ERdj8 governs the size of autophagosomes during the formation process

In macroautophagy, an intracellular degradation pathway, a double-membrane organelle called the autophagosome engulfs a variety of targets and delivers them to the lysosome/vacuole for degradation (Yang and Klionsky, 2010; Galluzzi et al., 2017). In the case of nonselective macroautophagy, autophagosomes are relatively homogenous in size (Mizushima, 2004). In selective autophagy, autophagosome size is rather heterogeneous. In the Cvt pathway in yeast, which enwraps the super complex of aminopeptidase I in an autophagy-related structure, the diameter is smaller (∼150 nm; Baba et al., 1997). By contrast, in the case of bacteria-targeted autophagy, larger autophagosomes (5–10 µm) can form (Singh et al., 2006; Yamaguchi et al., 2009). These observations raise two important questions: How is the size of selective autophagosomes determined, and why are nonselective autophagosomes mostly uniform in size? One study proposed that autophagosome size is determined by the total expression level of Atg8 (Xie et al., 2008). Another group proposed that actin assembly inside the autophagosome determines its shape (Mi et al., 2015). Despite these advances, we still do not fully understand how autophagosome size is determined.

The ER plays an important role as a platform for autophagosome formation (Axe et al., 2008; Ylä-Anttila et al., 2009; Hayashi-Nishino et al., 2009). ERdj proteins, a family of ER-localized DnaJ-like proteins, contain a J-domain that binds binding immunoglobulin protein (BiP/GRP-78) and promotes its ATPase activity (Kampinga and Craig, 2010; Otero et al., 2010). We have shown that some of the ERdjs are involved in protein folding, ER-associated protein degradation, or Ca²⁺ homeostasis (Yamamoto et al., 2010; Ushioda et al., 2016). In this study, we characterized ERdj8, the novel eighth member of the ERdj family that affects the size of the autophagosome membrane.

ERdj8/DNAJC16 is a type 1 membrane protein that contains a single transmembrane stretch along with DnaJ and thioredoxin-like domains (Fig. 1 A and data not shown). It was distributed within the ER network along with the ER luminal marker DsRed-KDEL and formed prominent puncta (Fig. S1 A, white arrows). Observation using a specific antibody against ERdj8 and structured illumination microscopy revealed that these punctate ERdj8-positive structures exhibited a dense meshwork morphology (Fig. 1 B). Importantly, these structures slightly excluded, but were still positive for, the typical ER marker protein GFP-Sec61β, indicating they were connected to the ER network and constituted a part of it (Fig. 1 B). These data suggest the existence of a specialized subdomain of the ER where ERdj8 primarily resides.

An ER subdomain enriched in the glycerophospholipid biosynthesis enzyme phosphatidylinositol synthase (PIS) exists (Kim et al., 2011; English and Voeltz, 2013). ERdj8-positive puncta were overlapped with PIS-GFP or closely adjacent (∼24%; Fig. 1 C). As this PIS-enriched domain is associated with autophagic machinery (Nishimura et al., 2017), we investigated their relationship. ATG13 forms a complex with ULK1, FIP200, and ATG101 that acts as a scaffold for ATG proteins (Mizushima et al., 2011; Karanasios et al., 2013), and 74% of the ATG13-positive structures were associated with endogenous ERdj8 (Fig. 1 D). The autophagy-specific class III PI3-kinase complex, which contains ATG14, is recruited to the ER, where it generates PI3P (Matsunaga et al., 2010; Itakura and Mizushima, 2010). 41% of ATG14 signals were associated with ERdj8 upon induction of autophagy (Fig. 1 E). Moreover, a part of ERdj8 puncta colocalized with FIP200 or ATG14 on the ER (Fig. S1, C and E). Together, these results led us to hypothesize that ERdj8 is associated with autophagosome formation.

We then examined the effect of ERdj8 overexpression by observing LC3, a marker of autophagosomes (Fig. 2 A, bottom panels; Kabeya et al., 2000). The size of LC3-positive structures was significantly increased by ERdj8 overexpression (Fig. 2, A and B) but not other members of the ERdj family (not shown). We noticed that some of the GFP-LC3–positive structures became elongated when ERdj8 was overexpressed (Fig. S1 F, bottom panels). Hence, we measured the distance to the distal tips of GFP-LC3–positive structures after refining the images through a deconvolution process. This analysis revealed that the population of larger GFP-LC3–positive structures (>0.4 µm) was significantly increased by ERdj8 overexpression (Fig. S1 F, red arrowheads). Electron microscopic analysis also revealed that overexpression of ERdj8 increased the size of autophagosomes (Fig. 2, C and D). Unlike wild-type ERdj8, point mutants in the ERdj8 DnaJ domain (H57Q) or TRX domain (C174A, C177A) did not increase the size of GFP-LC3–positive structures when overexpressed (Fig. 2, E and F).

We then examined the effect of ERdj8 overexpression on the isolation membrane, the precursor of the autophagosome. ATG5 associates with the isolation membrane, but upon completion of autophagosome formation, it detaches from the membrane (Mizushima et al., 2001). In COS-7 cells stably expressing YFP-ATG5, overexpression of ERdj8 doubled the lifetime of YFP-ATG5–puncta signal relative to control cells (Fig. 2 G). Syntaxin17 (STX17) is mostly associated with completed autophagosomes, and some GFP-LC3 puncta were colocalized with mCherry-STX17 (Fig. S1 D, upper panels, white arrows; Tsuboyama et al., 2016; Kumar et al., 2018; Kumar et al., 2019). ERdj8-BFP overexpression diminished the colocalization, implying that autophagosome formation was less efficient (Fig. 2 H). Therefore, the overall effect on autophagy flux by overexpression of ERdj8 was marginal (Fig. S1 G). Collectively, these observations indicate that ERdj8 overexpression delayed the transition from isolation membranes/phagophores to autophagosomes.

On the other hand, ERdj8 knockdown led to a marked increase in smaller LC3-positive structures (<0.4 µm in diameter) under starvation (Fig. 3 A, red arrowheads; and Fig. 3 B). Indeed, autophagosomes, which are labeled with GFP-LC3 by correlative light electron microscopy (CLEM), were significantly decreased in size by ERdj8 knockdown (Fig. 3, C and D). GFP-ATG14–positive puncta were also smaller (Fig. 3, E and F), and the lifetime of ATG5-positive puncta was reduced (Fig. 3 G), although autophagy flux was scarcely affected (Fig. S1 H). These data indicate that ERdj8 ablation led to formation of less-expanded autophagosomes.

To confirm this observation, we investigated whether ERdj8 knockdown affected the size capacity of autophagic targets by monitoring clearance of large protein aggregates (>2 µm in diameter) induced by mild puromycin treatment (Kirkin et al., 2009). As shown in Fig. 4, A and B, ERdj8 knockdown decreased the efficiency of clearance of large p62/SQSTM–positive protein aggregates relative to control.

To further support this notion, we employed latex beads coated with transfection reagents, which are introduced into cells via the endocytic pathway and are ultimately engulfed by autophagosomes; this process mimics the xenophagy pathway, which targets invading bacteria (Kobayashi et al., 2010). We used the mRFP-enhanced GFP (eGFP)-Galectin8 system to monitor engulfment of latex beads by autophagosomes (Fig. S2 A; Fujita et al., 2013; Maejima et al., 2013). Galectin8, a kind of lectin, is expressed in the cytoplasm and targeted to exposed carbohydrate chains within the damaged endosomal luminal side, which contains latex beads, and is eventually engulfed by the autophagosome, leading to fusion with the lysosome (Thurston et al., 2012). In the low-pH environment of the lysosome, the eGFP signal is attenuated due to its high pKa (acidity constant) of the protein (∼6.0; Kneen et al., 1998). Therefore, if mRFP-eGFP-Galectin8 surrounding the beads exhibits a weakened eGFP signal, it would indicate that the autophagic process had successfully completed in the lysosome. Two kinds of latex beads with different sizes (1 and 3 µm in diameter) were used for this assay. First, the 1-µm beads were introduced into HeLa. In Atg16L1-knockout cells, the GFP signal surrounding mRFP-positive beads was significantly stronger than in wild-type HeLa (control siRNA), indicating that bead autophagy was defective (Fig. S2, B and C). In contrast, eGFP signals in ERdj8-knockdown cells were attenuated relative to control, indicating successful autophagic engulfment (Fig. S2, B and C). However, when 3-µm beads were introduced, attenuation of eGFP-positive signals was suppressed in ERdj8-knockdown cells (Fig. 4, C and D). These data indicate that ERdj8 knockdown resulted in a defect in enwrapping of larger targets, but it preserved the cell’s ability to enwrap smaller ones.

Next, we explored the role of ERdj8 in the engulfment of physiologically relevant autophagy targets: damaged mitochondria (Youle and Narendra, 2011). Mitochondrial autophagy (mitophagy) occurs in response to loss of mitochondrial membrane potential following treatment with the uncoupler Carbonyl cyanide m-chlorophenyl hydrazone (Ding and Yin, 2012). After CCCP treatment, ∼80% of endogenous ERdj8-positive puncta were associated with Tomm20, a mitochondrial protein (Fig. S2 D, white arrows). CCCP treatment attenuated the signal from the mitochondria inner membrane protein cytochrome c oxidase subunit II (Cox-II), but this attenuation was suppressed in ERdj8-knockdown cells (Fig. 5, A–C). The survival rate was reduced after CCCP treatment of ERdj8-knockdown cells (data not shown). Thus, mitophagy does not proceed normally in ERdj8 knockdown cells.

Finally, we explored the role of ERdj8 at the organismal level in Caenorhabditis elegans. As with mammalian ERdj8 knockdown, knockdown of dnj-8 increased the abundance of smaller GFP::LGG-1 puncta (<1 µm in diameter); LGG-1 is an Atg8/LC3 paralog (Fig. S2, E and F; Meléndez and Levine, 2009). Next, we examined the effect of dnj-8 knockdown on mitophagy in C. elegans. The signal intensity and protein level of mitochondrially targeted GFP (GFP::Mito) in the body wall muscle was increased by dnj-8 knockdown as in atg-5 knockdown, while the mRNA level was unchanged (Fig. 5, D and E; and Fig. S2, G and H). In the fertilized egg of C. elegans, sperm-derived paternal mitochondria are selectively degraded by mitophagy during early embryogenesis (Sato and Sato, 2011; Al Rawi et al., 2011). The paternal mitochondria are spherical and have an average diameter of 400–500 nm, significantly smaller than the materially derived somatic mitochondria (Zhou et al., 2016). Hence, we monitored the fate of the paternal GFP mitochondria. At the one-cell stage, GFP::LGG-1 signals surrounded sperm-derived paternal mitochondria in both mock and dnj-8 (RNAi) embryos (Fig. 5 F). At the 32- to 64-cell stages, sperm-derived paternal mitochondria disappeared in both mock and dnj-8 (RNAi) embryos but accumulated in atg-5 (RNAi) embryos (Fig. 5 G). Together, these data indicate that ERdj8 is responsible for enwrapping larger mitochondria but is dispensable for enwrapping smaller mitochondria, even at the organismal level.

In this study, we showed that a novel membrane protein ERdj8 is localized mainly in an ER subdomain involved in autophagosome formation. Overexpression resulted in enlargement of the autophagosome. In contrast, knockdown limited the target size of autophagic engulfment. However, the overall effect on autophagy flux by overexpression and/or knockdown of ERdj8 was marginal. On the basis of these findings, we propose that ERdj8 may play a critical role in fine-tuning the autophagosome formation process.

ERdj8 is concentrated in a subdomain of the ER that we named the ERdj8 domain. Because ERdj8 is a membrane-spanning protein, it must be laterally translocated from the conventional ER tubular network, and we can reasonably conclude that the ERdj8 domain is connected to the ER network. Sec61β, an abundant ER-resident membrane protein, is scarce in the ERdj8 domain, although not totally absent (Fig. 1 B). PIS partly overlapped with ERdj8, but mainly seemed to surround the periphery of the ERdj8 domain (Fig. 1 C). In light of a recent report that autophagosome formation is required for de novo phospholipid synthesis (Schütter et al., 2020; Andrejeva et al., 2019) and takes place adjacent to a PIS-enriched domain (Nishimura et al., 2017), colocalization with PIS provided us with an important clue about the function of ERdj8. Indeed, ATG13, ULK1, FIP200, and ATG14L were associated with the ERdj8 domain (Fig. 1, D and E; and Fig. S1, B, C, and E). ATG2, which bridges the isolation membrane and the ER, acts as a lipid transfer protein, possibly by transferring lipids to the isolation membrane (Gómez-Sánchez et al., 2018; Kotani et al., 2018; Osawa et al., 2019; Valverde et al., 2019). Because the ERdj8 domain associates with the phosphoglycerolipid synthesis enzyme PIS and possibly CEPT1, abundant phosphoglycerolipid should be generated in its vicinity. Furthermore, overexpression of ERdj8 enlarges the PIS1-containing ER subdomain (data not shown), which may facilitate lipid supply and autophagosome enlargement. It is tempting to speculate that ERdj8 regulates some of these processes. The DnaJ and TRX domains of ERdj8 on the ER luminal side were important for the function of ERdj8 in the regulation of autophagy (Fig. 2, E and F), and unidentified ERdj8 effector proteins and/or BiP may play important roles in this regulation. We are currently seeking to identify these proteins, with the goal of understanding the underlying mechanism.

In summary, our study revealed that ERdj8 plays some critical role in regulation of autophagosome formation, shedding light on the mechanism by which autophagosome size is determined, and these observations provide novel insight into the determination of autophagosome size.

All cell lines were cultured in DMEM high glucose (GIBCO) supplemented with 10% FBS (Invitrogen). For starvation treatment, cells were cultured in Earle’s balanced salt solution (EBSS; Sigma) for the indicated periods. Lipofectamine 2000 (Invitrogen, P/N 52887) was used for transfection.

The following cell lines were described previously: COS-7; HeLa stably expressing eGFP-LC3; HeLa stably expressing mRFP-eGFP-LC3 (Kageyama et al., 2011); HeLa stably expressing GFP-ATG5 (Fujita et al., 2008); HeLa stably expressing ULK1-GFP (Kageyama et al., 2011); HeLa stably expressing mCherry-Parkin (Narendra et al., 2008); HeLa stably expressing eGFP-DFCP1; and COS-7 stably expressing YFP-ATG5 (Hamasaki et al., 2013).

HeLa stably expressing mRFP-eGFP-Galectin8, Atg16-knockout HeLa cells stably expressing mRFP-eGFP-Galectin8, and COS-7 cells stably expressing PIS-GFP were constructed using the pMRX retroviral vector (Saitoh et al., 2003). For retrovirus preparation, plasmids mRFP-eGFP-Galectin8 and PIS-GFP were cloned and transiently transfected into Plat-E cells using Lipofectamine 2000. Transfection medium was removed 4 h after transfection, and fresh medium was added to the plate. Cell supernatants were collected at 36 h and filtered through a 0.45-µm filter. Cells to be transduced were seeded 24 h before infection and then transduced with virus-containing supernatant supplemented with Polybrene. Cells were left to recover for 24 h in growth media before puromycin selection (1 µg/ml; Wako, 160–23151).

Atg16L1 KO HeLa cells were generated as follows. CRISPR gRNA sequences designed against the ATG16L1 gene were cloned into px330 (Cong et al., 2013). The target sequence was 5′-GCG​CCG​CTG​ACT​TCC​CCC​GC-3′. HeLa cells were transfected with px330 encoding the gRNA. After 24 h, cells were diluted and seeded in 96-well plates for isolation of single clones. Clones with mutations in both alleles were identified by immunoblotting and confirmed by sequencing of genomic DNA.

Cells were lysed in 1% Triton-X100 lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 0.1% [vol/vol] PIC-2 [ISTI, A-0014-20]) for 30 min on ice. The suspension was sedimented by centrifugation at 20,400 g for 20 min at 4°C. Lysates were mixed with SDS loading buffer (200 mM Tris-HCl, pH 6.8, 5% β-mercaptoethanol, 20% SDS, and 40% glycerol), incubated at 65°C for 20 min, and subjected to SDS-PAGE and Western blotting according to standard protocols.

The following primary antibodies were obtained from the indicated suppliers: rabbit anti-DNAJC16 (ERdj8; Proteintech, 17599–1-AP; Western blotting [WB], 1/500); mouse anti-ATG13 (Merck, MABC46; immunocytochemistry [IC], 1/100), mouse anti–Cox-II (Novex, 459200; WB, IC, 1/1,000); mouse anti-GAPDH (HyTest, 5G4; WB, 1/1,000); mouse anti-Tomm20 (Abcam, ab56783; WB, 1/1,000); mouse anti-LC3 (MBL, M186-3; IC, 1/500; WB, 1/1,000); mouse anti-tubulin α (MBL, PM054; WB, 1/1,000); mouse anti-Tomm20 (Abcam, ab56783; IC, 1/1,000); rabbit anti-p62 (MBL, PM066; IC, 1/500); rabbit anti-ATG16L1 (MBL, PM040Y; WB, 1/1,000); rabbit anti-DDDDK (MBL, PM020; WB, IC, 1/1,000); rabbit anti-FIP200 (Proteintech; 17250–1-AP, IC, 1/100); mouse anti-Calnexin, C-terminal (Enzo, ADI-SPA-860; IC, 1/1,000); rabbit anti-Calnexin, N-terminal (MBL, PM060; IC, 1/1,000), and rabbit anti-DNAJC16_744 (IBL; IC, 1/100). Rabbit anti-DNAJC16_744 was raised against the ERdj8 peptide CGLGSRPIKGKLSKLSL (Immuno-Biological Laboratories Co., Ltd.; IC, 1/100).

Secondary antibodies were as follows: HRP-conjugated goat anti-rabbit IgG (H+L; Invitrogen, 656120; WB, 1/2,000); HRP-conjugated goat anti-mouse IgG (H+L; Invitrogen, 626520; WB, 1/2,000); Alexa Fluor 488 goat anti-mouse IgG (H+L; Invitrogen, A11029; IC, 1/1,000); Alexa Fluor 647 goat anti-mouse IgG (H+L; Invitrogen, M21235; IC, 1/1,000); and Alexa Fluor 488 goat anti-mouse IgG (H+L; Invitrogen, A11034; IC, 1/1,000).

siRNA duplexes targeting ERdj8 (no. 1: sense, 5′-CCU​GCA​AAU​UCU​GUC​UGC​GUU​GGA​U-3′, antisense, 5′-AUC​CAA​CGC​AGA​CAG​AAU​UUG​CAG​G-3′; no. 2: sense, 5′-GAU​CCU​GGA​GCA​GAA​GAC​AAG​UUC​A-3′, antisense, 5′-UGA​ACU​UGU​CUU​CUG​CUC​CAG​GAU​C-3′) and siRNA negative control (Stealth RNAi) were purchased from Invitrogen. The Stealth RNAi oligonucleotides were transfected into cells using Lipofectamine RNAi MAX (Invitrogen). After 2 d, the cells were again transfected with the same siRNA and cultured for an additional 2 d before analysis.

The plasmids encoding mCherry-STX17, RFP-Sec61β, GFP-ATG14 (Hamasaki et al., 2013), and DsRed-KDEL (Ito and Nagata, 2016) were described previously. ERdj8-mRFP and ERdj8-BFP were purchased (OriGene Technologies, Inc.). To construct ERdj8-Flag, KIAA0962 (ERdj8/DNAJC16) cDNA was purchased (Kazusa DNA Research Institute). The primers used were forward, 5′-GCC​GCA​GGT​ACC​GGA​AGA​GAA​ATG​GAA​GTG​AGA​AAG​TTG-3′ and reverse, 5′-TCC​GCC​GGG​CCC​TCA​CTT​ATC​GTC​GTC​ATC​CTT​GTA​ATC​GTC​TAG​TTC​AGG​GGA-3′. Amplification with these primers fused a FLAG tag to the C-terminus of ERdj8. This insert was subcloned into the KpnI–ApaI site of pcDNA3.1(+) (Invitrogen). The H57Q for DnaJ domain mutant (CAT to CAG) and the TRX domain mutant C174A (TGC to GCC), C177A (TGC to GCC) were created using the PrimeSTAR Mutagenesis Basal Kit (TaKaRa).

Fluorescence microscopy was performed using an FV1000 confocal microscope system equipped with a 60×, 1.35 NA UPlanSApo or a 100×, 1.40 NA UPlanSApo oil objective lens (Olympus Corp.), a TCS SP-8 conformal laser-scanning fluorescence microscope (Leica) equipped with a 63× objective (HC PL APO 63×/1.40 OIL CS2; Leica) with Leica HyVolution Deconvolution Imaging (Huygens), a DeltaVision Elite fluorescence microscope (GE Healthcare Life Science) equipped with a 60× PlanAPO oil immersion objective lens (Olympus; NA 1.42), and a scientific complementary metal oxide semiconductor (CMOS) camera, a SpinSR10 Super Resolution Imaging System (Olympus) equipped with a 60× PlanAPO oil immersion objective lens (Olympus; NA 1.42), or a LSM700 (Zeiss) confocal laser scanning microscope equipped with an α PlanAPO 100×/1.46Oil differential interference contrast (DIC) microscope M27 (Zeiss).

For immunostaining, cells were fixed with 4% PFA for 10 min, quenched with PBS, permeabilized with 20 µg/ml digitonin, blocked with 5% normal goat serum, and mounted in mounting reagent (Mowiol 4–88, nonfluorescent glycerol, and 0.2 M Tris-Cl, pH 8.5). For observation of proteins fused to fluorescent tags, the permeabilization step was omitted.

For live cell imaging, cells were placed on a glass-bottom dish (Matsunami glass) and maintained at 37°C with 5% CO₂.

After 2 d of knockdown, 1.5 × 10⁵ cells were seeded on a coverslip (for immunofluorescence) or directly (for immunoblotting) in a six-well plate. After 48 h, beads were prepared by mixing 1-µm or 3-µm beads (PolySciences, Inc.) with Effectene transfection reagent (QIAGEN; Kobayashi et al., 2010). The bead mixture (100 µl) was further mixed with 1 ml of growth medium and then added to cells by replacing the medium. After incubation with the bead mixture for 1 h, the cells were washed twice with fresh medium to remove unattached beads, incubated for an additional 24 h, and imaged. The area corresponding to each bead was selected as the region of interest (ROI), and the signal intensity of GFP and RFP within the ROI was quantified using ImageJ.

After 2 d of knockdown, 1.0 × 10⁵ HeLa cells were seeded in six-well plates on a coverslip. After 48 h, the cells were treated with 5 µM puromycin (Wako, 160–23151) for 6 h to induce the formation of large p62 bodies. Puromycin was washed out and cultured for 17 h. Cells were fixed with 4% PFA, stained with p62 antibody, and subjected to microscopy.

After 2 d of knockdown, 1.5 × 10⁵ mCherry-Parkin stable HeLa cells were seeded in six-well plates on a coverslip (for immunofluorescence) or directly (for immunoblotting). After 48 h, the cells were treated with 5 µM CCCP (Sigma, C2759-250MG) for 24 h to induce mitophagy. Cells were fixed with 4% PFA, stained with Cox-II antibody, and subjected to microscopy or subjected to Western blotting.

After 2 d of knockdown of ERdj8 or overexpression of ERdj8-BFP or BFP, HeLa cells stably expressing eGFP-LC3 were cultured on glass-bottom dishes with a grid pattern (MatTek, P35G-2–14-C-GRID) and starved for 1 or 2 h. The cells were fixed with 4% (wt/vol) formaldehyde in 30 mM Hepes buffer (pH 7.4) containing 100 mM NaCl, 2 mM CaCl₂, and 1 µg/ml Hoechst 33342 for 30 min at room temperature; washed with 30 mM Hepes buffer (pH 7.4) containing 100 mM NaCl and 2 mM CaCl₂; and examined using a confocal laser scanning microscope (Leica, SP-8). The same specimens were further incubated with 2% (wt/vol) formaldehyde and 2.5% (wt/vol) glutaraldehyde in 30 mM Hepes buffer (pH 7.4) containing 100 mM NaCl and 2 mM CaCl₂ at 4°C for overnight. After three washes in 30 mM Hepes buffer (pH 7.4) containing 100 mM NaCl and 2 mM CaCl₂, the samples were postfixed with 1% (wt/vol) osmium tetroxide in 30 mM Hepes buffer (pH 7.4) containing 0.5% (wt/vol) potassium ferrocyanide, 100 mM NaCl, and 2 mM CaCl₂ for 1 h, washed three times in distilled water, dehydrated in ethanol, and embedded in Epon812 (TAAB Laboratories Equipment). Ultrathin sections (70-nm thick) were stained with saturated uranyl acetate and Reynolds lead citrate solution. The electron micrographs were taken with a JEOL JEM-1011 transmission electron microscope.

The transgenic lines used in this study were dkIs398 (Ppie-1::GFP::lgg-1, unc-119(+)) and dkIs698 (Pspe-11::hsp-6::mCherry, unc-119(+); Sato and Sato, 2011) and SJ4103 (myo-3::GFP(mit); Benedetti et al., 2006). RNAi experiments were conducted using the feeding method (Kamath and Ahringer, 2003). L4 larvae were treated with RNAi, and embryos dissected from the F1 adults or oocytes of the F1 adults were scored. For RNAi experiments, L4440 containing a cDNA fragment of dnj-8 or atg-5 was used. For imaging, adult worms, eggs, and dissected embryos were mounted on agarose pads with M9 buffer containing 10 mM levamisole and observed using a laser scanning confocal microscope LSM700 or FV1000. For immunoblotting, eggs were lysed in SDS loading buffer (200 mM Tris-HCl, pH 6.8, 20% SDS, and 40% glycerol) at −80°C and then at 100°C for 15 min. The suspension was sedimented by centrifugation at 20,400 g for 10 min at 4°C. Lysates were mixed with 5% β-mercaptoethanol, incubated at 100°C for 5 min, and subjected to SDS-PAGE and Western blotting according to standard protocols. RT-PCR was performed as described previously for worm experiments (Williams et al., 1992). The sequences of the oligonucleotides for GFP::mito were as follows: forward, 5′-GGA​GAA​GAA​CTT​TTC​ACT​GG-3′; reverse, 5′-CCA​TGC​CAT​GTG​TAA​TCC-3′. Each reaction was repeated at least five times to assess reproducibility.

Fig. S1 shows additional ERdj8 localization and ERdj8 overexpression or knockdown on autophagy flux data. Fig. S2 shows a scheme of mRFP-eGFP-Galectin8 engulfment assay, additional effect of ERdj8 knockdown on mRFP-eGFP-Galectin8 1-µm bead assay, and data on dnj-8 knockdown on C. elegans. Video 1 shows intracellular localization of ERdj8-RFP and YFP-ATG5.

We thank Dr. S. Yamaoka (Tokyo Medical and Dental University, Tokyo, Japan) for pMRX-IRES-puro vector; T. Kitamura (The University of Tokyo, Tokyo, Japan) for Plat-E cells; and R. Yamashita (Kyoto Sangyo University, Kyoto, Japan) for worm experiments. SpinSR10 at the Advanced Cell imaging Core, Osaka University, was utilized in this study.

This work is supported by the Japanese Society for the Promotion of Science KAKENHI (16K07347 to T. Noda and 18H04002 to K. Nagata); a grant from the Takeda Science Foundation (to K. Nagata); the Japan Science and Technology Agency CREST (JPMJCR13M6 to K. Nagata); and the joint research program of the Institute for Molecular and Cellular Regulation, Gunma University (16026 to Y.-H. Yamamoto).

The authors declare no competing financial interests.

Author contributions: Conceptualization, Y.-H. Yamamoto, T. Noda., and K. Nagata; Writing – Original Draft, Y.-H. Yamamoto, S. Bar-Nun, K. Nagata, and T. Noda. Writing – Review and Editing, T. Noda. Investigation, Y.-H. Yamamoto (most aspects of the study); S. Bar-Nun (topology analysis); A. Kasai (mitophagy analysis). T. Takino, M. Sugihara, and M. Sato. (worm experiments); T. Umemoto (bead assays); T. Hatta (mass spec); H. Omori, R. Arai., S. Waguri, (EM analysis); resources, R.I. Morimoto, M. Hamasaki, T. Yoshimori, K. Sato, and T. Natsume.