Cellular and Molecular Life Sciences (2021) 78:645–660 https://doi.org/10.1007/s00018-020-03521-y Cellular and Molecular Life Sciences ORIGINAL ARTICLE Receptor‑mediated endocytosis 8 (RME‑8)/DNAJC13 is a novel positive modulator of autophagy and stabilizes cellular protein homeostasis Anna S. Besemer1 · Joanna Maus1 · Mirjam D. A. Ax1 · Anna Stein1 · Stella Vo1 · Christian Freese1 · Karsten Nalbach1 · Christian von Hilchen1 · Ines F. Pfalzgraf1 · Ingrid Koziollek‑Drechsler1 · Beate Silva1 · Heike Huesmann1 · Fatima Boukhallouk2 · Luise Florin3 · Andreas Kern1 · Christian Behl1 · Albrecht M. Clement1 Received: 7 October 2019 / Revised: 20 March 2020 / Accepted: 6 April 2020 / Published online: 22 April 2020 © The Author(s) 2020 Abstract The cellular protein homeostasis (proteostasis) network responds effectively to insults. In a functional screen in C. elegans, we recently identified the gene receptor-mediated endocytosis 8 (rme-8; human ortholog: DNAJC13) as a component of the proteostasis network. Accumulation of aggregation-prone proteins, such as amyloid-β 42 (Aβ), α-synuclein, or mutant Cu/ Zn-superoxide dismutase (SOD1), were aggravated upon the knockdown of rme-8/DNAJC13 in C. elegans and in human cell lines, respectively. DNAJC13 is involved in endosomal protein trafficking and associated with the retromer and the WASH complex. As both complexes have been linked to autophagy, we investigated the role of DNAJC13 in this degradative pathway. In knockdown and overexpression experiments, DNAJC13 acts as a positive modulator of autophagy. In contrast, the overexpression of the Parkinson’s disease-associated mutant DNAJC13(N855S) did not enhance autophagy. Reduced DNAJC13 levels affected ATG9A localization at and its transport from the recycling endosome. As a consequence, ATG9A co-localization at LC3B-positive puncta under steady-state and autophagy-induced conditions is impaired. These data dem- onstrate a novel function of RME-8/DNAJC13 in cellular homeostasis by modulating ATG9A trafficking and autophagy. Keywords Proteostasis · RME-8 · DNAJC13 · Autophagy · ATG9A · Recycling endosome · C. elegans Abbreviations HEK H uman embryonic kidney ATG Autophagy-related HPV H uman papilloma virus C. elegans C aenorhabditis elegans HSC70 H eat shock cognate 70 DNAJC D naJ-binding domain C LC3 M icrotubule-associated protein 1 light chain GFP G reen fluorescent protein 3 Luc L uciferase Electronic supplementary material The online version of this PI(3)P P hosphatidylinositol-3-phosphate article (https ://doi.org/10.1007/s0001 8-020-03521 -y) contains RME-8 R eceptor-mediated endocytosis 8 supplementary material, which is available to authorized users. SNX18 S orting nexin 18 * Christian Behl SOD1 Cu/Zn-superoxide dismutase cbehl@uni-mainz.de SQSTM1 Sequestosome 1 * Albrecht M. Clement TBC T re2/Bub2/Cdc16 domain clement@uni-mainz.de VPS V acuolar protein sorting 1 WASH W iskott–Aldrich syndrome protein and Institute of Pathobiochemistry, University Medical Center of the Johannes Gutenberg-University, Duesbergweg 6, SCAR homolog 55128 Mainz, Germany 2 Institute of Medical Microbiology and Hygiene, University Medical Center of the Johannes Gutenberg-University, 55101 Mainz, Germany Introduction 3 Institute for Virology and Research Center for Immunotherapy (FZI), University Medical Center A fine-tuned, multi-component molecular network is in of the Johannes Gutenberg-University, 55101 Mainz, place to preserve protein homeostasis (proteostasis) on Germany the cellular level, particularly under acute and chronic Vol.:(012 3456789) 6 46 A. S. Besemer et al. challenges. Despite the flexibility of the system, the ability RME-8/DNAJC13 is involved in several endosomal func- to adapt to stressors generally declines with age [1, 2]. As a tions such as protein sorting [26, 31], endosomal tubulation consequence, aged cells are prone for the accumulation of [32], transport processes including the retrograde transport misfolded proteins which, in turn, exacerbates the imbalance from the endosome to the trans-Golgi network (TGN) [27, of the proteostasis network. This vicious cycle, resulting in 33, 34], the formation of endosomal degradative microdo- the accumulation of protein aggregates, is at least in part the mains [35], and the recycling of membrane receptors [36]. molecular basis for the development of several age-related In addition, it has been proposed that DNAJC13 might act neurodegenerative diseases, like Parkinson’s disease and as a scaffold to organize the retromer and WASH complexes, amyotrophic lateral sclerosis [3]. since DNAJC13 interacts with FAM21 [32], a component Macroautophagy (hereafter termed autophagy) is one of the WASH complex, and SNX1 [33, 34], a constituent of major pathway to remove misfolded proteins and its activ- the sorting nexin dimer tightly linked to the core retromer ity is reduced with age on the cellular and the organism trimer, consisting of VPS26, VPS35, and VPS29. levels [4]. Generally, autophagy directs cytosolic cargos, but There is growing evidence that retromer dysfunction is also invading pathogens, to lysosomal degradation [5] and linked to neurodegeneration [37, 38]. More specifically, its dysfunction has been linked to severe pathologic con- point mutations within DNAJC13 have been associated with ditions including neurodegenerative diseases [6]. The deg- rare familial cases of Parkinson’s disease (PD) with Lewy radative pathway can be induced by starvation conditions body pathology [39]. The expression of mutant DNAJC13 and the treatment with compounds such as rapamycin that resulted in the accumulation of α-synuclein in mammalian inhibit the Ser/Thr kinase mechanistic target of rapamycin cells, and in Drosophila [40, 41], pointing towards a role (mTOR) [7]. The inhibition of mTOR results in a transloca- of DNAJC13 in proteostasis. It is of note that the DNAJC tion of the ULK complex from the cytosol to the ER. The protein family as a subclass of HSP40 heat shock proteins detailed mechanism of phagophore initiation and elongation received attention due to the identification of various muta- is still enigmatic [8]. The processes depend on the ubiqui- tions that are linked to PD and other age-related neurological tin-like ligation complex ATG12-ATG5/ATG16L1 which is disorders [42]. responsible for the lipidation of MAP1LC3 (shortly LC3), In a previous screen, we uncovered that rme-8/DNAJC13 a member of the Atg8 protein family. Membrane-anchored is involved in maintaining proteostasis in acute heat stress LC3-II is essential for phagophore elongation, and acts also models in C. elegans [43, 44], and here, we transferred these as an attachment site for autophagy cargo receptors such findings in human cell lines. The accumulation of aggre- as SQSTM1/p62 [9]. Membranes and lipids are delivered gation-prone proteins upon rme-8/DNAJC13 knockdown from different sources within cells such as the endoplasmic and its functional association with the retromer and WASH reticulum [10], mitochondria [11, 12], the Golgi–endosomal complexes, both of which prompted us to investigate the role system [13], or lipid droplets [14, 15], indicating that lipid of RME-8/DNAJC13 in autophagy. We could demonstrate transport processes that are supported by RAB GTPases are that RME-8/DNAJC13, but not the PD-associated mutant essential for the maturation of phagophores to autophago- DNAJC13(N855S), is a positive modulator of this degrada- somes [16]. An important component for the maturation of tive process. Moreover, we found that DNAJC13 impacts autophagosomes and the delivery of lipids towards the grow- on the subcellular distribution of ATG9A. Since ATG9A ing phagophore is ATG9A [17, 18]. Several lines of evidence localization at and its transport from the recycling endo- suggested that the routing of ATG9A through the recycling some was reduced at DNAJC13 knockdown conditions, we endosome is an essential step for proper autophagosome suggest that DNAJC13 modulates autophagy by affecting biogenesis [19–24]. ATG9A trafficking. Rme-8 was initially identified as a component of the endo- cytosis machinery in C. elegans [25], belongs to the DNAJ/ HSP40 protein family, and is highly conserved [25–28]. Materials and methods The human ortholog of rme-8 is DNAJC13, a large protein containing 2243 amino acids. Due to the presence of the C. elegans central DNAJ domain, DNAJC13 binds HSC70 (heat shock cognate 70) [27, 29] and acts as a co-chaperone in mediat- Caenorhabditis elegans strains were obtained from the ing HSC70 cellular functions. Within the cell, DNAJC13 is Caenorhabditis Genetics Center (CGC), which is funded primarily localized at membranous structures. This interac- by NIH Office of Research Infrastructure Programs (P40 tion is mediated, at least in part, by the binding to phos- OD010440). Strains were cultivated on HB101 Escherichia phoinositides (PI) through its N-terminal sequence [30]. coli containing nematode growth medium (NGM) plates 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 647 according to the standard procedures. The large-scale, from I. Dikic, Frankfurt, Germany) by electroporation and unbiased RNAi screen of chromosome one in nematodes plated onto glass cover slips. Cells were cultivated for 48 h expressing a luciferase-GFP fusion protein (Luc-GFP) in and then treated with DMSO (control) or rapamycin (10 µM) the body wall muscle cells [45] was performed as described for 4 h. [44]. To analyze the effect of the newly identified gene rme- 8 on the paralysis of human Aβ42-expressing C. elegans, Generation of stable Luc‑GFP expressing HEK293A synchronized CL2006 (dvIs2 [pCL12(Punc-54-hAβ)/pRF4]) cells and heat stress experiments nematodes [46] were cultivated at 15 °C and placed on RNAi plates at the L4 stage. Worms were transferred onto A luciferase–GFP fusion construct [45] was subcloned in new plates daily and paralysis was tested by gently tipping the pLHCX vector (Invitrogen) (pLHCX:Luc-GFP). Viruses the nose of the nematode with a platinum wire. Worms that were produced as previously described [49] and specified moved their nose but were not able to move their body were in the Supplemental Methods. Virus-containing supernatant scored as paralyzed. Nematodes showing additional aberrant was incubated with HEK293A cells for 36 h, followed by an phenotypes were not included into the statistics. The staining incubation of transfected cells in complete medium contain- of amyloid structures by thioflavin was performed as previ- ing 250 µg/ml hygromycin (Life Technologies, 10687010) ously described [47]. to select for stably transfected cells. The efficiency of the rme-8 knockdown was monitored For heat stress experiments, stably transfected Luc-GFP by quantitative real-time PCR (qPCR). For RNA extraction cells were cultivated in six-well plates after transient trans- and quantification, RNA was isolated from total lysate of fection of siRNA by electroporation. 48 h after transfec- 15–20 worms using Absolutely RNA Nanoprep Kit (Agilent, tion, the medium was changed and plates were placed for 400753). Reverse transcription and qPCR were performed 1 h in a water bath inside an incubator with a tempera- as described in detail in the Supplemental Method section. ture of 41 °C. The heat stress paradigm was established as such that the stress did not alter cellular morphology Cell culture and induced luciferase accumulations in about 5% of cells after nonsense siRNA transfection. The GFP signal was Human embryonic kidney cells 293 (HEK293A or documented by a conventional Zeiss inverted microscope HEK293T) were cultivated in high glucose containing Dul- equipped with a monochrome camera (Spot RT, National becco’s modified Eagle’s medium (DMEM, Life Technolo- Diagnostics). At least 200 cells per experiment and condi- gies, 41965062) supplemented with 10% fetal calf serum tion were analyzed, and the number of cells with bright (FCS) (PAA Laboratories, A15-101), 1% sodium pyruvate fluorescent Luc-GFP accumulations was determined. Num- (Life Technologies, 1136-088), and a mixture of antibiot- bers were normalized and statistics were performed relative ics and antimycotics (Life Technologies, 15240-112). HeLa to the number of cells with accumulations under nonsense cells were cultivated in DMEM supplemented with 10% siRNA conditions. FCS, 1% Glutamax (Life Technologies, 35050061), 1% modified Eagle’s medium nonessential amino acids (Life HPV pseudovirion transduction Technologies, 11140050), and antibiotics (Life Technolo- gies, 15140122). Cultures were kept at 37 °C in a humidified HeLa cells were transfected with siRNA against DNAJC13 atmosphere containing 5% CO2. or nonsense siRNA using Lipofectamine RNAiMAX For transient expression, HEK293 cells were transfected (Invitrogen, 13778150) according to the manufacturer’s by calcium phosphate precipitation as described earlier [48] instructions. 24  h post-transfection, GFP-LC3B, and and specified in the supplemental material. A set of two siR- mCherry-ATG9A expressing plasmids were introduced by NAs against DNAJC13 or nonsense siRNA (Suppl. Table 1) Lipofectamine (Invitrogen, 11668019) for additional 24 h. (MWG Eurofines Genomics, Sigma) were introduced by Afterwards, human papillomavirus type 16 (HPV16) pseu- electroporation using the Nucleofactor 2B system (Lonza). dovirions were added. HPV pseudovirions were prepared 48 h after transfection cells were exposed to 4 µM bafilomy- as described earlier [50]. Briefly, HEK293TT (SV40 large cin A1 (LC Laboratories, B-1080) and/or 10 µM rapamycin T antigen expressing HEK293T cells) cells were co-trans- (Enzo, BML-A275-0025) for 4 h. Stock solutions were pre- fected with expression plasmids for HPV16 L1 and L2 as pared in DMSO (Roth, A994.2). HeLa cells were transfected well as with pcDNA3.1/Luciferase plasmids [51] and lysed with siRNA against DNAJC13 or nonsense siRNA and GFP- 48 h post-transfection. Pseudovirions were purified by an LC3 and mCherry-ATG9A expressing plasmids (kind gift Optiprep (Sigma, D1556) gradient centrifugation. The 1 3 6 48 A. S. Besemer et al. addition of 100 luciferase vector-positive pseudovirions per phosphate-buffered saline (PBS) (Sigma, D8537) and cell resulted in labeling of all HeLa cells with HPV virions. incubated with primary antibodies specifically detect- Transduction/infection efficiencies of HPV pseudovirions ing LC3B (Nano Tools, 0260-100), sequestosome 1/p62 were assessed by quantification of luciferase expression 24 h (SQSTM1) (Progen, GP62-C), early endosome antigen post-infection. Cells were lysed with cell culture lysis rea- 1 (EEA1) (BD Transduction Labs, 616457), ATG9A gent and relative luciferase activity was measured with the (Abcam, ab108338), cation-independent mannose- luciferase assay system (Promega, E4030) according to the 6-phosphate receptor (CI-M6PR) (Abcam, ab8093), lyso- manufacturer’s instructions, using the Tristar LB 941 lumi- somal-associated membrane protein 2 (LAMP2) (Abcam, nometer (Berthold Technologies). Luciferase activities were ab25631), trans-Golgi network protein 2 (TGN46) (ABD normalized to lactate-dehydrogenase (LDH) activity for cell Serotec, AHP500), ATG16L1 (MBL, PM04), and FLAG viability (CytoTox-One, Promega, G7891) and are depicted (Sigma, F1804). Subsequently, primary antibodies were as percentages of infection relative to the luciferase activity detected with species-specific secondary antibodies cou- in control-treated cells. pled to fluorescence dyes. After another washing step with PBS, DAPI (Calbiochem, 382061) or Hoechst 33342 Generation of expression constructs (Sigma, B2261) was added for nuclear counterstaining. Coverslips were mounted and analyzed with Leica SPE or For the generation of C-terminally tagged DNAJC13 SP5 confocal microscopes (at the Institute for Molecular constructs, FLAG and EGFP were amplified with for- Biology, Mainz, Germany) or a Zeiss 710 confocal micro- ward and reverse primers adding a BamHI and an XbaI scope. Images were carefully taken avoiding overexpo- site, respectively. DNAJC13 cDNA was excised from a sure within cells. In case of GFP-RAB11 and GFP-RAB7 DNAJC13 expression plasmid [26] (kind gift from K. transfected cells, only cells with a dotted distribution of Sekiguchi, Osaka, Japan) and fragments were re-ligated GFP were chosen for analysis and the GFP signal was in the pEF-BOS-EX vector (kind gift of S. Nagata, Kyoto, monitored. We intentionally analyzed unprocessed confo- Japan) (pEFBos:DNAJC13-FLAG; pEFBos:DNAJC13- cal single slice images using the plot profile tool within EGFP). DNAJC13 mutations were introduced in the the FIJI software [55]. Lines were drawn throughout cells pEF-BOS-DNAJC13 Vectors with the QuikChange Light- based on the distribution of ATG9A blinded to the GFP ning Site-directed Mutagenesis kit (Agilent, 210518). signals (Suppl. Figure 7C). At least 13 lines resulting in DNAJC13(N855S): forward (5′-gcgatgataaagctcact- about 85 puncta per cell and between 9 and 37 cells per gaaaaattcatacgatctcttaattgatcc-3′) and reverse primers condition were analyzed. Subsequently, intensity profiles (5′-ggatcaattaagagatcgtatgaatttttcagtgagctttatcatcgc-3′); were compared with peak intensities from co-stainings on DNAJC13(W20A): forward (5′-cgcttatacttccccctcgctgaat- the very same lines, and overlapping peaks were counted gttttgttgtgtagaaacatcg-3′) and reverse primers (5′-gcatgtttc- and normalized to all ATG9A-positive signals. Fluo- tacacaacaaaacattcagcgagggggaagtataagcg-3′). Wild-type rescence imaging for virus experiments was performed α-synuclein was amplified from a prion-promotor-driven using a Zeiss Axiovert 200 M microscope equipped with expression plasmid [52] (kind gift from G. Auburger, Frank- a Plan-Apochromat 100 × (1.4 NA). About eight ran- furt, Germany) and introduced in the pEGFP-N1 vector. domly selected and unprocessed images per condition in SOD1(G85R) was expressed within the pEGFP-N1 vector three independent experiments (about 100 cells in total) as described earlier [53]. The sequence of all constructs were analyzed using Axiovision co-localization software was verified by sequencing analysis (MWG). N-terminally 4.7 (Zeiss, Jena, Germany). A fixed threshold was set to tagged GFP-RAB11 [54] (generously provided by M. I. remove background staining for each experiment. Colombo, Mendoza, Argentina, via R. Prange, Mainz, Ger- many) and GFP-RAB7 (kind gift from A. Helenius, Zurich, Protein extraction, fractionation, and Western Switzerland, via M. Husmann, Mainz, Germany) were blotting expressed from pEGFP-C plasmids. Total cell extracts were prepared in lysis buffer contain- Immunocytochemistry and microscopy ing 62.5 mM Tris–HCl, 2% (w/v) SDS and 10% sucrose (pH 8) supplemented with EDTA-free protease (Roche, Cells were grown on glass coverslips and treated as 04693159001) and phosphatase inhibitor cocktails (Roche, described above. Cells were fixed with 4% paraformal- 04906837001) and sonicated four times at 60 Hz for 10 s dehyde (Sigma, P6148) and, if necessary, permeabilized on ice. Fractionation experiments after overexpression of with − 20 °C-cold methanol (95%) for 6 min, washed with aggregation-prone proteins were performed as described 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 649 earlier [56] and specified in the Supplemental Methods. DNAJC13 is involved in proteostasis also in human cells. For Western blotting, equal amounts of proteins were sepa- We generated an HEK293A cell line stably expressing lucif- rated by SDS-PAGE under denaturing conditions. Proteins erase-GFP and analyzed the generation of its aggregates were transferred on nitrocellulose membrane and proteins following a heat stress paradigm. Exposure to heat stress were detected by specific antibodies recognizing SQSTM1 under control conditions with nonsense siRNA or siRNA (Progen, GP62-C), LC3B (Novus, NB-100-2220), FLAG directed against glyceraldehyde-3-phosphate-dehydrogenase (Sigma, F1804), tubulin (Sigma, T9026), SOD1 (Epito- (GAPDH) had only a very mild effect on luciferase–GFP mics, 2018-1), ATG9A (Abcam, ab108338), histone H3 inclusion formation and did not induce overt morphological (Abcam, ab47915), α-synuclein (Abcam, ab27766), and changes in these cells (Fig. 1c). In contrast, the knockdown DNAJC13 (kind gift from M. Farrer, Vancouver, and P. of heat shock factor 1 (HSF1) and heat shock cognate 70 McPherson, Montreal, Canada). Species-specific second- (HSC70) resulted in a significantly increased number of ary antibodies coupled to horseradish peroxidase (Dianova) cells with luciferase-GFP accumulations compared to non- were used to develop the Western blot with the Fusion sys- sense siRNA and served as positive control (Fig. 1c; Suppl. tem (Peqlab) or the Amersham Imager 600 (GE). Densi- Figure 1A, B). Under the same conditions, the number of tometry was performed with the AIDA software (Raytest) cells with luciferase-GFP inclusions treated with DNAJC13 or ImageJ. siRNA was significantly increased compared to nonsense siRNA controls (Fig. 1c). Statistical analysis The relevance of DNAJC13 for maintaining proteostasis was also shown by the analysis of mutant Cu/Zn-superoxide For the comparison of two groups, variance was deter- dismutase (SOD1) and α-synuclein aggregation in human mined and Student’s t test (two-tailed, unpaired) was applied cell lines. Mutant SOD1 variants cause familial forms of accordingly. In some cases, Mann–Whitney U test or the amyotrophic lateral sclerosis and α-synuclein is the main one-way ANOVA with Bonferroni or Games–Howell cor- constituent of Lewy bodies which are a prominent pathologi- rection for multiple comparisons was used (SPSS software, cal hallmark of PD. The ectopic expression of SOD1 carry- IBM; SigmaStat, Systat; Prism). Statistical analysis for virus ing the point mutation SOD1(G85R) [53, 57] or α-synuclein experiments was performed using Statistical Software R [58] results in the formation of inclusions. HEK293T (2017, version 3.3.1) from R Core Team (R: A language and cells transiently overexpressing either SOD1(G85R) or environment for statistical computing. R Foundation for Sta- α-synuclein were additionally co-transfected with nonsense tistical Computing, Vienna, Austria). All values were tested or DNAJC13 siRNA (Fig. 1d, e; Suppl. Figure 1C, D). The for normality using the Shapiro–Wilk test. As the values separation of cell lysates in a soluble and an aggregate- were not normally distributed, Wilcoxon rank-sum test was enriched fraction revealed that reduced DNAJC13 protein applied. levels resulted in an accumulation of mutant SOD1 and α-synuclein within the aggregate fraction (Fig. 1d; Suppl. Figure 1D). Concomitant with the biochemical data, the Results number of SOD1(G85R)-transfected cells bearing aggre- gates is increased under DNAJC13 knockdown conditions rme‑8/DNAJC13 knockdown affects proteostasis (Fig. 1e), emphasizing that diminished DNAJC13 levels in C. elegans and HEK293 cells facilitated protein aggregation; DNAJC13 knockdown did not promote SOD1(WT) aggregation (Fig. 1d, e). Previously, we detected the gene rme-8 in an RNAi screen using C. elegans expressing the protein folding sensor Decreased DNAJC13 levels reduce autophagic luciferase-GFP to identify modulators of proteostasis [43, activity under basal and stimulated conditions 44]. To confirm the role of RME-8 in proteostasis, nema- todes expressing human amyloid-β 42 (Aβ42) in body wall It has been well established that protein aggregates can be muscle cells were employed. This transgenic line (CL2006) efficiently removed by autophagy and that, in turn, age- is characterized by the age-dependent appearance of Aβ42 related or mutation-induced impairment of autophagic aggregates and an associated progressive paralysis [46]. activity cause accumulation of misfolded proteins [6]. In Upon rme-8 RNAi treatment, the number of aggregates was addition, recent evidence that the precise interaction of the increased compared to controls (eV) as shown by thioflavin WASH and retromer complexes is required for autophagy staining (Fig. 1a). Consistently, the paralysis phenotype was [59] prompted us to investigate whether DNAJC13 is aggravated in Aβ42-expressing nematodes (Fig. 1b). involved in autophagy. To determine autophagic flux, cells 1 3 6 50 A. S. Besemer et al. 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 651 ◂Fig. 1 Knockdown of rme-8/DNAJC13 affects proteostasis. a, b C. Figure 3A–C). Furthermore, we employed RPE-1 cells elegans expressing Aβ42 in body wall muscle cells (CL2006) were stably expressing GFP-LC3B-RFP-LC3B(ΔG) [60] and fed with rme-8 RNAi or bacteria containing empty vector (eV) start- ing from the L4 stage. a Two-day adult animals show an increased determined the GFP/RFP ratio in individual cells upon non- number of thioflavin-positive accumulations (arrows) in the head sense and DNAJC13 siRNA treatment and basal or starva- region under rme-8 knockdown conditions. About 40 worms were tion induced autophagy conditions. We observed that the counted and results were represented in a box-plot (n = 41/42; Mann– GFP/RFP ratio was increased upon DNAJC13 knockdown Whitney U test: *p ≤ 0.05). Micrographs represent maximum inten- sity projections of the head region of worms (scale bar: 20  µm). b in either condition (Suppl. Figure 4) comparable to ATG3 CL2006 under rme-8 knockdown conditions developed an aggra- siRNA-treated cells, indicating that autophagy is inhibited. vated paralysis compared to control nematodes (eV). In total, 151 (eV) and 202 (rme-8) worms were analyzed (n = 4; mean ± SEM; t DNAJC13 overexpression augments autophagic flux test: *p ≤ 0.05, **p ≤ 0.01). c Phase-contrast and epi-fluorescent pic- tures of HEK293A stably expressing Luc-GFP under control condi- tions or after heat stress and siRNA treatment. At least 200 cells per Since knockdown of DNAJC13 resulted in decreased experiment and condition were analyzed and the number of cells with autophagy, we investigated whether the overexpression of bright fluorescent Luc-GFP accumulations (arrowheads) was deter- DNAJC13 might positively affect autophagic activity. Ele- mined. The inset shows an enlargement of the boxed cell. Numbers were normalized and statistics were performed relative to the number vated protein levels of DNAJC13 resulted in an enhanced of cells with accumulations under nonsense siRNA conditions (n = 3; flux of LC3B-II and SQSTM1 compared to controls (Fig. 2e, mean ± SEM; t test: *p ≤ 0.05, ***p ≤ 0.001) (scale bar: 20  µm). d, f). The increased autophagic activity upon DNAJC13 over- e HEK293T cells were transiently transfected with EGFP-tagged expression was not mediated through an elevated transcrip- wild-type SOD1(WT) or ALS-causing mutant SOD1(G85R) and nonsense siRNA or DNAJC13 siRNA. d Cell lysates were separated tion of autophagy-related genes like ATG7, ATG3, LC3B, or in a soluble (S) and an aggregate-enriched fraction (P) and the rela- SQSTM1 (Suppl. Figure 5A), despite LC3B-I and SQSTM1 tive distribution of the SOD1-EGFP variants were quantified (n = 3; protein levels were increased (Suppl. Figure 5C). To analyze mean ± SD; t test: *p ≤ 0.05) (lower right panel). Endogenous SOD1 if the direct interaction of DNAJC13 with phosphoinositides and histone H3 served as markers for the soluble and the aggregate- enriched fraction, respectively (upper panel). Representative West- (PI), most prominently to PI(3)P, is relevant for this effect, ern blots showing the expression levels of transgenic SOD1-EGFP we generated an FLAG-tagged mutant DNAJC13(W20A). variants and DNAJC13 levels in total cell lysates (lower left panel). e This variant shows a dramatically reduced interaction with Confocal images of HEK293T cells transfected with SOD1(WT) and membranes [30]. The analysis of accumulated LC3B-II after SOD1(G85R) and nonsense or DNAJC13 siRNA. Images were taken with the same exposure conditions. Open arrows point to cells with- bafilomycin A1 treatment suggested that DNAJC13(W20A) out aggregates; full arrows point to aggregates. The number of trans- augmented autophagic flux comparable to increased levels fected cells with SOD1(G85R) aggregates was determined (n = 4; of DNAJC13(WT) (Suppl. Figure 5D, E). In contrast, over- one-way ANOVA with Bonferroni correction; *p ≤ 0.05) (scale bar: expression of the PD-related DNAJC13(N855S) mutant 10 µm) did not increase the autophagic activity (Suppl. Figure 5F, G), indicating an autophagy defect in Parkinson’s causing were exposed to bafilomycin A 1, a compound that prevents DNAJC13(N855S) mutant. It is of note that DNAJC13 itself the acidification of lysosomes and thereby leads to the accu- is not degraded by autophagy as the protein levels of endog- mulation of autophagosomes and autophagy targets, such enous DNAJC13 as well as of DNAJC13-FLAG after over- as SQSTM1, an autophagy cargo receptor, and the lipidated expression were not significantly affected by bafilomycin A1 form of LC3B, LC3B-II. When HEK293 or human primary treatment (Suppl. Figure 5H, I). fibroblasts (IMR90) were treated with DNAJC13 siRNA, the autophagic activity was decreased compared to cells treated ATG9A subcellular distribution is altered with nonsense siRNA (Fig. 2a–d; Suppl. Figure 2). Reduced upon DNAJC13 knockdown DNAJC13 levels also led to lower autophagic flux under rapamycin treatment (Fig. 2a, c). The transient knockdown As ATG9A is shuttling through the endosomal compart- of DNAJC13 did not alter the expression of genes involved ment [18, 20, 21, 23] and DNAJC13 is involved in the in autophagy like LC3B or SQSTM1, nor were the protein trafficking of endosomal cargos [31–34, 36], we analyzed levels of LC3B-I and SQSTM1 affected under basal condi- whether DNAJC13 plays a role in ATG9A trafficking and tions (Suppl. Figure 5A, B). investigated the subcellular distribution of ATG9A fol- These data were independently confirmed in HeLa lowing DNAJC13 knockdown. In nonsense siRNA-treated cells transiently transfected with GFP-LC3B, showing a HEK293A cells, ATG9A was concentrated in a perinuclear reduced number of LC3B-positive puncta per transfected compartment as described earlier [61] whereas the distribu- cell under basal conditions and after rapamycin treatment tion was more diffuse in cells treated with siRNA against upon DNAJC13 knockdown compared to controls (Suppl. DNAJC13 (Suppl. Figure 6A), pointing towards a change 1 3 6 52 A. S. Besemer et al. 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 653 ◂Fig. 2 DNAJC13 positively modulates autophagy. a–d HEK293A in lysosomes (LAMP2-positive compartment) (Suppl. Fig- cells were transfected with nonsense (nons) or DNAJC13 siRNA. ure 8B) or the endoplasmic reticulum (CLIMP63-positive 48  h after transfection cells were treated with bafilomycin A 1 and/ or rapamycin for additional 4 h. a–c Cell extracts were separated on compartment) (Suppl. Figure 8C). SDS-PAGE and transferred on nitrocellulose membrane. Expression Upon autophagy induction, ATG9A vesicles shut- levels of LC3B-II and SQSTM1 were quantified and normalized to tle between the trans-Golgi, the endosome, or the “ATG9 tubulin. The lower band of the LC3B-I duplet in some blots may rep- compartment” and the autophagosome precursor [61]. It resent a processing intermediate [71]. Autophagic flux under basal conditions (b) or rapamycin treatment (c) was determined by subtrac- was previously suggested that ATG9A does not integrate tion of LC3B-II and SQSTM1 levels without bafilomycin A (ctr) in the phagophore membrane, but ATG9A vesicles deliver 1 from LC3-II and SQSTM1 levels with bafilomycin A1 (Bafi), respec- membranes and important proteins towards the growing tively (n = 4 [basal], n = 3 [rapamycin]; mean ± SEM; t test: *p ≤ 0.05, phagophore [18]. In this process, the routing of ATG9A ***p ≤ 0.001). d Confocal images of HEK293A cells transfected with nonsense or DNAJC13 siRNA as above and stained for SQSTM1 through the recycling endosome is an essential step for and LC3B. The nucleus is detected by DAPI (scale bar: 20  µm). e, proper autophagosome biogenesis [20–23]. After induction f Autophagic flux in HEK293T cells transiently overexpressing of autophagy with rapamycin in control cells (nonsense DNAJC13-FLAG was determined as in a, b (n = 4; mean ± SEM; t siRNA), ATG9A co-localization with RAB11-positive endo- test: *p ≤ 0.05) somal structures was reduced (Fig. 3a) which is consistent with earlier reports [21]. In contrast, co-localization under of the distribution of ATG9A-containing subcellular com- DNAJC13 knockdown conditions was unchanged at low partments. Total ATG9A protein levels were comparable level after rapamycin treatment when normalized to DMSO- between both conditions based on Western blot analysis treated cells (Fig. 3b), indicating that DNAJC13 affects the (Suppl. Figure 6B). Moreover, we could demonstrate a par- trafficking of ATG9A to and from the recycling endosome. tial co-localization of FLAG-tagged DNAJC13 with endog- Our data indicate that the modulation of autophagy by enous ATG9A by immunofluorescence analysis (Suppl. Fig- DNAJC13 is mediated through an altered ATG9A traf- ure 6C). It is of note that besides DNAJC13, also VPS35 ficking. This direct functional link between DNAJC13 and partially co-localizes with ATG9A confirming the previous ATG9A was further supported by experiments where we data [59]. The co-localization of DNAJC13 and ATG9A is assayed autophagic flux in cells overexpressing DNAJC13 further supported by experiments employing sucrose density upon transient knockdown of ATG9A. The reduction of gradient centrifugation. Endogenous DNAJC13 co-migrates ATG9A protein levels resulted in a comparable decrease of in the same fractions as endogenous ATG9A as well as, in autophagic activity in control cells as well as in cells overex- part, VPS35 (Suppl. Figure 6D). These data suggest that pressing DNAJC13 (Suppl. Figure 9B), suggesting that the DNAJC13, ATG9A, and VPS35 reside within the same effect of DNAJC13 overexpression depends on a sufficient cellular compartment and that DNAJC13 interferes with amount of ATG9A. ATG9A transport through the retromer and possibly the WASH complex [59]. DNAJC13 knockdown affects ATG16L1 trafficking To analyze ATG9A subcellular distribution in more and ATG9A co‑localization with LC3B detail, its localization in the trans-Golgi network, the late endosome, and the recycling endosome was studied by The recycling endosome serves as a hub for compiling immunofluorescence labeling with anti-TGN46 antibodies ATG9A vesicles. One example for an uptake of cargo in or the expression of GFP-tagged RAB7 and RAB11, respec- ATG9A vesicles at the recycling endosome is ATG16L1, tively. The co-localization of ATG9A to all three compart- a protein required for LC3B lipidation. Although ATG9A ments was reduced upon DNAJC13 knockdown under basal and ATG16L1 enter the recycling endosome on different autophagy conditions (Fig. 3; Suppl. Figure 7A, B). The routes, the proteins traffic towards the phagophore in the overall distribution of the trans-Golgi network (TGN46), the same vesicles [62]. As ATG9A trafficking from the recy- late (RAB7), and recycling (RAB11) endosome were not cling endosome was impaired upon DNAJC13 knockdown, substantially altered when DNAJC13 levels were reduced. we hypothesize that, as a consequence, the formation of In contrast, the distribution of the mannose-6-phosphate autophagosomes is affected. To investigate this, we ana- receptor (CI-M6PR) which is transported in a DNAJC13- lyzed the trafficking of ATG16L1 and the co-localization dependent manner is changed from a perinuclear staining of ATG9A with LC3B under autophagy-stimulating con- to a more disperse localization throughout the cell (Suppl. ditions. It had previously been shown that ATG16L1 is Figure 8A), confirming earlier data [32]. Reduced DNAJC13 present at the phagophore, but not on autophagosomes levels did not result in an untypical localization of ATG9A [63]. To monitor ATG16L1 trafficking, we analyzed the 1 3 6 54 A. S. Besemer et al. Fig. 3 Knockdown of DNAJC13 alters localization of ATG9A at overlap of ATG9A intensity peaks with GFP-RAB11 was quantified. the recycling endosome. Confocal images of HEK293A cells trans- The quantification method is described in detail in the Materials and fected with nonsense or DNAJC13 siRNA and co-transfected with Methods section. At least 13 regions of interest (ROIs) covering GFP-RAB11 under steady-state conditions (a) or rapamycin treat- about 85 intensity peaks per cell were analyzed in 9–37 cells per con- ment (b, c). The box in each merged panel is enlarged and the fluo- dition (mean ± SD; t test: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001) (scale rescent intensities along the white line are represented in the dia- bar: 15 µM) gram, whereby ATG9A is shown in red throughout all panels. The number of ATG16L1-positive dots per cells under basal in the cytosol. Most of these dots were WIPI2 positive conditions and after autophagy stimulation with rapamy- (Suppl. Figure 9A), indicating that these represent pha- cin. ATG16L1 was usually localized in one or two adja- gophores. In this assay, less ATG16L1 dots per cell were cent bright dots close to the nucleus (Fig. 4a, b). After counted under DNAJC13 knockdown compared to control autophagy induction, ATG16L1 dots were distributed cells (Fig. 4a–c). 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 655 Comparable to ATG16L1, ATG9A is not integrated in DNAJC13 is linked to the retromer and the WASH com- the membrane of mature autophagosome, but a substan- plexes which represent molecular machineries for the sort- tial portion is co-localized with LC3B upon autophagy ing of endosomal proteins and the formation of distinct induction [18]. We analyzed the localization of transiently endosomal subdomains by forming actin-polymerization expressed GFP-LC3B and mCherry-ATG9A upon nonsense patches, respectively [65, 66]. The retromer complex con- or DNAJC13 siRNA treatment under basal conditions and sists of the classical cargo selection complex (VPS26, upon autophagy induction employing HeLa cells infected VPS35, and VPS29) that is linked to a sorting nexin dimer with human papillomavirus (HPV) [64]. Analysis of GFP (SNX1/SNX2; SNX5/SNX6) [67]. The WASH complex and mCherry fluorescence revealed that LC3B and ATG9A is a protein pentamer that is involved in the formation of were partially co-localized under control conditions which branched actin networks on endosomes [66]. By interact- was significantly reduced (Fig.  4d, f) upon DNAJC13 ing with SNX1 [33, 34], a constituent of the sorting nexin knockdown (Fig.  4h). When autophagy was stimulated dimer, and FAM21 [32], a component of the WASH com- by treatment with HPV pseudoviruses, a general increase plex, DNAJC13 is believed to orchestrate the interaction in GFP-LC3B and mCherry-ATG9A co-localization was and thus the function of both complexes. observed. In line with the data above, reduced DNAJC13 A coordinated action of WASH and retromer complexes levels resulted in a significant decrease of LC3B and ATG9A has also been implicated in autophagosome formation by co-localization compared to nonsense siRNA-treated cells regulating ATG9A trafficking from endosomal membranes. after autophagy stimulation (Fig. 4e, f). It is of note that ATG9A is an important component for the maturation of despite HPV virions were bound to and taken up by almost autophagosomes and the only transmembrane protein of the all HeLa cells independent of siRNA treatment, the delivery autophagy core machinery [17]. Upon autophagy induc- of virion DNA towards the nucleus and thereby the expres- tion, ATG9A vesicles shuttle between the trans-Golgi, the sion of luciferase as a marker protein was impaired in cells endosome, or the “ATG9 compartment” and the autophago- treated with DNAJC13 siRNA (Fig. 4g). some precursor [61]. It was previously suggested that ATG9A does not integrate in the phagophore membrane, but ATG9A vesicles deliver membranes and other cargos Discussion towards the growing phagophore [18]. In this process, the routing of ATG9A through the recycling endosome We uncovered rme-8/DNAJC13 as a novel factor to main- is an essential step for proper autophagosome biogenesis tain proteostasis in C. elegans and in human cells. Upon [19–24]. reduction of RME-8/DNAJC13 protein levels, accumula- The impairment of ATG9A localization and trafficking tion of aggregation-prone and age-related disease-causing upon knockdown of DNAJC13 is reminiscent of the proteins were found to be increased. As this phenotype ATG9A phenotype after the expression of the PD-related points towards a defect in protein degradative pathways, VPS35(D6120N) mutant. In these exper iments, we investigated the role of DNAJC13 in autophagy. Alto- VPS35(D620N) did destabilize neither the retromer nor gether, our data now provide first evidence that DNAJC13, the WASH complexes [59, 68], but resulted in an impaired but not the PD-related mutant variant DNAJC13(N855S), recruitment of the completely assembled WASH complex is a positive modulator of autophagy. Mechanistically, the to endosomes which in turn altered ATG9A trafficking lack of DNAJC13 interferes with the trafficking of ATG9A [59]. We now showed that the knockdown of DNAJC13 to and from the recycling endosome. As a consequence, the caused a similar phenotype than an impairment of localization of ATG9A at LC3-positive-structures under endosomal WASH recruiting: reduced DNAJC13 levels steady-state and autophagy-induced conditions is impaired (1) affected localization of ATG9A at the late and suggesting a functional cross-talk of DNAJC13 with this key recycling endosomes, (2) impaired ATG9A transport from autophagy protein. the recycling endosome upon autophagy stimulation, and RME8/DNAJC13 is primarily localized at the early (3) resulted in a reduced localization of ATG9A with endosome membrane which serves as a switchboard for LC3B under steady state and autophagy-stimulating protein and membrane traffic. RME-8/DNAJC13 affects conditions. It is of note that we additionally observed a retrograde, clathrin-mediated transport from endosomes reduced number of ATG16L1-positive puncta per cells towards the trans-Golgi network [33], endosomal tubula- after pharmacological autophagy induction in DNAJC13 tion [32], and the recycling of membrane receptors via the knockdown cells (summarized in Fig. 5). These data are recycling endosome or through a direct route [36]. RME-8/ consistent with the view that ATG16L1 gets hooked on 1 3 6 56 A. S. Besemer et al. 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 657 ◂Fig. 4 DNAJC13 affects the formation of ATG16L1 puncta and the ATG9A vesicles trafficking from the recycling endosome co-localization of ATG9A with LC3. HEK293A cells transfected towards the site of autophagosome formation [62] and with a nonsense or b DNAJC13 siRNA and treated with rapamycin were stained with antibodies detecting ATG16L1 (green) and DAPI that an impaired ATG9A trafficking from the recycling (blue). Images represent maximum projections of image stacks. The endosome alters ATG16L1 localization at phagophores. lower panels are enlargements of the boxed areas. All cells carrying Several PD-causing mutations are located within genes a single or a double point in close proximity to the nucleus (open like VPS35 [69, 70] and DNAJC13 [39] that facilitate arrows) represent cells with a low autophagic activity; multiple, accu- mulated ATG16L1 dots (closed arrows) reflect a high autophagic endosomal protein trafficking. Increasing evidence sug- activity (scale bar: 10 µm). c In average, dots in 60 cells per condi- gests that mutant DNAJC13 expression results in an tion and experiment were counted (n = 4; box-plot; one-way ANOVA altered endosomal protein transport which affects also with Bonferroni correction: *p ≤ 0.05; ***p ≤ 0.001). d, e HeLa cells protein degradation. As an example, the expression of the were transfected with nonsense (nons) or siRNA against DNAJC13 and GFP-LC3B (in green) and mCherry-ATG9A (in red). Representa- DNAJC13(N855S) variant causes defects in SNX1 mem- tive deconvoluted images of z-stacks are shown without infection brane dynamics [40] and actin cytoskeleton organization (d) and 5  h post-HPV pseudovirion addition (e). DNA was labeled which is reminiscent of an disturbed WASH function with Hoechst and is shown in blue (scale bar: 10 µm). f Analysis of [41]. In the latter, altered endosomal trafficking resulted GFP-LC3B and mCherry-ATG9A resulted in a decreased co-local- ization of pixels upon knockdown of DNAJC13. The infection with in the accumulation of α-synuclein. In line with these HPV virions resulted in an increased co-localization of ATG9A and experiments, we do not observe a significant activation of LC3B and the knockdown of DNAJC13 affects this co-localization. autophagic activity upon DNAJC13(N855S) expression About eight images per condition were analyzed in three independ- compared to DNAJC13(WT) overexpression. This and the ent experiments each (about 100 cells in total). Values were nor- malized to nonsense siRNA transfected cells without virus (n = 24; above-mentioned data indicate a dominant-negative mode mean ± SEM; Wilcoxon rank-sum test: *p ≤ 0.05). g HeLa cells of action of mutant DNAJC13(N855S) causing late-onset treated with nonsense or DNAJC13 siRNA were infected with HPV Parkinson’s disease. These data are consistent with the virions containing a luciferase expression plasmid. Only cells in view that the chronic expression of factors challenging the which the plasmid reached the nucleus expressed luciferase. Rela- tive infection was measured by luciferase activity and normalized by autophagic pathway, like the expression of mutant variants LDH measurements. Control siRNA infection rate was set to 100% of DNAJC13, results in increased proteotoxic stress that in (n = 4; mean ± SD; t test: ***p ≤ 0.001). h DNAJC13 protein levels turn significantly contributes to the development of late- were determined after transfection of HeLa cells with nonsense or onset neurodegenerative diseases. DNAJC13 siRNA by Western blot. Tubulin served as a loading con- trol Fig. 5 Proposed mode of action of DNAJC13 in autophagy by interfering with ATG9A trafficking. See details in the “Results” and “Discussion” section (W Wash complex, R retromer complex, S sorting nexin dimer) 1 3 658 A. S. Besemer et al. Acknowledgements Open Access funding provided by Projekt DEAL. 8. Mercer TJ, Gubas A, Tooze SA (2018) A molecular perspec- We greatly appreciate the gift of reagents from M. Farrer (Vancouver, tive of mammalian autophagosome biogenesis. J Biol Chem Canada), B. Levine (Dallas, USA), P. McPherson (Montreal, Canada), 293:5386–5395 S. Nagata (Kyoto, Japan), K. Sekiguchi (Osaka, Japan), I. Dikic, G. 9. Abada A, Elazar Z (2014) Getting ready for building: signaling Auburger (both Frankfurt, Germany), M. I. Colombo (Mendoza, and autophagosome biogenesis. EMBO Rep 15:839–852 Argentina), and A. Helenius (Zurich, Switzerland). C. Pietrzik (Mainz, 10. Hayashi-Nishino M, Fujita N, Noda T, Yamaguchi A, Yoshimori Germany) helped with the generation of Luc-GFP cell lines. We thank T, Yamamoto A (2009) A subdomain of the endoplasmic reticu- T. Benz for excellent technical assistance. lum forms a cradle for autophagosome formation. Nat Cell Biol 11:1433–1437 Author contributions AMC, CB, AK, and LF designed experiments 11. Hailey DW, Rambold AS, Satpute-Krishnan P, Mitra K, Sougrat and supervised data analysis; MDAA, HH, and CF performed C. ele- R, Kim PK, Lippincott-Schwartz J (2010) Mitochondria supply gans experiments; ASB, JM, AS, SV, KN, CvH, IKD, IFP, BS, and membranes for autophagosome biogenesis during starvation. AMC performed cell culture experiments and analyzed data; FB and PLoS ONE 141:656–667 LF executed and analyzed HPV experiments; AMC and CB planned 12. Hamasaki M, Furuta N, Matsuda A, Nezu A, Yamamoto A, Fujita and supervised the project and wrote the manuscript. N, Oomori H, Noda T, Haraguchi T, Hiraoka Y et al (2013) Autophagosomes form at ER–mitochondria contact sites. Nature Funding This work was supported in part by the CRC1080 “Neuronal 495:389–393 homeostasis” (AMC, CB), the CRC1177 “Selective autophagy” (CB), 13. Ohashi Y, Munro S (2010) Membrane delivery to the yeast and the Grant FL 696/2-1 (LF) (all DFG). ASB, JM, and AS were autophagosome from the Golgi-endosomal system. Mol Biol Cell fellows of the Focus Program Translational Neuroscience of the Uni- 21:3998–4008 versity of Mainz (FTN). 1 4. Bekbulat F, Schmitt D, Feldmann A, Huesmann H, Eimer S, Juretschke T, Beli P, Behl C, Kern A (2020) RAB18 loss interferes with lipid droplet catabolism and provokes autophagy network Compliance with ethical standards adaptations. J Mol Biol 432:1216–1234 15. Shpilka T, Elazar Z (2015) Lipid droplets regulate autophagosome Conflict of interest The authors report no conflict of interest. biogenesis. Autophagy 11:2130–2131 16. Kern A, Dikic I, Behl C (2015) The integration of autophagy This article is licensed under a Creative Commons Attri- and cellular trafficking pathways via RAB GAPs. Autophagy Open Access bution 4.0 International License, which permits use, sharing, adapta- 11:2393–2397 tion, distribution and reproduction in any medium or format, as long 17. Noda T, Kim J, Huang WP, Baba M, Tokunaga C, Ohsumi Y, as you give appropriate credit to the original author(s) and the source, Klionsky DJ (2000) Apg9p/Cvt7p is an integral membrane protein provide a link to the Creative Commons licence, and indicate if changes required for transport vesicle formation in the Cvt and autophagy were made. The images or other third party material in this article are pathways. J Cell Biol 148:465–480 included in the article’s Creative Commons licence, unless indicated 1 8. Orsi A, Razi M, Dooley HC, Robinson D, Weston AE, Collinson otherwise in a credit line to the material. If material is not included in LM, Tooze SA (2012) Dynamic and transient interactions of Atg9 the article’s Creative Commons licence and your intended use is not with autophagosomes, but not membrane integration, are required permitted by statutory regulation or exceeds the permitted use, you will for autophagy. Mol Biol Cell 23:1860–1873 need to obtain permission directly from the copyright holder. To view a 19. Corcelle-Termeau E, Vindelov SD, Hamalisto S, Mograbi B, copy of this licence, visit http://creati vecom mons. org/licens es/by/4.0/. Keldsbo A, Brasen JH, Favaro E, Adam D, Szyniarowski P, Hof-man P et al (2016) Excess sphingomyelin disturbs ATG9A traf- ficking and autophagosome closure. Autophagy 12:833–849 20. Imai K, Hao F, Fujita N, Tsuji Y, Oe Y, Araki Y, Hamasaki M, References Noda T, Yoshimori T (2016) Atg9A trafficking through the recy- cling endosomes is required for autophagosome formation. J Cell 1. Kaushik S, Cuervo AM (2015) Proteostasis and aging. Nat Med Sci 129:3781–3791 21:1406–1415 21. Longatti A, Lamb CA, Razi M, Yoshimura S, Barr FA, Tooze SA 2. Labbadia J, Morimoto RI (2015) The biology of proteostasis in (2012) TBC1D14 regulates autophagosome formation via Rab11- aging and disease. Annu Rev Biochem 84:435–464 and ULK1-positive recycling endosomes. J Cell Biol 197:659–675 3. Klaips CL, Jayaraj GG, Hartl FU (2018) Pathways of cellular 22. Puri C, Renna M, Bento CF, Moreau K, Rubinsztein DC (2013) proteostasis in aging and disease. J Cell Biol 217:51–63 Diverse autophagosome membrane sources coalesce in recycling 4. Hansen M, Rubinsztein DC, Walker DW (2018) Autophagy as a endosomes. PLoS ONE 154:1285–1299 promoter of longevity: insights from model organisms. Nat Rev 2 3. Soreng K, Munson MJ, Lamb CA, Bjorndal GT, Pankiv S, Carls- Mol Cell Biol 19:579–593 son SR, Tooze SA, Simonsen A (2018) SNX18 regulates ATG9A 5. Khaminets A, Behl C, Dikic I (2016) Ubiquitin-dependent and trafficking from recycling endosomes by recruiting Dynamin-2. independent signals in selective autophagy. Trends Cell Biol EMBO Rep 19:e44837 26:6–16 2 4. Takahashi Y, Tsotakos N, Liu Y, Young MM, Serfass J, Tang Z, 6. Menzies FM, Fleming A, Rubinsztein DC (2015) Compromised Abraham T, Wang HG (2016) The Bif-1-Dynamin 2 membrane autophagy and neurodegenerative diseases. Nat Rev Neurosci fission machinery regulates Atg9-containing vesicle generation at 16:345–357 the Rab11-positive reservoirs. Oncotarget 7:20855–20868 7. Ravikumar B, Vacher C, Berger Z, Davies JE, Luo S, Oroz LG, 25. Zhang Y, Grant B, Hirsh D (2001) RME-8, a conserved J-domain Scaravilli F, Easton DF, Duden R, O’Kane CJ et al (2004) Inhibi- protein, is required for endocytosis in Caenorhabditis elegans. tion of mTOR induces autophagy and reduces toxicity of polyglu- Mol Biol Cell 12:2011–2021 tamine expansions in fly and mouse models of Huntington disease. 26. Fujibayashi A, Taguchi T, Misaki R, Ohtani M, Dohmae N, Nat Genet 36:585–595 Takio K, Yamada M, Gu J, Yamakami M, Fukuda M et al (2008) Human RME-8 is involved in membrane trafficking through early endosomes. Cell Struct Funct 33:35–50 1 3 Receptor-mediated endocytosis 8 (RME-8)/DNAJC13 is a novel positive modulator of autophagy… 659 27. Girard M, Poupon V, Blondeau F, McPherson PS (2005) The 4 6. Link CD (1995) Expression of human beta-amyloid peptide in DnaJ-domain protein RME-8 functions in endosomal trafficking. transgenic Caenorhabditis elegans. Proc Natl Acad Sci USA J Biol Chem 280:40135–40143 92:9368–9372 28. Silady RA, Kato T, Lukowitz W, Sieber P, Tasaka M, Somerville 47. Wu Y, Wu Z, Butko P, Christen Y, Lambert MP, Klein WL, Link CR (2004) The gravitropism defective 2 mutants of Arabidopsis CD, Luo Y (2006) Amyloid-beta-induced pathological behaviors are deficient in a protein implicated in endocytosis in Caenorhab- are suppressed by Ginkgo biloba extract EGb 761 and ginkgolides ditiselegans. Plant Physiol 136:3095–4103 discussion 3002 in transgenic Caenorhabditis elegans. J Neurosci 26:13102–13113 29. Chang HC, Hull M, Mellman I (2004) The J-domain protein 48. Liebl MP, Kaya AM, Tenzer S, Mittenzwei R, Koziollek-Drechsler Rme-8 interacts with Hsc70 to control clathrin-dependent endo- I, Schild H, Moosmann B, Behl C, Clement AM (2014) Dimeri- cytosis in Drosophila. J Cell Biol 164:1055–1064 zation of visinin-like protein 1 is regulated by oxidative stress 30. Xhabija B, Vacratsis PO (2015) Receptor-mediated endocytosis 8 and calcium and is a pathological hallmark of amyotrophic lateral utilizes an N-terminal phosphoinositide-binding motif to regulate sclerosis. Free Radic Biol Med 72:41–54 endosomal clathrin dynamics. J Biol Chem 290:21676–21689 49. Isbert S, Wagner K, Eggert S, Schweitzer A, Multhaup G, Weggen 3 1. Girard M, McPherson PS (2008) RME-8 regulates trafficking of S, Kins S, Pietrzik CU (2012) APP dimer formation is initiated the epidermal growth factor receptor. FEBS Lett 582:961–966 in the endoplasmic reticulum and differs between APP isoforms. 32. Freeman CL, Hesketh G, Seaman MN (2014) RME-8 coordinates Cell Mol Life Sci 69:1353–1375 the activity of the WASH complex with the function of the ret- 5 0. Schneider MA, Scheffer KD, Bund T, Boukhallouk F, Lambert romer SNX dimer to control endosomal tubulation. J Cell Sci C, Cotarelo C, Pflugfelder GO, Florin L, Spoden GA (2013) The 127:2053–2070 transcription factors TBX2 and TBX3 interact with human papil- 3 3. Popoff V, Mardones GA, Bai SK, Chambon V, Tenza D, Burgos lomavirus 16 (HPV16) L2 and repress the long control region of PV, Shi A, Benaroch P, Urbe S, Lamaze C et al (2009) Analysis HPVs. J Virol 87:4461–4474 of articulation between clathrin and retromer in retrograde sorting 5 1. Schneider MA, Spoden GA, Florin L, Lambert C (2011) Identi- on early endosomes. Traffic 10:1868–1880 fication of the dynein light chains required for human papilloma- 3 4. Shi A, Sun L, Banerjee R, Tobin M, Zhang Y, Grant BD (2009) virus infection. Cell Microbiol 13:32–46 Regulation of endosomal clathrin and retromer-mediated endo- 5 2. Gispert S, Del Turco D, Garrett L, Chen A, Bernard DJ, Hamm- some to Golgi retrograde transport by the J-domain protein RME- Clement J, Korf HW, Deller T, Braak H, Auburger G et al (2003) 8. EMBO J 28:3290–3302 Transgenic mice expressing mutant A53T human alpha-synuclein 35. Norris A, Tammineni P, Wang S, Gerdes J, Murr A, Kwan KY, show neuronal dysfunction in the absence of aggregate formation. Cai Q, Grant BD (2017) SNX-1 and RME-8 oppose the assembly Mol Cell Neurosci 24:419–429 of HGRS-1/ESCRT-0 degradative microdomains on endosomes. 53. Witan H, Gorlovoy P, Kaya AM, Koziollek-Drechsler I, Neumann Proc Natl Acad Sci USA 114:E307–E316 H, Behl C, Clement AM (2009) Wild-type Cu/Zn superoxide dis- 36. Gomez-Lamarca MJ, Snowdon LA, Seib E, Klein T, Bray SJ mutase (SOD1) does not facilitate, but impedes the formation of (2015) Rme-8 depletion perturbs Notch recycling and predisposes protein aggregates of amyotrophic lateral sclerosis causing mutant to pathogenic signaling. J Cell Biol 210:303–318 SOD1. Neurobiol Dis 36:331–342 3 7. McMillan KJ, Korswagen HC, Cullen PJ (2017) The emerging 5 4. Savina A, Fader CM, Damiani MT, Colombo MI (2005) Rab11 role of retromer in neuroprotection. Curr Opin Cell Biol 47:72–82 promotes docking and fusion of multivesicular bodies in a cal- 3 8. Vagnozzi AN, Pratico D (2019) Endosomal sorting and traffick- cium-dependent manner. Traffic 6:131–143 ing, the retromer complex and neurodegeneration. Mol Psychiatry 55. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, 24:857–868 Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B et al 39. Vilarino-Guell C, Rajput A, Milnerwood AJ, Shah B, Szu-Tu C, (2012) Fiji: an open-source platform for biological-image analysis. Trinh J, Yu I, Encarnacion M, Munsie LN, Tapia L et al (2014) Nat Methods 9:676–682 DNAJC13 mutations in Parkinson disease. Hum Mol Genet 56. Witan H, Kern A, Koziollek-Drechsler I, Wade R, Behl C, Clem- 23:1794–1801 ent AM (2008) Heterodimer formation of wild-type and amyo- 4 0. Follett J, Fox JD, Gustavsson EK, Kadgien C, Munsie LN, Cao trophic lateral sclerosis-causing mutant Cu/Zn-superoxide dis- LP, Tatarnikov I, Milnerwood AJ, Farrer MJ (2019) DNAJC13 mutase induces toxicity independent of protein aggregation. Hum p.Asn855Ser, implicated in familial parkinsonism, alters mem- Mol Genet 17:1373–1385 brane dynamics of sorting nexin 1. Neurosci Lett 706:114–122 57. Bruijn LI, Becher MW, Lee MK, Anderson KL, Jenkins NA, 41. Yoshida S, Hasegawa T, Suzuki M, Sugeno N, Kobayashi J, Uey- Copeland NG, Sisodia SS, Rothstein JD, Borchelt DR, Price DL ama M, Fukuda M, Ido-Fujibayashi A, Sekiguchi K, Ezura M et al et al (1997) ALS-linked SOD1 mutant G85R mediates damage to (2018) Parkinson’s disease-linked DNAJC13 mutation aggravates astrocytes and promotes rapidly progressive disease with SOD1- alpha-synuclein-induced neurotoxicity through perturbation of containing inclusions. Neuron 18:327–338 endosomal trafficking. Hum Mol Genet 27:823–836 5 8. Lazaro DF, Rodrigues EF, Langohr R, Shahpasandzadeh H, 42. Roosen DA, Blauwendraat C, Cookson MR, Lewis PA (2019) Ribeiro T, Guerreiro P, Gerhardt E, Krohnert K, Klucken J, DNAJC proteins and pathways to parkinsonism. FEBS J Pereira MD et al (2014) Systematic comparison of the effects of 286:3080–3094 alpha-synuclein mutations on its oligomerization and aggregation. 4 3. Kern A, Spang N, Huesmann H, Behl C (2018) Novel modulators PLoS Genet 10:e1004741 of proteostasis: RNAi screen of chromosome I in a heat stress 59. Zavodszky E, Seaman MN, Moreau K, Jimenez-Sanchez M, paradigm in C. elegans. Cells 7:E49 Breusegem SY, Harbour ME, Rubinsztein DC (2014) Mutation 44. Spang N, Feldmann A, Huesmann H, Bekbulat F, Schmitt V, Hie- in VPS35 associated with Parkinson’s disease impairs WASH bel C, Koziollek-Drechsler I, Clement AM, Moosmann B, Jung complex association and inhibits autophagy. Nat Commun 5:3828 J et al (2014) RAB3GAP1 and RAB3GAP2 modulate basal and 6 0. Kaizuka T, Morishita H, Hama Y, Tsukamoto S, Matsui T, Toyota rapamycin-induced autophagy. Autophagy 10:2297–2309 Y, Kodama A, Ishihara T, Mizushima T, Mizushima N (2016) An 45. Kern A, Ackermann B, Clement AM, Duerk H, Behl C (2010) autophagic flux probe that releases an internal control. Mol Cell HSF1-controlled and age-associated chaperone capacity in neu- 64:835–849 rons and muscle cells of C. elegans. PLoS ONE 5:e8568 6 1. Young AR, Chan EY, Hu XW, Kochl R, Crawshaw SG, High S, Hailey DW, Lippincott-Schwartz J, Tooze SA (2006) Starvation 1 3 660 A. S. Besemer et al. and ULK1-dependent cycling of mammalian Atg9 between the 68. McGough IJ, Steinberg F, Jia D, Barbuti PA, McMillan KJ, Hee- TGN and endosomes. Free Radic Biol Med 119:3888–3900 som KJ, Whone AL, Caldwell MA, Billadeau DD, Rosen MK et al 6 2. Puri C, Renna M, Bento CF, Moreau K, Rubinsztein DC (2014) (2014) Retromer binding to FAM21 and the WASH complex is ATG16L1 meets ATG9 in recycling endosomes: additional roles perturbed by the Parkinson disease-linked VPS35(D620N) muta- for the plasma membrane and endocytosis in autophagosome bio- tion. Curr Biol 24:1670–1676 genesis. Autophagy 10:182–184 6 9. Vilarino-Guell C, Wider C, Ross OA, Dachsel JC, Kachergus JM, 63. Mizushima N, Kuma A, Kobayashi Y, Yamamoto A, Matsubae Lincoln SJ, Soto-Ortolaza AI, Cobb SA, Wilhoite GJ, Bacon JA M, Takao T, Natsume T, Ohsumi Y, Yoshimori T (2003) Mouse et al (2011) VPS35 mutations in Parkinson disease. Am J Hum Apg16L, a novel WD-repeat protein, targets to the autophagic Genet 89:162–167 isolation membrane with the Apg12–Apg5 conjugate. J Cell Sci 70. Zimprich A, Benet-Pages A, Struhal W, Graf E, Eck SH, Offman 116:1679–1688 MN, Haubenberger D, Spielberger S, Schulte EC, Lichtner P et al 6 4. Ishii Y (2013) Electron microscopic visualization of autophago- (2011) A mutation in VPS35, encoding a subunit of the retromer somes induced by infection of human papillomavirus pseudoviri- complex, causes late-onset Parkinson disease. Am J Hum Genet ons. Biochem Biophys Res Commun 433:385–389 89:168–175 6 5. Mukadam AS, Seaman MN (2015) Retromer-mediated endoso- 7 1. Klionsky DJ, Abdelmohsen K, Abe A, Abedin MJ, Abeliovich H, mal protein sorting: the role of unstructured domains. FEBS Lett Acevedo Arozena A, Adachi H, Adams CM, Adams PD, Adeli K 589:2620–2626 et al (2016) Guidelines for the use and interpretation of assays for 66. Seaman MN, Gautreau A, Billadeau DD (2013) Retromer-medi- monitoring autophagy (3rd edition). Autophagy 12:1–222 ated endosomal protein sorting: all WASHed up! Trends Cell Biol 23:522–528 Publisher’s Note Springer Nature remains neutral with regard to 67. Chen KE, Healy MD, Collins BM (2019) Towards a molecu- jurisdictional claims in published maps and institutional affiliations. lar understanding of endosomal trafficking by Retromer and Retriever. Traffic 20:465–478 1 3