in Saccharomyces cerevisiae

Autophagy, an intracellular degradation system, is highly conserved among eukaryotes from yeast to mammalian cells. In the yeast Saccharomyces cerevisiae, most Atg (autophagy-related) proteins, which are essential for autophagosome formation, are recruited to a restricted region close to the vacuole, termed the vacuole-isolation membrane contact site (VICS), upon induction of autophagy. Subsequently, the isolation membrane (IM) expands and sequesters cytoplasmic materials to become a closed autophagosome. In S. cerevisiae, the ubiquitin-like protein Atg8 is C-terminally conjugated to the phospholipid phosphatidylethanolamine (PE) to generate Atg8-PE. During autophagosome formation, Atg8-PE is cleaved by Atg4 to release delipidated Atg8 (Atg8G116) and PE. Although delipidation of Atg8-PE is important for autophagosome formation, it remains controversial whether the delipidation reaction is required for targeting of Atg8 to the VICS or for subsequent IM expansion. We used an IM visualization technique to clearly demonstrate that delipidation of Atg8-PE is dispensable for targeting of Atg8 to the VICS, but required for IM expansion. Moreover, by overexpressing Atg8G116, we showed that the delipidation reaction of Atg8-PE by Atg4 plays an important role in efficient expansion of the IM other than supplying unlipidated Atg8G116. Finally, we suggested the existence of biological membranes at the Atg8-labeled structures in Atg8-PE delipidation-defective cells, but not at those in atg2Δ cells. Taken together, it is likely that Atg2 is involved in localization of biological membranes to the VICS, where Atg4 is responsible for IM expansion.


Introduction
Eukaryotic cells have an intracellular degradation system, macroautophagy (hereafter autophagy), which is conserved from yeast to mammals [1]. When autophagy is induced by starvation or other stresses, most Atg proteins are recruited to a dot close to the vacuole, termed the vacuole-isolation membrane contact site (VICS) [2]. Next, the isolation membrane (IM), a membrane sac, expands by engulfing cytoplasmic materials and subsequently closes to form the autophagosome, a double-membrane organelle [3,4]. The outer membrane of a completed autophagosome then fuses with the vacuole (lysosome in mammals) to release the autophagic body and its inner membrane compartment for degradation [3]. Currently, 41 Atg (autophagy-related) proteins have been identified in Saccharomyces cerevisiae, of which 19 are essential for autophagosome formation.
Among the Atg proteins involved in autophagosome formation, Atg8, a ubiquitin-like protein, is a reliable marker for monitoring progression of autophagy because it localizes to the VICS, IM, and autophagosome (hereafter, autophagy-related structures) and is ultimately transported into the vacuole [5][6][7]. Initially, Atg8 is synthesized with an arginine (residue 117) at the C-terminus, but is processed by the cysteine protease Atg4 to produce glycine-exposed Atg8 (Atg8 G116 ) [8]. Subsequently, Atg8 undergoes a ubiquitin-like conjugation reaction involving Atg7 and Atg3, an E1-and an E2-like enzyme, respectively; the Atg16ÁAtg5-Atg12 complex serves as an E3-like enzyme in this reaction. Finally, Atg8 is conjugated to the phospholipid phosphatidylethanolamine (PE), which anchors it to membranes [9]. Atg4 also enzymatically cleaves the amide bond between Atg8 and PE to facilitate their recycling and promote autophagosome formation [5,8].
Cleavage of the arginine residue of Atg8 by Atg4 (hereafter, 'cleavage') is required for Atg8-PE production, which is in turn necessary for targeting of Atg8 to autophagy-related structures. The requirement for Atg8 cleavage can be bypassed by cells expressing Atg8 G116 instead of full-length Atg8. Atg8 G116 cells can produce Atg8-PE in the absence of Atg4 cleavage activity, but autophagy is still defective [5,10,11]. Thus, cleavage of Atg8-PE by Atg4 (hereafter, 'delipidation') is also important for autophagy. However, it remains unknown whether the defect in delipidation affects targeting of Atg8 to the VICS or subsequent IM expansion.
In this study, we bypassed the cleavage reaction using Atg8 G116 cells and visualized the IM as a cup-shaped structure under the fluorescence microscope by overexpressing precursor Ape1, a selective cargo of autophagosomes [2]. This analysis showed that the delipidation reaction did not affect Atg8 targeting to the VICS and that IM expansion was severely impaired in delipidation mutants. We conclude that delipidation of Atg8-PE is important for efficient IM expansion.
Gene disruptions of ATG2, ATG4, ATG11, and ATG17 were performed by homologous recombination. For disruption of ATG2, the DNA fragment was PCR-amplified using plasmid pFA6a-hphNT1 as template and Atg2_del_Univ_F and Atg2_del_Univ_R as primers. For ATG4 disruption, genomic DNA obtained from ORY0804 was used as a template, and APG04N-500F and APG04C+600R were used as primers. For disruption of ATG11 and ATG17, genomic DNA obtained from GYS891 was used as a template and ATG11N-800F/ ATG11C+500R and APG17N-600F/APG17C+600R as primers, respectively.

Alkaline phosphatase (ALP) assay
The ALP assay was performed as described previously [14]. Cells were grown to mid-log phase and shifted to SD(-N) medium (0.17% Difco™ yeast nitrogen base w/o amino acids and ammonium sulfate, 2% glucose) to induce autophagy. Lysates were prepared by disrupting the cells with glass beads in the ice-cold ALP assay buffer (250 mM Tris-HCl (pH9.0), 10 mM MgSO 4 , 10 μM ZnSO 4 , 1 mM phenylmethylsulfonyl fluoride) and the cell debris was removed by centrifugation at 2,000 × g for 5 min. ALP activity in the lysate was assayed with 5.5 mM 1-naphthyl phosphate as a substrate for 10 min at 30˚C, and the reaction was stopped by addition of the ALP stop buffer (2 M glycine-NaOH (pH 11.0)). The fluorescence intensity (excitation: 360nm, emission: 465nm) was measured using an RF-5300PC spectrofluorophotometer (Shimadzu).

Fluorescence microscopy
Cells were cultured in SDCA medium to a density of about 5×10 7 cells/ml, and autophagy was induced by addition of rapamycin. Cells were harvested, spun at RT with a microcentrifuge, and subjected to fluorescence microscopy using an IX83 inverted system microscope (Olympus) equipped with a UPlanSApo100×/1.40 Oil (Olympus) and a CoolSNAP HQ CCD camera (Nippon Roper). A U-FGFP and U-FMCHE filter sets (Olympus) were used for GFP/mNeon-Green and octadecyl rhodamine B (R18, Invitrogen)/FM4-64 staining visualization, respectively. Images were acquired using the MetaVue imaging software (Molecular Devices). For determination of IM length, fluorescence intensities of mNeonGreen-labeled Atg proteins were measured using the 'linescan' function of the MetaView software, and the full width at half maximum was calculated manually.
For R18 staining, cells were stained with 10 μg/ml of R18 (1 mg/ml stock dissolved in dimethyl sulfoxide) for 10 min in nutrient-rich medium at 30˚C. After cells were washed three times with fresh medium, rapamycin was added to induce autophagy. FM 4-64 staining was performed as previously described [16].

Results
Localization of Atg8 to autophagy-related structures requires Atg8 cleavage but not Atg8-PE delipidation Cleavage of Atg8 by Atg4 is a prerequisite for Atg8-PE formation, which is in turn essential for localization of Atg8 to autophagy-related structures; thus, autophagosome formation is abolished in atg4Δ cells [8,17]. When Atg8 lacking the C-terminal arginine residue (Atg8 G116 ) is expressed in atg4Δ cells, cleavage of Atg8 can be bypassed; Atg8-PE is produced in the absence of Atg4, but autophagosome formation is still defective in this mutant [5]. We examined the autophagic activity of Atg8 G116 atg4Δ cells by performing Ape1 maturation, GFP-Atg8 cleavage, and alkaline phosphatase assays. The results revealed that these cells are defective in autophagy (Fig 1).
It remains controversial whether delipidation of Atg8-PE is required for targeting of Atg8 to autophagy-related structures [5,10,11]. To address this issue, we first examined localization of Atg8 and Atg8 G116 by fluorescence microscopy. In GFP-Atg8 expressing wild-type cells, GFP-Atg8 was visualized as puncta close to the vacuole, which corresponded to autophagyrelated structures (Fig 2A). By contrast, the abundance of puncta was markedly reduced in atg4Δ cells (Fig 2A); this observation was supported by counting the number of puncta per cell (Fig 2B). This result is reasonable because lipidation of GFP-Atg8 is inhibited in the absence of Atg4. Next, we examined the localization of GFP-Atg8 G116 . GFP-Atg8 G116 was detected as puncta in wild-type and atg4Δ cells (Fig 2A). Quantification of the puncta revealed that the number of puncta was significantly smaller in atg4Δ cells than in wild-type cells ( Fig  2B). This decrease is probably caused by lacking the localization of GFP-Atg8 G116 to at least autophagosomes in atg4Δ cells because autophagosome formation is severely defective in Atg8 G116 atg4Δ cells [5]. Moreover, GFP-Atg8 G116 exhibited a vacuolar membrane pattern in atg4Δ cells, as previously reported (Fig 2A, arrows) [5].
Autophagy-related structures disappear in atg11Δ atg17Δ cells because Atg11 and Atg17 are scaffold proteins playing a role in recruitment of Atg proteins [17,18]. To determine whether the puncta observed in GFP-Atg8 G116 -expressing cells represented autophagy-related GYS608) cells carrying an empty or Atg4-expressing plasmid were grown in SDCA medium to mid-log phase and treated with rapamycin for 2 h. Western blot analysis was performed using anti-Atg8 and anti-Ape1 structures, we disrupted ATG11 and ATG17 genes in these strains. In ATG4 cells expressing GFP-Atg8 G116 , the number of puncta decreased in atg11Δ, atg17Δ, and atg11Δ atg17Δ cells (Fig 3). In atg4Δ cells expressing GFP-Atg8 G116 , the number of puncta decreased upon disruption of ATG11, ATG17, or both (Fig 3A). Quantification revealed that the number of the puncta was at the basal level in atg11Δ, atg17Δ, and atg11Δ atg17Δ cells (Fig 3B). Therefore, the antisera. Slower-migrating bands in the upper panel correspond to Atg8/Atg8 G116 (Unlipidated Atg8), and faster-migrating bands correspond to Atg8-PE. Slower-migrating bands in the lower panel correspond to the precursor form of Ape1 (prApe1), and faster-migrating bands correspond to the mature Ape1 (mApe1). (B) ATG8 atg4Δ (KVY13) or ATG8 G116 atg4Δ (GYS608) cells carrying the indicated plasmids were grown in SDCA medium to mid-log phase and treated with rapamycin for 2 h. Western blot analysis was performed with anti-GFP antibodies. Slower-migrating bands represent GFP-tagged Atg8 (GFP-Atg8), and faster-migrating bands represent cleaved GFP. (C) ATG8 G116 (YOC5308) and ATG8 G116 atg4Δ (YOC5309) cells were grown in SDCA medium to mid-log phase and shifted to SD(-N) medium and incubated for 4 h at 30˚C. Then autophagic activity was measured by the alkaline phosphatase assay. Error bars indicate standard deviations. N.S. indicates not significant. *P < 0.05, **P < 0.01 (two-tailed Student's t-test) (n = 3).

Atg8-PE delipidation is important for IM expansion
The number of GFP-Atg8 G116 dots observed in ATG4 cells markedly decreased in atg4Δ cells (Fig 2), and these dots corresponded to autophagy-related structures (Fig 3). Together, these observations support the fact that the autophagosome is hardly generated in Atg8 G116 atg4Δ cells. Next, we examined whether the IM expands in Atg8 G116 atg4Δ cells by the prApe1-overexpression system, which enables visualization of the IM as a cup-shaped structure [2]. In this experiment, Atg8 and Atg8 G116 were visualized by fusing green fluorescent mNeonGreen to their N-termini.
Cup-shaped IMs were visualized in wild-type cells expressing mNeonGreen-Atg8, but no autophagy-related structures were observed in atg4Δ cells due to their defect in cleavage ( Fig  4A). mNeonGreen-Atg8 G116 was visualized as a cup-shaped structure in wild-type cells, but as dots in atg4Δ and atg2Δ cells (Fig 4A). We observed no significant difference in IM lengths between wild-type cells expressing mNeonGreen-Atg8 and those expressing mNeonGreen-Atg8 G116 (Fig 4B), indicating that cleavage by Atg4 does not have a large impact on IM lengths. The average IM lengths in mNeonGreen-Atg8 G116 expressing wild-type, atg4Δ, and atg2Δ cells were 0.83 μm, 0.43 μm, and 0.41 μm, respectively (Fig 4B), indicating that the IM lengths in atg4Δ cells were significantly shorter than those in wild-type cells. In atg2Δ cells, Atg8 G116 can localize to the VICS, but the IM does not expand [2,17]. We detected no significant differences in IM lengths between Atg8 G116 expressing atg4Δ and atg2Δ cells (Fig 4B). Thus, IM expansion was reduced to the basal level without delipidation.
In mammalian cells, the IM can expand without lipidation of Atg8 or Atg8 itself [19][20][21]. Therefore, we examined the IMs labeled with other marker proteins in Atg8 G116 ATG4 and Atg8 G116 atg4Δ cells. Previously, our group have shown that Atg1 and Atg16 are localized to the IM [2]. Therefore, we examined localization of Atg1 and Atg16 in Atg8 G116 ATG4 or Atg8 G116 atg4Δ cells overexpressing prApe1. In Atg8 G116 ATG4 cells, Atg1 and Atg16 were visualized as cup-shaped structures, whereas they are observed as dots in Atg8 G116 atg4Δ cells (Fig 5A). The IM lengths of the structures in Atg8 G116 atg4Δ cells were significantly shorter than those in Atg8 G116 ATG4 cells (Fig 5B). These results show that the IMs labeled with Atg1 and Atg16 cannot fully expand without delipidation. Taken together, we conclude that delipidation of Atg8-PE is important for IM expansion.

Atg8-PE delipidation plays an important role in IM expansion other than supplying unlipidated Atg8
Most Atg8 G116 expressed in atg4Δ cells was detected as Atg8-PE (Fig 6, lane 2) [5]. We hypothesized that the shortage of unlipidated Atg8 G116 caused a defect in IM expansion in Atg8 G116expressing atg4Δ cells as shown in Fig 4. To address this issue, we overexpressed Atg8 or Atg8 G116 in Atg8 G116 background cells. In Atg8 G116 ATG4 cells, overexpressed Atg8 and Atg8 G116 were mostly detected as an unlipidated form, and Atg8-PE was detected as minor bands (Fig 6, lanes 3 and 4). Overexpression of Atg8 and Atg8 G116 in Atg8 G116 ATG4 cells did not affect the maturation of Ape1 (Fig 6, lanes 3 and 4). When Atg8 was overexpressed in Atg8 G116 atg4Δ cells, most Atg8 was detected as an unlipidated form (Fig 6, lane 5). On the  other hand, Atg8-PE as well as unlipidated Atg8 was detected in Atg8 G116 atg4Δ cells overexpressing Atg8 G116 (Fig 6, lane 6). Nevertheless, the maturation of Ape1 was severely defective in Atg8 G116 atg4Δ cells overexpressing Atg8 G116 (Fig 6, lane 6). These results indicate that only the presence of unlipidated Atg8 G116 in addition to Atg8-PE is insufficient to recover autophagic activity in Atg8 G116 atg4Δ cells, suggesting that the shortage of unlipidated Atg8 G116 is not the main cause of a defect in autophagy in these cells.
Next, we measured IM lengths in Atg8 G116 atg4Δ cells overexpressing Atg8 or Atg8 G116 . The IM lengths in both strains were significantly longer than those in Atg8 G116 atg4Δ cells but significantly shorter than those in Atg8 G116 ATG4 cells (Fig 4). This result suggests that the defect in IM expansion in Atg8 G116 atg4Δ cells cannot be fully recovered by supplying unlipidated Atg8 G116 . Taken together, we conclude that delipidation of Atg8 itself plays an important role in efficient IM expansion.

Autophagic membranes are present in cells defective in Atg8-PE delipidation, but not in atg2Δ cells
We tested several kinds of lipophilic fluorescent dyes and found that octadecyl rhodamine B (R18) preferentially stained the ER labeled with Dpm1-GFP, an ER transmembrane marker, with slight staining of the vacuolar membrane, under nutrient-rich conditions (Fig 7A). In R18-stained cells, GFP-Atg8 was visualized as a dot close to the vacuole without rapamycin treatment (Fig 7B). A cup-shaped IM emerged after treatment with rapamycin, and the IM was labeled with R18 (Fig 7B), suggesting that lipids constituting the IM can be labeled by R18 staining. Because the vacuolar membrane was slightly stained with R18, we also explored the possibility that the IM is stained with FM 4-64, a lipophilic dye that labels the vacuolar membrane. The vacuolar membrane was stained with FM 4-64 and treated with rapamycin. In contrast to R18, the IM was not stained with FM 4-64 ( Fig 7C). Therefore, R18 preferentially labels IM lipids, whereas FM 4-64 does not.
Cup-shaped IMs were stained with R18 in ATG4 cells expressing mNeonGreen-Atg8 or mNeonGreen-Atg8 G116 (Fig 7D). In mNeonGreen-Atg8 G116 expressing atg4Δ cells, mNeon-Green-labeled structures were also stained with R18, and there were no significant differences in frequency of R18-labeling relative to wild-type cells (Fig 7D and 7E). On the other hand, mNeonGreen-labeled structures in atg2Δ cells barely stained with R18 (Fig 7D and 7E). These results suggest that mNeonGreen-Atg8 G116 expressing atg4Δ cells are capable of recruiting lipids to the IM, whereas atg2Δ cells are not.

Discussion
In this study, we demonstrated that delipidation of Atg8-PE by Atg4 is dispensable for targeting of Atg8-PE to the VICS, but required for expansion of the IM (Fig 4). We also obtained a Autophagic activity of Atg8 G116 atg4Δ cells overexpressing Atg8/Atg8 G116 . ATG8 G116 atg4Δ (GYS608) cells expressing indicated proteins were grown in SDCA medium to mid-log phase, and then treated with rapamycin for 2 h. Western blot analysis was performed with anti-Atg8 and anti-Ape1 antisera. Short-and long-exposed images were shown. Slower-migrating bands in the upper and middle panels correspond to unlipidated Atg8, and faster-migrating bands correspond to Atg8-PE. Slower-migrating bands in the lower panel correspond to the precursor form of Ape1 (prApe1), and faster-migrating bands correspond to the mature Ape1 (mApe1). ox indicates overexpression. https://doi.org/10.1371/journal.pone.0181047.g006 Atg4 is important for efficient expansion of isolation membranes by cleaving lipidated Atg8 in yeast result suggesting that biological membranes exist at the Atg8-labeled structures in delipidation-defective cells (Fig 7). From these facts, we think that Atg4 is involved in efficient expansion of the IM by cleaving Atg8-PE localized at the VICS.
Here we show that IM expansion is impaired in Atg8 G116 atg4Δ cells at the same level as in Atg8 G116 atg2Δ cells (Fig 4). On the other hand, previous studies showed that a small number of closed autophagosomes are formed in Atg8 G116 atg4Δ cells, whereas no autophagosomes are detected in atg2Δ cells [5,22,23]. These facts suggest that the IM in Atg8 G116 atg4Δ cells has an ability to become a closed autophagosome, whereas that of Atg8 G116 atg2Δ cells does not. Thus, Atg8-PE delipidation by Atg4 is unlikely to play a role in closure of the IM; instead, the delipidation reaction is mainly involved in IM expansion. We also found that autophagyrelated structures in Atg8 G116 atg4Δ cells were stained with lipophilic dye R18, whereas those in Atg8 G116 atg2Δ cells were not (Fig 7D and 7E). Based on these observations, we hypothesize that Atg2 is involved in recruitment of lipids to the VICS, and that subsequent delipidation of Atg8-PE by Atg4 serves to expand the IM. The minimal autophagic activity in Atg8 G116 atg4Δ cells might be explained by the presence of lipids in the autophagy-related structures.
In ATG4 cells, Atg8-PE is produced, and GFP-Atg8 localizes to autophagy-related structures (Figs 1A and 2A), suggesting that Atg8 localized to autophagy-related structures is conjugated to PE. On the other hand, when we visualized the IM, we demonstrated that the IM length of Atg8 G116 atg4Δ cells did not significantly differ from that in Atg8 G116 atg2Δ cells (Fig  4), indicating that Atg8-PE delipidation is required for IM expansion. Therefore, we assume that delipidation activity of Atg4 is initially repressed at the VICS, but derepressed upon IM expansion.
Previous reports proposed that delipidation supplies Atg8 G116 to promote autophagosome formation [5,11]. However, overexpression of Atg8 G116 cannot rescue the defect in autophagic activity in atg4Δ atg8Δ cells [10,11]. We examined the level of unlipidated Atg8 G116 in Atg8 G116 -overexpressing Atg8 G116 atg4Δ cells and detected a certain amount of unlipidated Atg8 G116 (Fig 6). However, IM expansion and autophagic activity were not rescued in these cells (Figs 4 and 6), suggesting that the role of Atg8-PE delipidation is not limited to supplying unlipidated Atg8 G116 . Based on these results, we propose three possible scenarios: (1) a cycle of lipidation and delipidation is involved in IM expansion; (2) because accumulation of Atg8-PE at the VICS may disturb IM expansion, surplus Atg8-PE must be delipidated by Atg4; or (3) the site of delipidation occurs is important, i.e., Atg8-PE must be delipidated at a specific membrane, and then Atg8 G116 plays a role at the site of delipidation. Notably, IM lengths slightly increased by overexpression of Atg8 or Atg8 G116 in Atg8 G116 atg4Δ cells (Fig 4), whereas autophagic activity remained unchanged (Fig 6, lanes 2, 5 and 6). These facts suggest that excess unlipidated Atg8/Atg8 G116 affects the shape of the Atg8-labeled structures but does not improve the activity of autophagosome formation. observed by fluorescence microscopy. (B) Wild-type cells harboring pRS314[GFP-Atg8] and pYEX-BX [prApe1] plasmids were grown in SDCA medium containing CuSO 4 and stained with R18. After the cells were washed with fresh medium, they were treated with rapamycin for 3 h. (C) Wild-type cells harboring pRS314 [GFP-Atg8] and pYEX-BX[prApe1] plasmids were grown in SDCA medium containing CuSO 4 and stained with FM 4-64. After the cells were incubated with fresh medium without dye for 30 min, they were treated with rapamycin for 3 h. (D) mNG-Atg8 atg4Δ (YOC5272) and mNG-Atg8 G116 atg4Δ (YOC5330) cells carrying an empty or Atg4-expressing plasmid and mNG-Atg8 G116 atg4Δ atg2Δ (YOC5331) cells carrying an Atg4-expressing plasmid were grown to mid-log phase in SDCA medium containing CuSO 4 , and then stained with R18. After the cells were washed with fresh medium, they were treated with rapamycin for 1 h. Scale bar: 2 μm. (E) Frequencies of mNG-positive structures stained with R18. Error bars indicate standard deviations. N.S., not significant. **P < 0.01 (two-tailed Student's t-test). At least 40 cells were counted for each experiment (n = 4). https://doi.org/10.1371/journal.pone.0181047.g007 Lipidation of Atg8 mediates the tethering and hemifusion of liposomes in vitro [24]. Based on this fact, Atg8-PE is thought to be involved in IM expansion. Here, we report that delipidation of Atg8-PE is also essential for IM expansion, implying that Atg8 G116 plays a role in this step. Addition of delipidated Atg8 to the in vitro Atg8-PE conjugation reaction may modulate the hemifusion activity of Atg8-PE.
The ER marker proteins GFP-HDEL and Dpm1-GFP do not label the IM [2], indicating that these proteins do not transit to the IM from the ER. However, we found that a lipophilic dye R18 labeled the IM (Fig 7B). This result could be interpreted in two ways. First, the dye could be transported to the IM via canonical vesicular trafficking pathways from the ER. Alternatively, lipids could be supplied from the ER to the IM directly. We prefer the latter 'lipid flow' model because we have never observed any vesicle-like structures around the IM, despite intensive observation by electron microscopy (data not shown). Future work should seek to clarify the mechanisms involved in transport of lipids from the ER to the IM. It is worth noting that we cannot exclude the possibility that other organelles supply lipids to the IM.
Further studies of Atg4 will be necessary to reveal the mechanisms of IM expansion mediated by delipidation of Atg8-PE.