Atg8L/Apg8L is the fourth mammalian modifier of mammalian Atg8 conjugation mediated by human Atg4B, Atg7 and Atg3 Isei Tanida, Yu-shin Sou, Naoko Minematsu-Ikeguchi, Takashi Ueno and Eiki Kominami Molecular Cell Biology, Department of Biochemistry, Juntendo University School of Medicine, Tokyo, Japan Ubiquitylation and ubiquitylation-like reactions are post-translational modifications that play indispensable roles in many cellular events. Atg8 ⁄ Apg8 ⁄ Aut7 is a ubiquitin-like (Ubl) protein essential for autophagy in the yeast Saccharomyces cerevisiae [1]. Genetic analyses of yeast ATG gene mutants have suggested that the C-terminus of Atg8 is cleaved by Atg4 or a protease activated by Atg4 ⁄ Apg4 to expose the C-terminal Gly of Atg8 [2–4]. Subsequently, Atg8 is activated by Atg7 ⁄ Apg7 ⁄ Gsa7 ⁄ Cvt2, an E1-like enzyme [5–8], transferred to Atg3 ⁄ Apg3 ⁄ Aut1, an E2-like enzyme [3], and finally conjugated to phosphatidylethanolamine [3]. This conjugation reaction is essential for auto- phagy under conditions of starvation and in the cyto- plasm for the vacuole-targeting (Cvt) pathway under nutrient-rich conditions. To date, three Atg8 homologs have been character- ized in mammals: LC3 (microtubule-associated protein 1 light chain 3, MAP1-LC3) [9,10], 4-aminobutyrate A - receptor associated protein (GABARAP) [11–13], and Golgi-associated ATPase enhancer of 16 kDa (GATE- 16) [12–14]. Following cleavage by human Atg4B ⁄ autophagin 1, the C-terminal Gly of each of these Atg8 homologs is exposed, as is also observed for yeast Atg8 [13,15]. Human Atg4A ⁄ autophagin 2 also cleaves the C-terminus of GATE-16 [16]. Following C-terminal cleavage, each Atg8 homolog is activated by human Atg7, an E1-like enzyme, to form a transient E1-sub- strate intermediate [12,17,18]. Each Atg8 homolog is subsequently transferred to human Atg3, an E2-like enzyme, to form a transient E2-substrate intermediate, and modified to the membrane-bound forms, LC3-II, GABARAP-PL (GABARAP-II), and GATE-16-II [12,19]. Recently, it has been shown that LC3-II and GABARAP-PL are protein–phospholipid conjugates [15], with phosphatidylethanolamine thought to be the Keywords autophagy; GABARAP; GATE-16; LC3; ubiquitylation-like modification Correspondence E. Kominami, Department of Biochemistry, Juntendo University School of Medicine, 2-1-1, Hongo, Bunkyo-ku, Tokyo 113-8421, Japan Fax: +81 3 5802 5889 Tel: +81 3 5802 1031 E-mail: kominami@med.juntendo.ac.jp (Received 23 December 2005, revised 4 April 2006, accepted 5 April 2006) doi:10.1111/j.1742-4658.2006.05260.x Murine Atg8L ⁄ Apg8L has significant homology with the other known mammalian Atg8 homologs, LC3, GABARAP and GATE-16. However, it is unclear whether murine Atg8L modification is mediated by human Atg4B, Atg7 and Atg3. Expression of Atg8L in HEK293 cells led to clea- vage of its C-terminus. In vitro, the C-terminus of Atg8L was cleaved by human Atg4B, but not human Atg4A or Atg4C. Atg8L-I formed an E1-substrate intermediate with Atg7 C572S , and an E2-substrate intermediate with Atg3 C264S . A modified form of Atg8L was detected in the pelletable fraction in the presence of lysosomal protease inhibitors under nutrient-rich conditions. Cyan fluorescent protein (CFP)–Atg8L colocalized with yellow fluorescent protein (YFP)–LC3 in HeLa cells in the presence of the inhibi- tors. However, little accumulation of the modified form of Atg8L was observed under conditions of starvation. These results indicate that Atg8L is the fourth modifier of mammalian Atg8 conjugation. Abbreviations CFP, cyan fluorescent protein; GABARAP, 4-aminobutyrate A receptor-associated protein; GABARAP-PL, GABARAP–phospholipid conjugate; GFP, green fluorescent protein; LC3, human microtubule-associated protein 1 light chain 3; LC3-I, soluble unmodified form of LC3; LC3-II, LC3–phospholipid conjugate; PL, phospholipid; PVDF, poly(vinylidene difluoride); TRX, thioredoxin; YFP, yellow fluorescent protein. FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 2553 phospholipid bound to LC3 [13,20]. LC3-II and GABARAP-PL are both deconjugated by human Atg4B [15]. It is unclear whether the target of GATE- 16 is a phospholipid. Recently, a fourth mammalian Atg8 homolog has been reported, mouse Atg8L ⁄ Apg8L [21]. The amino acid sequence of mouse Atg8L ⁄ Apg8L shows 100% and 54% identity to those of human Atg8L and yeast Atg8, respectively, and 33%, 60% and 86% identity to the amino acid sequences of LC3, GATE-16 and GABARAP, respectively. In addition, Atg8L shares a conserved Gly at its C-terminus. Following chemical modification of Atg8L with a C-terminal vinyl sulfone, mouse Atg8L–vinyl sulfone was shown to react with Atg4B, suggesting that mouse Atg8L is a substrate of Atg4B [21]. However, no direct evidence that mouse Atg8L is cleaved by Atg4B has yet been reported. Fur- thermore, both the homology between Atg8L and the three other mammalian Atg8 homologs and the con- served C-terminal Gly in all four proteins suggest that Atg8L may also be a substrate of mammalian Atg7 and Atg3. Therefore, we examined whether Atg8L is a substrate of these three enzymes involved in mamma- lian Atg8 conjugation. Results The C-terminus of Atg8L is cleaved in HEK293 cells, and the Gly116 of Atg8L is essential for cleavage The C-termini of yeast Atg8, mammalian LC3, GABA- RAP and GATE-16 are post-translationally cleaved to expose a Gly residue, which is essential for ubiquitin-like modification [3,10,12]. This consensus Gly116 is con- served in Atg8 and its mammalian homologs [21], and is present in Atg8L. To determine whether the C-terminus of Atg8L is post-translationally cleaved, we constructed an Atg8L expression vector tagged with a Myc epitope at its N-terminus and a 3xFLAG epitope at its C-termi- nus (Fig. 1A, Myc–Atg8Lwt)3xFLAG). Following transfection of HEK293 cells with this construct, the lysates were analyzed by SDS ⁄ PAGE. A protein of 18 kDa corresponding to Myc–Atg8L, a C-terminal cleaved form of Myc–Atg8Lwt)3xFLAG, was recog- nized by immunoblotting with anti-Myc, but not with anti-FLAG (Fig. 1B). We next investigated whether the Gly116 residue of Atg8L is essential for cleavage of its C-terminus by changing Gly116 of Myc–Atg8Lwt)3xFLAG to Ala by site-directed mutagenesis (Fig. 1A, Myc–Atg8L- GA)3xFLAG), and expressing the mutant protein in HEK293 cells (Fig. 1B, GA). The mobility of mutant Myc–Atg8LGA)3xFLAG on SDS ⁄ PAGE was slower than that of wild-type Myc–Atg8Lwt)3xFLAG. More- over, the mutant protein was recognized by immuno- blotting with both anti-Myc and anti-FLAG (Fig. 1B). Similar results were obtained when the Gly116 residue was deleted by site-directed mutagenesis (Fig. 1A, Myc–Atg8LDG)3xFLAG; Fig. 1B, DG). These results suggest that the C-terminus of Atg8L is post-transla- tionally cleaved in HEK293 cells, and that Gly116 of Atg8L is essential for the cleavage of its C-terminus. The cleaved form of Atg8L was designated Atg8L-I. Atg4B cleaves the C-terminus of Atg8L in vitro The three previously identified mammalian Atg8 homo- logs, LC3, GABARAP, and GATE-16, were shown to be cleaved by human Atg4B (hAtg4B) in vitro [13,15]. Although we showed that the C-terminus of Atg8L was cleaved soon after its translation in HEK293 cells, the enzyme responsible for this activity could not be identi- fied. Therefore, we examined whether Atg4B has pro- teolytic activity on the C-terminus of Atg8L. FLAG– hAtg4B was expressed in HEK293 cells, the cells were lysed, and the lysate was fractionated by ultracentrifuga- tion, with the resulting supernatant used as the enzyme mixture. FLAG–hAtg4B in the supernatant was recog- nized by immunoblotting with anti-FLAG (Fig. 1C, FLAG–hAtg4B). The substrate, wild-type thioredoxin (TRX)–Atg8L)3xFLAG, consisting of Atg8L with TRX at its N-terminus and the 3xFLAG epitope at its C-terminus, was expressed in Escherichia coli, and the supernatant of this cell lysate was used as the substrate solution. When we incubated the two supernatants con- taining FLAG–hAtg4B with TRX–Atg8L)3xFLAG, we found that the C-terminus of Atg8L was cleaved, as shown by immunoblotting with anti-FLAG (Fig. 1C). Using anti-TRX, we confirmed that the N-terminal TRX tag within each substrate remains unchanged, indicating that the C-terminal FLAG tag was cleaved by hAtg4B (Fig. 1C). When an active site mutant of hAtg4B, hAtg4B C74A [15], was used instead of wild-type hAtg4B, little cleavage occurred (Fig. 1D, lane 2 versus lane 1). When a mutant in which the Gly116 residue of Atg8L had been deleted, TRX–Atg8LDG)3xFLAG, was employed instead of the wild type, little C-terminal cleavage occurred (Fig. 1D, lane 5 versus lane 1). In addition to hAtg4B, hAtg4A ⁄ hApg4A ⁄ autophag- in-2, hAtg4C ⁄ hAutl1 ⁄ autophagin-3 and hAtg4D ⁄ autophagin-4 have also been reported [16,22]. Of these three Atg4 homologs, hAtg4A has been shown to cleave the C-terminus of GATE-16 [16], and autophag- in-3 ⁄ hAtg4C ⁄ hAutl1, but not hAtg4D, has been shown to exert N-ethylmaleimide-sensitive proteolytic activity Ubiquitylation-like modification of murine Atg8L I. Tanida et al. 2554 FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS on the synthetic substrate Mca-Thr-Phe-Gly-Met-Dpa- NH 2 [22]. Therefore, hAtg4A and hAtg4C may also cleave the C-terminus of Atg8L. When hAtg4A or hAtg4C was used instead of hAtg4B for in vitro diges- tion, little cleavage of the C-terminal 3xFLAG tag of TRX–Atg8L)3xFLAG occurred (Fig. 1, lanes 3 and 4 versus lane 1). These results indicate that Atg4B cleaves the C-terminus of Atg8L in vitro. Atg8L-I forms an E1-substrate intermediate with human Atg7 C572S In the ubiquitlyation-like modification steps, an active site Cys within an E1 enzyme temporally conjugates to a substrate to form an E1-substrate intermediate via a thiol–ester bond. Owing to the rapid turnover of an E1 reaction, it is difficult to recognize such an interme- diate in sufficient quantity. When an active site Cys residue of human Atg7 is changed to Ser, a stable O-ester bond instead of a thiol–ester bond will be formed between the enzyme and substrate(s). Previ- ously, we showed that the mammalian Atg8 homologs LC3, GATE-16 and GABARAP form E1-substrate intermediates with an active site mutant human Atg7 C572S , in which the active site Cys572 of human Atg7 was changed to Ser12. If Atg8L-I is a substrate of human Atg7, Atg8L-I will form an E1-substrate intermediate with human Atg7 C572S . Therefore, we investigated whether Atg8L-I also forms an intermedi- ate with human Atg7 C572S . When Myc–Atg8Lwt)3x- FLAG and Atg7 C572S were expressed together, an Atg7 C572S –Myc–Atg8L (E1-substrate) intermediate was observed (Fig. 2). This intermediate was also detected by immunoblotting with anti-Myc, but not with anti- Fig. 1. The C-terminus of Atg8L is cleaved in vivo and in vitro.(A) Schematic representation of mutant proteins of Myc–Atg8L)3x- FLAG. The arrowhead indicates a Gly residue predicted to be essential for ubiquitylation-like reactions. Myc–Atg8Lwt)3xFLAG represents wild-type Atg8L protein tagged with the Myc epitope at its N-terminus and with the 3xFLAG epitope at the C-terminus. Myc–Atg8LGA)3xFLAG represents a mutant protein in which the Gly116 of Atg8L was changed to Ala, and Myc–Atg8LDG)3xFLAG represents a mutant protein in which the Gly116 of Atg8L was deleted. (B) Cleavage of the C-terminus of Atg8L. Wild-type or mutant Myc–Atg8L)3xFLAG proteins were transiently expressed in HEK293 cells. The cells were lysed, total proteins were separated by SDS ⁄ PAGE, and wild-type and mutant Myc–Atg8L)3xFLAG were recognized by immunoblotting with anti-Myc (a-Myc) and anti- FLAG (a-FLAG). wt, GA, and DG were the same as in (A). (C) In vit- ro assay for Atg4B cleavage of the C-terminus of Atg8L. Wild-type TRX–Atg8L)3xFLAG was expressed in Escherichia coli; the cells were lysed and centrifuged, and the supernatants were used as the substrate. The supernatants of HEK293 cells transfected with pTag2B–hATG4B (hAtg4B wild) were used as enzymes; the negat- ive control consisted of the supernatants of HEK293 cells transfect- ed with pCMV–Tag2B (control). Following incubation of 1 lgof enzyme solution with 10 lg of substrate solution for the indicated times (incubation time), the reactions were stopped, and the total proteins in each mixture were separated by SDS ⁄ PAGE. TRX– Atg8L)3xFLAG was recognized by immunoblotting with anti-thiore- doxin (TRX), and FLAG–hAtg4B and the C-terminal 3-FLAG tag of TRX–Atg8L)3xFLAG were recognized with anti-FLAG. TRX– Atg8L)3xFLAG, uncleaved form of TRX–Atg8L)3xFLAG; TRX– Atg8L-I, cleaved form of TRX–Atg8L)3xFLAG; FLAG–hAtg4B, FLAG-tagged Atg4B cysteine protease. (D) In vitro assay for the cleavage of the C-terminus of Atg8L by an active site mutant hAtg4 C74A or other Atg4 homologs. As an enzyme solution, supern- atants of HEK293 cells expressing an active site mutant hAtg4B C74A (hAtg4B C74A), wild-type hAtg4A (hAtg4A wild) or wild-type hAtg4C (hAtg4C) were employed instead of wild-type hAtg4B (hAtg4B wild). Supernatant containing mutant TRX– Atg8LDG)3xFLAG (DG), in which the Gly116 of Atg8L was deleted, was used as the substrate. The enzyme solution and the substrate solution were mixed and incubated for 30 min. A B C D I. Tanida et al. Ubiquitylation-like modification of murine Atg8L FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 2555 FLAG. These results indicate that, like the other Atg8 homologs, Atg8L-I forms an E1-substrate intermediate with human Atg7 C572S . Atg8L-I forms an E2-substrate intermediate with green fluorescent protein (GFP)–Atg3 C264S ; this is dependent on Atg7 Owing to the rapid turnover of the E2 reaction, it is difficult to recognize such an E2-substrate intermediate with an E2-like enzyme. When the active site Cys resi- due of the E2-like enzyme is replaced by Ser, a stable O-ester bond instead of a thiol–ester bond will be formed between the enzyme and substrate(s); this is dependent on an E1-like enzyme. Previously, we showed that LC3, GABARAP and GATE-16 form E2-substrate intermediates with an active site mutant human Atg3 C264S and that this activity was dependent on human Atg7 [19]. If Atg8-L is a substrate of Atg3, Atg8L-I will form an E2-substrate intermediate with human Atg3 C264S that is dependent on Atg7. There- fore, we investigated whether Atg8L-I also forms an Atg7-dependent E2-substrate intermediate with human Atg3 and whether formation of this intermediate occurs via the Gly116 in Atg8L. When wild-type Myc–Atg8Lwt)3xFLAG, GFP–Atg3 C264S and wild- type Atg7 were expressed in HEK293 cells, a GFP– Atg3 C264S –Myc–Atg8L (E2-substrate) intermediate was formed (Fig. 3, lane 2). When mutant Myc– Atg8LDG)3xFLAG was substituted for wild-type Atg8L (Fig. 3, lane 3), no intermediate was observed, as demonstrated by immunoblotting with anti-GFP. When Atg7 was not overexpressed (endogenous Atg7 alone), no intermediate was observed, while endo- genous Atg7 was detected by anti-Atg7 (Fig. 3, lane Fig. 2. Atg8L-I forms an E1-substrate intermediate with human Atg7 C572S . Myc-tagged Atg8L)3xFLAG (Myc–Atg8L)3xFLAG) was transiently expressed with wild-type (Atg7 wt) or mutant human Atg7 (Atg7 C572S). The cells were lysed, the proteins were separ- ated by SDS ⁄ PAGE, and human Atg7 was recognized by immuno- blotting with anti-human Atg7 (WB:a-Atg7), while Myc–Atg8L-I was recognized by immunoblotting with anti-Myc (WB:a-Myc). Atg7– Atg8L-I intermediate, E1-substrate intermediate between human Atg7 C572S and Atg8L; Atg7, human Atg7; Myc–Atg8L-I, Myc-tagged Atg8L-I. Fig. 3. Atg8L-I forms an E2-substrate intermediate with human Atg3 C264S , which is dependent on human Atg7. Myc–Atg8Lwt)3x- FLAG (wt) was transiently expressed together with green fluores- cent protein (GFP)–Atg3 C264S in the presence (Atg7+) or absence (Atg7–) of human Atg7. DG represents mutant Myc–Atg8LDG)3x- FLAG lacking the Gly116 of Atg8L (see Fig. 1A). After lysing the cells, total proteins were separated by SDS ⁄ PAGE. Atg7 was recognized by immunoblotting with anti-human Atg7 (WB:a-Atg7), GFP–Atg3 C264S and its E2-substrate intermediate were recognized with anti-GFP (WB:a-GFP), and Myc-tagged Atg8L proteins were recognized by immunoblotting with anti-Myc (WB:a-Myc). The faint band in lane 3 (DG) of the middle panel (WB:a-GFP) is GFP– Atg3 C264S -endogenous Atg8 homolog intermediate(s). Ubiquitylation-like modification of murine Atg8L I. Tanida et al. 2556 FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 1). These results indicate that Atg8L-I forms an Atg7- dependent E2-substrate intermediate with human Atg3 via its Gly116. Atg8L-II is increased in the presence of E64d and pepstatin A, inhibitors of lysosomal proteases, in HeLa cells LC3-I and GABARAP-I have been shown to be modi- fied by Atg7 and Atg3 to form the respective protein– phospholipid conjugates, LC3-II and GABARAP-PL [15]. We have reported the accumulation of LC3-II and GABARAP-PL in HeLa cells incubated with E64d [23], a membrane-permeable inhibitor of cathep- sins B, H, and L, and pepstatin A [24], an inhibitor of cathepsins D and E, for 24 h under nutrient-rich con- ditions [15,25]. These results suggest that a modified form of Atg8L-I may accumulate under the same con- ditions. When HeLa cells expressing Myc–Atg8L were incubated with E64d and pepstatin A for 24 h under nutrient-rich conditions, LC3-II and GABARAP-PL accumulated (Fig. 4A, WB:a-LC3 and a-GABARAP). Under these conditions, we observed two bands by immunoblotting with anti-Myc (Fig. 4A, WB:a-Myc). One band corresponded to Myc–Atg8L-I, while the other showed faster mobility. This second band, repre- senting a modified form of Atg8L, was designated Atg8L-II. We have also shown that the modified forms of LC3 and GABARAP, LC3-II and GABARAP-PL, are present in the pellet after subcellular fraction- ation [15]. Therefore, we hypothesized that, if Atg8L-II is a modified form and not a degradation product of Atg8L-I, it would be present in the pellet of inhibitor-treated HeLa cells. Therefore, we centri- fuged cell lysates at 100 000 g for 1 h and examined the subcellular localization of Myc–Atg8L-I and Myc–Atg8L-II by immunoblotting with anti-Myc. Our results indicated that Myc–Atg8L-II was present in the pellet, whereas Myc–Atg8L-I was present in the supernatant (Fig. 4B), suggesting that Myc– Atg8L-II is a modified form of Myc–Atg8L-I, and not a degradation product. We next investigated the intracellular localization of Atg8L-II under these conditions using cyan fluorescent protein (CFP)–Atg8L. HeLa cells expressing both CFP–Atg8L and yellow fluorescent protein (YFP)– LC3 were cultured under nutrient-rich conditions in the presence of E64d and pepstatin A for 24 h, and punctate signals of both CFP–Atg8L (Fig. 4C,e,h,k) and YFP–LC3 (Fig. 4C,d,g,j) were observed by fluor- escent microscopy. Like Myc–Atg8L, CFP–Atg8L can form E1-substrate and E2-substrate intermediates with human Atg7 C572S and human Atg3 C264S , respectively (Fig. 4D,E), and little degradation of CFP–Atg8L was observed even in the presence of these inhibitors for 24 h under nutrient-rich conditions (Fig. 4F), indica- ting that the fluorescence signal reflects intact tagged protein. Merging of the images indicated that most of the puncta of CFP–Atg8L were colocalized with those of YFP–LC3 (Fig. 4,f,I,l). In the absence of inhibitors, only a few puncta of either type were observed (Fig. 4C,a–c). Little Atg8L-II accumulates under conditions of starvation even in the presence of E64d and pepstatin A in HeLa cells LC3-I is significantly lipidated to form LC3-II under conditions of starvation [10], and LC3-II is degraded in the lysosome [25]. Therefore, in the presence of E64d and pepstatin A, LC3-II shows significant accu- mulation under conditions of starvation [25]. It is poss- ible that, like LC3, Atg8L is modified to Atg8L-II under conditions of starvation. Therefore, we investi- gated whether Atg8L-II accumulates in HeLa cells expressing Myc–Atg8L)3xFLAG incubated in Krebs– Ringer buffered medium for 4 h under conditions of starvation in the presence of E64d and pepstatin A, conditions under which LC3-II has been reported to accumulate [25] (Fig. 5, LC3-II). However, little Atg8L-II accumulation was observed (Fig. 5, lane 4 versus lane 3). Even when cells were incubated with the inhibitors under nutrient-rich conditions for a short time—4 h compared with 24 h (Fig. 4A)—little accumulation of Atg8L-II occurred (Fig. 5, lane 2). Discussion Here, we have shown that Atg8L is a substrate of reac- tions mediated by human Atg4B, Atg7, and Atg3. We found that the C-terminus of Atg8L is post-transla- tionally cleaved and that the Gly116 in Atg8L is essen- tial for this reaction, which is mediated by human Atg4B, a cysteine protease, in vitro. We also found that Atg8L forms an E1-substrate intermediate with an E1-like enzyme, the active site mutant Atg7 C572S , and an E2-substrate intermediate with an E2-like enzyme, the active site mutant Atg3 C264S , with the latter reac- tion dependent on Atg7. All these reactions are similar to a series of reactions of three other mammalian Atg8 homologs: LC3, GABARAP, and GATE-16. In addi- tion, we showed that a modified form of Atg8L, Atg8L-II, accumulates in HeLa cells in the presence of lysosomal protease inhibitors under nutrient-rich con- ditions, comparable to the accumulation of LC3-II I. Tanida et al. Ubiquitylation-like modification of murine Atg8L FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 2557 A B C abc def ghi jkl Fig. 4. Atg8L-II, a modified form of Atg8L, accumulates in HeLa cells in the presence of inhibitors of lysosomal proteases, E64d and pepstatin A. (A) Accumulation of Atg8L-II in the presence of E64d and pepstatin A. Myc–Atg8L)3xFLAG was transiently expressed in HeLa cells, and the transfectants were incubated for 24 h in the presence (+) or absence (–) of E64d and pepstatin A. The cells were lysed, and total proteins were separated by SDS ⁄ PAGE. Endogenous LC3 and GABARAP were recognized by immunoblotting with anti- LC3 and 4-aminobutyrate A -receptor associated protein (GABARAP), respectively (WB:a-LC3 and a-GABARAP). Myc–Atg8L was recognized with anti-Myc (WB:a-Myc). (B) Subcellular localization of Myc–Atg8L-II. Cell lysates of inhibitor-treated HeLa cells (A, E64d and pepstatin A+) were fractionated into pellet (Ppt) and soluble (Sup) fractions by ultracentrifugation at 100 000 g for 1 h [12,15]. LC3-I, soluble unlip- idated form of LC3; LC3-II, lipidated membrane-bound form of LC3; GABARAP-I, unlipidated form of GABARAP; GABARAP-PL, lipidated form of GABARAP; Myc–Atg8L-I, soluble form of Myc–Atg8L; Myc–Atg8L-II, membrane-bound form of Myc–Atg8L. (C) Intracellular local- ization of cyan fluorescent protein (CFP)–Atg8L. HeLa cells expressing both CFP–Atg8L and yellow fluorescent protein (YFP)–LC3 were cultured in the absence (dimethylsulfoxide (DMSO)) or presence of E64d and pepstatin A (E64d and pepstatin A) for 24 h. After fixing, cyan and yellow fluorescence in HeLa cells were observed with a Zeiss Axioplan2 fluorescence microscope with filters XF114-2 and XF104-2. The deconvoluted images are shown in (C). In a, d, g, and j, yellow fluorescent images correspond to YFP–LC3. In b, e, h, and k, cyan fluorescent images correspond to CFP–Atg8L. Merged images (Merge) are shown in e, f, i, and l. Arrowheads indicate colocaliz- ation of YFP–LC3 and CFP–Atg8L. (D) Formation of an E1-substrate intermediate with Atg7 C572S and CFP–Atg8L. CFP–Atg8L was trans- iently expressed with wild-type (Atg7 wt) or mutant human Atg7 (Atg7 C572S). The cells were lysed, the proteins were separated by SDS ⁄ PAGE, and human Atg7 was recognized by immunoblotting with anti-human Atg7 (WB:a-Atg7), and CFP–Atg8L was recognized by immunoblotting with anti-GFP (WB:a-GFP). Atg7–Atg8L intermediate, E1-substrate intermediate between human Atg7 C572S and Atg8L; Atg7, human Atg7; CFP–Atg8L, CFP-tagged Atg8L. (E) Formation of an E2-substrate intermediate with Atg3 C264S S and CFP–Atg8L. CFP– Atg8L was transiently expressed together with GFP–Atg3 C264S in the presence of human Atg7 (Atg7+). After lysing the cells, total pro- teins were separated by SDS ⁄ PAGE. Atg7 was recognized by immunoblotting with anti-human Atg7 (WB:a-Atg7), GFP–Atg3 C264S and its E2-substrate intermediate were recognized with anti-Atg3 (WB:a-Atg3), and CFP–Atg8L was recognized by immunoblotting with anti-Atg8L (WB:a-Atg8L). (F) There was little degradation of CFP–Atg8L in the presence of E64d and pepstatin A for 24 h under nutrient-rich condit- ions. CFP–Atg8L and YFP–LC3 were transiently expressed in HeLa cells, and the transfectant was incubated for 24 h in the presence (+) or absence (–) of E64d and pepstatin A. The cells were lysed, and total proteins were separated by SDS ⁄ PAGE. YFP–LC3 and CFP–Atg8L were recognized by immunoblotting with anti-GFP (WB:a-GFP). CFP–Atg8L was recognized with anti-Atg8L (WB:a-Atg8L). Positions of molecular weight markers for SDS ⁄ PAGE are indicated on the right of the panel. Ubiquitylation-like modification of murine Atg8L I. Tanida et al. 2558 FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS and GABARAP-PL under identical conditions. More- over, we found that Atg8L-II was fractionated in the pellet. These results indicate that Atg8L is a substrate of Atg4B, Atg7, and Atg3, and that Atg8L is the authentic fourth modifier in the mammalian Atg8 con- jugation system. Previously, we showed that LC3-II and GABARAP-PL are protein–phospholipid conju- gates in vivo [15], and that these two conjugates, as well as GATE-16-II, are localized to a membrane compartment [10,12]. Like LC3-II and GABARAP- PL, Atg8L-II was fractionated in the pelletable fraction, suggesting that Atg8L-II may be a protein–phospholipid conjugate that is localized to a membrane compartment. The modified form, Myc–Atg8L-II, was observed only in the presence of E64d and pepstatin A for 24 h under nutrient-rich conditions (Fig. 4A), whereas little Myc–Atg8L was observed for a shorter period (4 h) even in the presence of these inhibitors. In contrast, LC3-II accumulates in the presence of E64 and pepstatin A within only 4 h under both nutrient-rich and starva- tion conditions. There are two possible reasons for the low level of Myc–Atg8L-II accumulation for 4 h in the presence of these inhibitors. The first is that modifica- DE F Fig. 4. (Continued). I. Tanida et al. Ubiquitylation-like modification of murine Atg8L FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 2559 tion of Atg8L-I to Atg8L-II may occur much more slowly than that of LC3-I to LC3-II. Therefore, Myc– Atg8L-II was observed only after a longer incubation time (24 h) in the presence of E64d and pepstatin A. The other possibility is that Myc–Atg8L-II may be very unstable compared with LC3-II. We have reported that Atg4B is a delipidating enzyme for LC3-II and GABA- RAP-PL in addition to being a cysteine protease that cleaves the C-termini of mammalian Atg8 homologs. Therefore, if Atg4B delipidates Myc–Atg8L-II more effectively than LC3-II, there will be less accumulation of Myc–Atg8L-II compared with LC3-II. These points will be clarified in future studies by in vitro assays of lipidation and delipidation of Atg8L. Experimental procedures Materials, and biochemical and molecular biological techniques Molecular biological and biochemical techniques were per- formed as described [25]. To clone mouse Atg8L cDNA by PCR, we used high-fidelity KOD plus DNA polymerase (Toyobo, Osaka, Japan). A plasmid containing mouse Atg8L cDNA was transfected into HEK293 cells with Lipofectamine 2000 reagent according to the manufac- turer’s protocol (Invitrogen, Carlsbad, CA). E. coli strain DH5a cells, the hosts for plasmids and protein expression, were grown in Luria broth medium in the presence of anti- biotics as required. Restriction enzymes were purchased from Toyobo and New England Biolabs (Beverly, MA). Oligonucleotides were synthesized by Invitrogen. pGEM-T was purchased from Promega (Madison, WI), and p3xFLAG–CMV14 was obtained from Sigma-Aldrich (St Louis, MO). Cloning of mouse Atg8L cDNA and site-directed mutagenesis A DNA fragment encoding the entire open reading frame of ATG8L, according to a cDNA sequence in GenBank (accession number: BG244294) [21], was amplified by high- fidelity PCR from a Marathon ready mouse brain cDNA library (BD Biosciences Clontech, Palo Alto, CA) and inserted into pGEM-T. The resulting plasmid was designa- ted pGEM–mATG8L. Using the Gene-Editor in vitro site-directed mutagenesis system (Promega) and the plasmid pGEM–mATG8L, Gly116 in mouse Atg8L was replaced by Ala in accordance with the manufacturer’s protocol, and the resulting plasmid was designated pGEM–mATG8L G116A . Gly116 in mouse Atg8L of pGEM–mATG8L was deleted by site-directed mutagenesis, and the resulting plasmid was designated pGEM–mATG8LDG. Mammalian expression vectors for N-terminal Myc-tagged and C-terminal 3xFLAG-tagged versions of Atg8L, Myc–Atg8Lwt)3xFLAG, Myc–Atg8L- GA)3xFLAG and Myc–Atg8LDG)3xFLAG were con- structed from pGEM–mATG8L and p3xFLAG–CMV14 by high-fidelity PCR, and designated pMyc–mATG8L) 3xFLAG, pMyc–mATG8LGA)3xFLAG, and pMyc– mATG8LDG)3xFLAG, respectively. An expression vector for N-terminal TRX-tagged and C-terminal 3xFLAG- tagged Atg8L in E. coli, TRX–Atg8L)3xFLAG, was con- structed based on pThioHisA (Invitrogen) and designated pTRX–ATG8L)3xFLAG. A mammalian expression vector for N-terminal CFP-tagged Atg8L designated pCFP– ATG8L was generated by inserting the open reading frame of mAtg8L excised from pGEM–mATG8L into pCFP-C1 (BD Biosciences Clontech). The DNA sequences of new constructs were confirmed using a ABI PRISM TM 310 Gen- etic Analyzer (Applied Biosystems, Foster City, CA). The expression vectors pCMV–hAPG7, pCMV–hAPG7CS, pGFP–hAPG3, pGFP–hAPG3CS, pTag2B–HsATG4B, pTag2B–HsATG4B C74A , pTag2B–HsATG4A and pTag2B– HsAUTL1 for human Atg7, mutant Atg7 C572S , GFP–Atg3, mutant GFP–Atg3 C264S , wild-type FLAG–hAtg4B, mutant FLAG–hAtg4B C74A , wild-type FLAG–hAtg4A and wild- Fig. 5. There was little modification of Atg8L under conditions of starvation, even in the presence of inhibitors of lysosomal proteases. Myc–Atg8L)3xFLAG was transiently expressed in HeLa cells. The transfectant was transferred to Krebs–Ringer buffered medium (KRB, 4 h, starvation) or Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (DMEM, 10% fetal bovine serum, nutrient-rich), and incubated for 4 h in the presence (+) or absence (–) of inhibitors. Little Myc–Atg8L-II was detected by immunoblotting with anti-Myc under either condition. Endogenous LC3 was recognized by immunoblotting with antibodies to LC3. As a control, the membrane was stained with Coomassie Brilliant Blue. LC3-II, LC3–phospholipid conjugate. Ubiquitylation-like modification of murine Atg8L I. Tanida et al. 2560 FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS type FLAG–hAtg4C ⁄ AutL1, respectively, were as described [15,17,19]. Cell culture The HEK293 and HeLa cell lines were purchased from the ATCC (Manassas, VA) and cultured in 60-mm dishes in Dulbeccos’s modified Eagles’s medium (DMEM, Invitro- gen) containing 10% fetal bovine serum (Invitrogen) in a humidified 5% CO 2 atmosphere at 37 °C. Antibodies For preparation of antiserum against mouse Atg8L, rabbits were immunized with a TRX–Atg8L fusion protein. Anti- human GABARAP, LC3, Atg7 and Atg3 were prepared and purified as described [12,17,19,26]. Anti-Myc was pur- chased from Cell Signaling (Beverly, MA), anti-FLAG M2 was purchased from Sigma-Aldrich, anti-TRX was pur- chased from Santa Cruz Biotechnology (Santa Cruz, CA), and anti-GFP was purchased from BD Biosciences Clontech. Immunoblotting analyses Protein concentrations were determined using the bicincho- ninic acid protein assay reagent (Pierce, Rockford, IL). Immunoblotting analyses were carried out according to standard protocols using a chemiluminescent method with SuperSignal West Dura Extended Duration Substrate or SuperSignal West Pico Chemiluminescent Substrate (Pierce). In vitro assay for cleavage of the C-terminus of TRX–Atg8L)3xFLAG by Atg4B The in vitro assay for cleavage of the C-terminus of mouse Atg8L was performed as previously described [15]. Fluorescence microscopy HeLa cells expressing YFP–hLC3 and CFP–mAtg8L were fixed according to the manufacturer’s protocol (BD Bio- sciences Clontech). Briefly, after transfection of pYFP–LC3 and pCFP–ATG8L into HeLa cells using Lipofecta- mine2000 (Invitrogen), cells were incubated for 24 h under nutrient-rich conditions in the presence or absence of E64d and pepstatin A. Thereafter, cells were washed twice with NaCl ⁄ P i , and fixed in NaCl ⁄ P i containing 4% paraformal- dehyde for 10 min at room temperature. Fixed cells were washed three times with NaCl ⁄ P i , and then mounted on slide-glasses with SlowFade Light antifade reagent (50% glycerol solution) (Invitrogen). Fluorescence of YFP and CFP in the cells was monitored with a Zeiss Axioplan2 fluorescence microscope (Carl Zeiss, Thornwood, NY) equipped with an ORCA-ER CCD camera (Hamamatsu Photonics, Tokyo, Japan). For deconvolution of the ima- ges, a Zeiss Axioplan2 fluorescence microscope (Carl Zeiss) and an Aqua C-imaging system (Hamamatsu Photonics) were employed. Acknowledgements This work was supported in part by grants-in-aid 15590254 (to IT), 09680629 (to TU) and 12470040 (to EK) for Scientific Research, and grant-in-aid 12146205 (to EK) for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports, and Cul- ture of Japan, and The Science Research Promotion Fund from the Japan Private School Promotion Foun- dation (to EK). References 1 Ohsumi Y (2001) Molecular dissection of autophagy: two ubiquitin-like systems. Nat Rev Mol Cell Biol 2, 211–216. 2 Tsukada M & Ohsumi Y (1993) Isolation and character- ization of autophagy-defective mutants of Saccharo- myces cerevisiae. FEBS Lett 333, 169–174. 3 Ichimura Y, Kirisako T, Takao T, Satomi Y, Shimonishi Y, Ishihara N, Mizushima N, Tanida I, Kominami E, Ohsumi M, et al. (2000) A ubiquitin-like system mediates protein lipidation. Nature 408, 488–492. 4 Kirisako T, Ichimura Y, Okada H, Kabeya Y, Mizushi- ma N, Yoshimori T, Ohsumi M, Takao T, Noda T & Ohsumi Y (2000) The reversible modification regulates the membrane-binding state of Apg8 ⁄ Aut7 essential for autophagy and the cytoplasm to vacuole targeting path- way. J Cell Biol 151, 263–276. 5 Mizushima N, Noda T, Yoshimori T, Tanaka Y, Ishii T, George MD, Klionsky DJ, Ohsumi M & Ohsumi Y (1998) A protein conjugation system essential for auto- phagy. Nature 395, 395–398. 6 Kim J, Dalton VM, Eggerton KP, Scott SV & Klionsky DJ (1999) Apg7p ⁄ Cvt2p is required for the cytoplasm- to-vacuole targeting, macroautophagy, and peroxisome degradation pathways. Mol Biol Cell 10, 1337–1351. 7 Yuan W, Stromhaug PE & Dunn WA Jr (1999) Glu- cose-induced autophagy of peroxisomes in Pichia pas- toris requires a unique E1-like protein. Mol Biol Cell 10, 1353–1366. 8 Tanida I, Mizushima N, Kiyooka M, Ohsumi M, Ueno T, Ohsumi Y & Kominami E (1999) Apg7p ⁄ Cvt2p: a novel protein-activating enzyme essential for autophagy. Mol Biol Cell 10, 1367–1379. 9 Mann SS & Hammarback JA (1994) Molecular characterization of light chain 3. A microtubule binding subunit of MAP1A and MAP1B. J Biol Chem 269, 11492–11497. I. Tanida et al. Ubiquitylation-like modification of murine Atg8L FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS 2561 10 Kabeya Y, Mizushima N, Ueno T, Yamamoto A, Kiri- sako T, Noda T, Kominami E, Ohsumi Y & Yoshimori T (2000) LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing. EMBO J 19, 5720–5728. 11 Wang H, Bedford FK, Brandon NJ, Moss SJ & Olsen RW (1999) GABA A -receptor-associated protein links GABA A receptors and the cytoskeleton. Nature 397, 69–72. 12 Tanida I, Komatsu M, Ueno T & Kominami E (2003) GATE-16 and GABARAP are authentic modifiers mediated by Apg7 and Apg3. Biochem Biophys Res Commun 300, 637–644. 13 Kabeya Y, Mizushima N, Yamamoto A, Oshitani- Okamoto S, Ohsumi Y & Yoshimori T (2004) LC3, GABARAP and GATE16 localize to autophagosomal membrane depending on form-II formation. J Cell Sci 117, 2805–2812. 14 Sagiv Y, Legesse-Miller A, Porat A & Elazar Z (2000) GATE-16, a membrane transport modulator, interacts with NSF and the Golgi v-SNARE GOS-28. EMBO J 19, 1494–1504. 15 Tanida I, Sou YS, Ezaki J, Minematsu-Ikeguchi N, Ueno T & Kominami E (2004) HsAtg4B ⁄ HsApg4B ⁄ autophagin-1 cleaves the carboxyl termini of three human Atg8 homologues and delipidates microtubule- associated protein light chain 3- and GABAA receptor- associated protein–phospholipid conjugates. J Biol Chem 279, 36268–36276. 16 Scherz-Shouval R, Sagiv Y, Shorer H & Elazar Z (2003) The COOH terminus of GATE-16, an intra-Golgi trans- port modulator, is cleaved by the human cysteine pro- tease HsApg4A. J Biol Chem 278, 14053–14058. 17 Tanida I, Tanida-Miyake E, Ueno T & Kominami E (2001) The human homolog of Saccharomyces cerevisiae Apg7p is a protein-activating enzyme for multiple sub- strates including human Apg12p, GATE-16, GABARAP, and MAP-LC3. J Biol Chem 276, 1701–1706. 18 Tanida I, Tanida-Miyake E, Nishitani T, Komatsu M, Yamazaki H, Ueno T & Kominami E (2002) Murine Apg12p has a substrate preference for murine Apg7p over three Apg8p homologs. Biochem Biophys Res Commun 292, 256–262. 19 Tanida I, Tanida-Miyake E, Komatsu M, Ueno T & Kominami E (2002) Human Apg3p ⁄ Aut1p homologue is an authentic E2 enzyme for multiple substrates, GATE-16, GABARAP, and MAP-LC3, and facilitates the conjugation of hApg12p to hApg5p. J Biol Chem 277, 13739–13744. 20 Bampton E, Goemans C, Dhevahi Niranjan D, Mizu- shima N & Tolkovsky A (2005) The dynamics of auto- phagy visualised in live cells: from autophagosome formation to fusion with endo ⁄ lysosomes. Autophagy 1, 23–36. 21 Hemelaar J, Lelyveld VS, Kessler BM & Ploegh HL (2003) A single protease, Apg4B, is specific for the autophagy-related ubiquitin-like proteins GATE-16, MAP1-LC3, GABARAP, and Apg8L. J Biol Chem 278, 51841–51850. 22 Marino G, Uria JA, Puente XS, Quesada V, Bordallo J & Lopez-Otin C (2003) Human autophagins, a family of cysteine proteinases potentially implicated in cell degradation by autophagy. J Biol Chem 278, 3671–3678. 23 Ueno T, Ishidoh K, Mineki R, Tanida I, Murayama K, Kadowaki M & Kominami E (1999) Autolysosomal membrane-associated betaine homocysteine methyltrans- ferase. Limited degradation fragment of a sequestered cytosolic enzyme monitoring autophagy. J Biol Chem 274, 15222–15229. 24 Dean RT (1977) Lysosomes and protein degradation. Acta Biol Med Ger 36, 1815–1820. 25 Tanida I, Minematsu-Ikeguchi N, Ueno T & Kominami E (2005) Lysosomal turnover, but not a cellular level, of endogenous LC3 is a marker for autophagy. Autophagy 1, 84–91. 26 Asanuma K, Tanida I, Shirato I, Ueno T, Takahara H, Nishitani T, Kominami E & Tomino Y (2003) MAP- LC3, a promising autophagosomal marker, is processed during the differentiation and recovery of podocytes from PAN nephrosis. FASEB J 17, 1165–1167. Ubiquitylation-like modification of murine Atg8L I. Tanida et al. 2562 FEBS Journal 273 (2006) 2553–2562 ª 2006 The Authors Journal compilation ª 2006 FEBS . Atg8L/Apg8L is the fourth mammalian modifier of mammalian Atg8 conjugation mediated by human Atg4B, Atg7 and Atg3 Isei Tanida, Yu-shin Sou, Naoko Minematsu-Ikeguchi, Takashi Ueno and Eiki. it is unclear whether murine Atg8L modification is mediated by human Atg4B, Atg7 and Atg3. Expression of Atg8L in HEK293 cells led to clea- vage of its C-terminus. In vitro, the C-terminus of Atg8L. of Atg8L-II occurred (Fig. 5, lane 2). Discussion Here, we have shown that Atg8L is a substrate of reac- tions mediated by human Atg4B, Atg7, and Atg3. We found that the C-terminus of Atg8L is