A conserved glycine residue in the C-terminal region of human ATG9A is required for its transport from the endoplasmic reticulum to the Golgi apparatus

Biochemical and Biophysical Research Communications 479 (2016) 404e409

Contents lists available at ScienceDirect

Biochemical and Biophysical Research Communications journal homepage: www.elsevier.com/locate/ybbrc

A conserved glycine residue in the C-terminal region of human ATG9A is required for its transport from the endoplasmic reticulum to the Golgi apparatus Catherine Staudt, Florentine Gilis, Virginie Tevel, Michel Jadot, Marielle Boonen* URPhyM-Laboratoire de Chimie Physiologique, University of Namur, 61 Rue de Bruxelles, 5000 Namur, Belgium

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 September 2016 Accepted 19 September 2016 Available online 20 September 2016

ATG9A is the only polytopic protein of the mammalian autophagy-related protein family whose members regulate autophagosome formation during macroautophagy. At steady state, ATG9A localizes to several intracellular sites, including the Golgi apparatus, endosomes and the plasma membrane, and it redistributes towards autophagosomes upon autophagy induction. Interestingly, the transport of yeast Atg9 to the pre-autophagosomal structure depends on its self-association, which is mediated by a short amino acid motif located in the C-terminal region of the protein. Here, we investigated whether the residues that align with this motif in human ATG9A (V515-C519) are also required for its trafficking in mammalian cells. Interestingly, our findings support that human ATG9A self-interacts as well, and that this process promotes transport of ATG9A molecules through the Golgi apparatus. Furthermore, our data reveal that the transport of ATG9A out of the ER is severely impacted after mutation of the conserved V515-C519 motif. Nevertheless, the mutated ATG9A molecules could still interact with each other, indicating that the molecular mechanism of self-interaction differs in mammalian cells compared to yeast. Using sequential amino acid substitutions of glycine 516 and cysteine 519, we found that the stability of ATG9A relies on both of these residues, but that only the former is required for efficient transport of human ATG9A from the endoplasmic reticulum to the Golgi apparatus. © 2016 Elsevier Inc. All rights reserved.

Keywords: ATG9A Subcellular trafficking Endoplasmic reticulum Self-interaction

1. Introduction In macroautophagy, a portion of the cell cytoplasm (including damaged organelles and/or cytosolic macromolecules) is engulfed within an autophagosome and degraded upon fusion of this structure with a lysosome. Many autophagy-related (ATG) proteins have been identified [1,2]. They mediate the formation, elongation and fusion of the phagophore, i.e. the elongation membrane that progressively surrounds cytoplasmic components to form an autophagosome. Interestingly, the only transmembrane proteins of this family are Atg27 in yeast, and Atg9/ATG9A in both yeast and mammalian cells. Mammalian ATG9A cycles between endosomes and the trans-Golgi network (TGN) under basal conditions and, upon autophagy induction, is recruited on autophagosomes, but not inserted into their membrane [3e5]. Albeit the precise molecular function of ATG9A is not yet fully understood, it has been

* Corresponding author. E-mail address: [email protected] (M. Boonen). http://dx.doi.org/10.1016/j.bbrc.2016.09.097 0006-291X/© 2016 Elsevier Inc. All rights reserved.

reported that its knockdown impairs the macroautophagy process at an early stage, and that ATG9A knockout mice die during the neonatal starvation period [3,4,6,7]. Intriguingly, the finding that MAP kinase P38a controls the transport of ATG9A between the TGN and peripheral sites (e.g. endosomes), and that disruption of this regulatory mechanism impairs autophagosome formation, has led to the proposal that the autophagic pathway depends on ATG9A trafficking [8]. This assumption is further supported by the observation that inactivation or overexpression of several other proteins that control autophagosome formation upon autophagy induction impacts the subcellular trafficking of ATG9A. These include the clathrin adaptor protein AP-2, Bif-1 (a membrane curvatureinducing protein), Dynamin 2, Beclin 1 (mammalian ortholog of yeast Atg6), UVRAG (UV radiation resistance-associated gene protein), TBC1D14 (Tre-BubCDC16 domain-containing protein) and TRAPPC8 (mammalian orthologue of the yeast autophagy-specific TRAPP subunit Trs85) [9e13]. Mutational analyses have revealed that several intramolecular determinants are involved in the subcellular trafficking of Atg9/ ATG9A. For instance, we recently demonstrated that anterograde

C. Staudt et al. / Biochemical and Biophysical Research Communications 479 (2016) 404e409

transport of newly synthesized human ATG9A molecules through the Golgi apparatus is dependent upon a conserved LYM motif located in its C-terminal region [14]. Indeed, we showed that mutation of this motif does not impact the sorting of ATG9A from the endoplasmic reticulum (ER), nor its arrival in the cis-Golgi, but largely reduces the amount of ATG9A proteins that can proceed through the medial- and trans-Golgi cisternae. Following up on the very intriguing observation that this defect does not prevent ATG9A from reaching endosomes and the trans-Golgi, we found that ATG9A can travel to these sites by an atypical route, bypassing the Golgi apparatus to reach the plasma membrane. From there ATG9A can be endocytosed and subsequently transported to the TGN [14]. The endocytosis process is mediated, at least partly, by clathrin and the clathrin adaptor protein complex AP-2 [11,15], and it has recently been shown that mutation of the YQRL and LLV motifs located in the N-terminal region of ATG9A decreases Golgi residency and binding to AP-2 [16]. In Pichia pastoris, transport of Atg9 from a peripheral ER-like compartment is prevented by deletion of its third loop [17]. Similarly, deletion of 15 amino acids within this region in human ATG9A (DL340-L354) results in the retention of the protein in the endoplasmic reticulum and extensive degradation by the proteasomal system [14]. Lastly, a self-interaction motif has been identified in the C-terminal region of yeast Atg9 [18]. Deletion of amino acids L766 to C770, or mutation of the most conserved residues G767 and C770 within this sequence severely impairs the self-assembly and subcellular trafficking of the protein to the pre-autophagosomal structure in yeast. This conserved motif is found in the C-terminal region of human ATG9A, at position 515e519 (V515GDTC519, referred to as V515-C519 hereafter). However, the role of this sequence in mammalian cells has not been investigated so far. Here, we used a combination of biochemical and fluorescence microscopy methods to investigate whether amino acids V515-C519 control the intracellular trafficking of ATG9A in human cells. 2. Material and methods 2.1. Material Unless otherwise specified, chemicals were obtained from Sigma-Aldrich, as well as mouse anti-GAPDH, mouse anti-Tubulin, and rabbit anti-Hemagglutinin (HA) antibodies. Mouse and rabbit anti-MYC antibodies were bought from Cell Signaling and rabbit anti-ATG9A from Novus Biologicals. Rabbit anti-VPS26 and antiCalnexin were obtained from Abcam, mouse anti-GM130 from BD Biosciences and mouse anti-GOLGIN97 from Santa Cruz. Rabbit antiSEC61b was gifted by Bernhard Dobberstein's group (ZMBH, Heidelberg). HRP (Horseradish Peroxidase)-coupled and Alexa Fluor™coupled secondary antibodies were obtained from DAKO and Life Technologies, respectively. Cell culture media, trypsin and antibiotics were obtained from LONZA, and fetal bovine serum from Sigma. The pcDNA3.1(þ) construct containing the cDNA encoding wild-type human ATG9A fused to a MYC tag has been described previously [14]. Deletion of the codons coding for amino acids 515e519, and substitutions of the codons coding for glycine 516 and cysteine 519 by alanine codons were introduced in the wild-type cDNA using classical molecular biology methods. HA-tagged constructs of ATG9A (wild-type or mutated) were obtained by insertion of a linker coding for HA and a STOP codon via the NotI restriction site located at the C-terminal end of ATG9A in pcDNA3.1(þ). 2.2. Experimental procedures All manipulations (cell culture, transfection, western blotting,

405

immunofluorescence and co-immunoprecipitation assays) were conducted as described in Staudt et al., 2016 [14], except that statistical significance of all Pearson's correlation analyses were assessed using non-paired Student's t-test. 3. Results 3.1. Deletion of amino acids V515-C519 in human ATG9A, or dual substitution of G516 and C519 by alanine residues, severely impairs the transport of ATG9A from the ER to the Golgi apparatus When expressed in HeLa cells, two forms of ATG9A are visualized by western blotting: one form of approximately 92 kDa, and one form of approximately 104 kDa. Using pulse/chase labelling and endoglycosidase treatments, we have documented previously that the former is synthesized first and that it harbors a high-mannose N-linked glycan. This oligosaccharidic chain is then further converted into a complex-type glycan upon passage through the Golgi apparatus, which raises the apparent molecular mass of the protein to ~104 kDa [14]. To investigate whether the conserved sequence that drives the self-association and trafficking of yeast Atg9 (L766GYVC770) is involved in the transport of ATG9A in mammalian cells, we first deleted the corresponding amino acids, V515GDTC519, in a MYC-tagged human ATG9A (Fig. 1A, DV515-C519). 24 h after transfection of HeLa cells, the wild-type and mutated proteins were detected by western blotting, using an anti-MYC antibody. Interestingly, this analysis revealed that the DV515-C519 mutant exhibits a lower level of expression compared to the wild-type protein, and that it does not acquire a complex-type oligosaccharide (Fig. 1B, only the 92 kDa form is detected for the mutated protein). This finding indicates that DV515-C519 fails to travel through the Golgi apparatus. Similar results were obtained after replacement of the two most conserved residues of this motif, glycine 516 and cysteine 519, by alanine residues (G516A/C519A, Fig. 1AeB). In order to identify the underlying cause of this failure to acquire a complex N-glycan in the Golgi apparatus, we used immunofluorescence and the Pearson's correlation method to characterize the subcellular localization of the G516A/C519A mutant relative to the wild-type protein. Fig. 1C shows that these substitutions induce the accumulation of ATG9A in the ER. Indeed, while the wild-type protein was detected in endosomes (r ¼ 0.46 ± 0.02 with the endosomal marker VPS26), to some extent in the cis-Golgi (r ¼ 0.22 ± 0.02 with GM130), as well as in the TGN (r ¼ 0.58 ± 0.02 with GOLGIN97), it poorly co-localized with the ER marker calnexin (0.10 ± 0.01) (Fig. 1C, data obtained from Staudt et al., 2016). By contrast, a large overlap was detected between this ER marker protein and the G516A/ C519A mutant (r ¼ 0.59 ± 0.02). Consistently, lower co-localization levels were detected between the mutant protein and endosomal, cis- and trans-Golgi markers (p < 0.0001 when compared to the wild-type protein using a non-paired Student's t-test, n ¼ 30 cell images, collected in 3 independent experiments). These observations demonstrate that the G516A/C519A mutations severely impair the sorting of ATG9A to post-ER compartments and its transport to the Golgi apparatus. 3.2. Alanine substitutions of glycine 516 and cysteine 519 do not prevent the oligomerization of human ATG9A Interestingly, transport of ATG9A-G516A/C519A through the Golgi apparatus could be restored, at least to some extent, when the mutated protein (MYC tagged) was co-expressed with wild-type ATG9A (HA tagged). This was reflected by the acquisition of a complex-type N-glycan on the mutated protein, indicative of its passage through the medial-Golgi (Fig. 2A, arrowhead). Of note, this was not observed when two mutated proteins, HA- or MYC-tagged,

406

C. Staudt et al. / Biochemical and Biophysical Research Communications 479 (2016) 404e409

Fig. 1. Deletion of amino acids V515GDTC519 or substitution of G516 and C519 by alanine residues impairs anterograde trafficking of ATG9A to the Golgi apparatus. A. Schematic representation of the mutations introduced in the C-terminal region of human ATG9A, fused to a C-terminal MYC tag. B. cDNA constructs encoding the following ATG9A-MYC proteins were transiently expressed in HeLa cells: wild-type ATG9A, ATG9A missing amino acids V515GDTC519 (DV515-C519), or ATG9A containing alanine residues at G516 and C519 positions (G516A/C519A). Twenty-four hours post-transfection, the cells were lysed and the proteins of interest were detected by western blotting using an anti-MYC antibody.

C. Staudt et al. / Biochemical and Biophysical Research Communications 479 (2016) 404e409

407

oligomerization. 3.3. The stability of ATG9A and its sorting from the ER are dependent upon glycine 516, whereas mutation of cysteine 519 only affects protein stability The results presented in Fig. 1B show that lower levels of DV515C and of G516A/C519A (double mutant) were detected in HeLa cells, compared to the wild-type protein. This could reflect degradation by the proteasomal system after synthesis in the ER, due to protein instability. In accordance with this hypothesis, Fig. 3 shows that the intracellular levels of the mutated proteins (both DV515C519 and G516A/C519A) largely increased after treatment with MG132, a proteasomal inhibitor, indicating that they are indeed extensively degraded by this system (Fig. 3, left panel). Lastly, to better comprehend the role of glycine 516 and cysteine 519 in the transport of ATG9A from the ER to the Golgi apparatus, we mutated these residues separately. It is well-known that palmitoylation of cysteine residues regulates the subcellular trafficking of many proteins, including transmembrane proteins [19]. Intriguingly, though, whereas both mutants (C519A and G516A) accumulated in the presence of MG132, indicative of their degradation by the proteasomal system, only G516A failed to travel through the medial Golgi (Fig. 3, right panel, a complex-type oligosaccharidic chain is detected on C519A, but not on G516A). Of note, the strong impact of MG132 on the maturation of the C519A mutant might be caused by the association/entrapment of the Golgi-bound ATG9A molecules with the large amount of incorrectly folded ATG9A proteins that accumulate in the ER in the presence of the drug. Taken together, these results indicate that glycine 516 and cysteine 519 are both required for proper folding of ATG9A in the ER. However, glycine 516 only is critical for transport to the Golgi apparatus. 519

Fig. 2. Human ATG9A molecules oligomerize independently of residues G516 and C519, and can be transported from the ER to the Golgi apparatus as a complex. A. Detection of MYC- or HA-tagged ATG9A (wild-type [wt] or G516A/C519A), by western blotting after co-transfection of the indicated constructs in HeLa cells for 24 h. The arrowhead shows the acquisition of complex N-glycans on the mutated proteins when co-expressed with wt proteins. B. Immunoprecipitation (IP) of ATG9A-MYC (wt or G516A/C519A) from HeLa cells transiently co-transfected (for 24 h) with the indicated MYC- or HA-tagged ATG9A constructs, using a mouse anti-MYC antibody. Precipitated proteins were detected by western blotting using rabbit anti-MYC or anti-HA antibodies. Input: 4% of the starting sample. A non-relevant antibody (mouse IgG) was used as a control (Ctl) of nonspecific protein immunoprecipitation and absorption on the protein G agarose beads. Arrowheads show that wild-type and mutated ATG9A forms that have acquired complex N-glycans co-immunoprecipitate. IB, immunoblotting; IP, immunoprecipitation.

were co-expressed (Fig. 2A, middle lane). To explain this finding, we postulated that mutated ATG9A molecules could oligomerize with wild-type proteins, and that this process promoted their exit from the ER. In accordance with this view, MYC-tagged G516A/C519A mutants and HA-tagged wild-type proteins could be coimmunoprecipitated (Fig. 2B, right panel). Similarly, MYC-tagged G516A/C519A could pull down G516A/C519A mutant proteins tagged with HA (Fig. 2B, middle panel). Importantly, the mutated proteins, which acquired complex N-glycans when co-expressed with wildtype proteins, could co-immunoprecipitate wild-type proteins bearing complex type N-glycans as well (Fig. 2B, arrowheads in right panels). This observation indicates that, when transport of the mutated protein through the Golgi is promoted by its association with wild-type proteins, at least some of these proteins remain associated in post-ER compartments. Taken together, these results highlight that, similarly to yeast Atg9, newly synthesized human ATG9A molecules oligomerize. However, the molecular mechanism(s) supporting this oligomerization process differ in mammalian cells as, by contrast to yeast Atg9, mutation of the conserved glycine and cysteine residues of the V515-C519 motif does not prevent

4. Discussion In yeast, self-interaction/association of Atg9 in a multiprotein complex is critical for transport of Atg9 to the pre-autophagosomal structure [18]. Our findings demonstrate that human ATG9A proteins self-interact in mammalian cells as well, and that this process promotes the anterograde transport of ATG9A molecules from the ER to the Golgi apparatus (present work and Staudt et al., 2016 [14]). However, our results reveal that mutation of the residues that support self-association of Atg9 in yeast (corresponding to G516 and C519 in the C-terminal region of human ATG9A) does not prevent self-association in human cells. Human ATG9A only shares ~30% amino acid identity with Atg9 from Saccharomyces cerevisiae, and exhibits important differences of structure with the yeast protein. For example, the N-terminal domain is much shorter in human ATG9A, and its C-terminal region contains ~90 additional amino acids compared to its yeast ortholog. The regions that differ in the human protein may contain alternative or additional selfinteraction determinants. It is not uncommon that a combination of motifs located in cytosolic, luminal or transmembrane domains participate to the oligomerization of membrane proteins [20e22]. It is also worth considering that other factors could support selfinteraction of ATG9A in mammalian cells, such as the composition and/or organization of the ER membrane, or the recruitment of

GAPDH was used as a loading control. C. The intracellular localization of wild-type ATG9A (wt) and ATG9A-G516A/C519A was assessed by immunofluorescence in HeLa cells. Pearson's correlation analyses were conducted to assess the extent of co-localization with markers of the ER (Calnexin), cis-Golgi (GM130), trans-Golgi (GOLGIN97), and endosomes (VPS26). Means ± SEM calculated from 30 cell images, collected in 3 independent experiments, are shown on the graphs. A non-paired Student's t-test was used to evaluate statistical significance of the results. Of note, the Pearson's correlation coefficients measured for the wild-type protein have also been documented in our recent article [14]. The datasets collected for ATG9A-G516A-C519A were obtained simultaneously to those of the wild-type protein but have not been reported in Staudt et al., 2016.

408

C. Staudt et al. / Biochemical and Biophysical Research Communications 479 (2016) 404e409

Fig. 3. Substitution of G516 by alanine is sufficient to prevent the transport of ATG9A from the ER to the Golgi apparatus. HeLa cells were transiently transfected with the indicated MYC-tagged ATG9A proteins and treated overnight with 5 mM of MG132, an inhibitor of proteasomal degradation. The intracellular levels of wild-type (wt) and mutated ATG9A were assessed in control and treated cells by western blotting using an anti-MYC antibody, and GAPDH was used as a loading control. Note that 20 mg and 10 mg of total proteins were loaded in the non-treated () and treated (þ) lanes, respectively. NT refers to Non Transfected cells and IB to immunoblotting.

binding partners that could stabilize the association of ATG9A molecules. While G516 and C519 are not critical for self-interaction of ATG9A, mutation of these residues severely impacts the subcellular trafficking of the protein. Substitution of C519 by an alanine induces a decrease of ATG9A stability in the ER, which translates into a lower level of ATG9A reaching the Golgi apparatus compared to wild-type proteins. However, the mutated protein remains able to travel through the Golgi apparatus, as indicated by acquisition of a complex-type N-glycan. These findings suggest that any modification of C519, such as a palmitoylation, is unlikely to be a prerequisite for transport of ATG9A from the ER to the Golgi apparatus. The impact of the G516A substitution on ATG9A is two-fold: it strongly decreases the stability of ATG9A, and it prevents its anterograde sorting to the Golgi apparatus. Intriguingly, this glycine residue and its surrounding amino acids do not form any consensus sequence that could be readily identified as an ER exit signal (e.g. [D/E]X[D/E], di-aromatic or di-hydrophobic motifs). While it is possible that G516 belongs to a new kind of motif signaling the packaging into Golgibound transport carriers (e.g. COPII vesicles), other hypotheses are worth considering to understand the impact of this amino acid substitution on the sorting of ATG9A. Among amino acids, glycine residues have the greatest conformational flexibility, which is why they are often involved in tight turn structures. Therefore, the loss of this amino acid in ATG9A might result in a change of conformation or orientation of the C-terminal portion of ATG9A, which could have consequences on its anterograde transport to the Golgi. For instance, the mutated protein could be unable to present an ER-exit signal to the ERexport machinery. The C-terminal region of ATG9A contains three motifs of the [D/E]X[D/E]-type, which might serve as recognition signals for binding to COPII coat components such as Sec24 isoforms. However, when we mutated each of these putative signals individually, ATG9A molecules were still able to acquire complex-type oligosaccharides, suggesting that neither of these motifs is solely responsible for the exit of ATG9A out of the ER (data not shown). The C-terminal region of ATG9A also contains several motifs of the RXR-type, which are known to retain transmembrane proteins in the ER when correctly exposed [23,24]. This exposure might arise in response to some structural change of the C-terminal tail induced by mutation of G516. Alternatively, substitution of this glycine could, directly or indirectly, disrupt the recognition of ATG9A by key subcellular trafficking regulators. For example, it has been reported that P38a regulates the transport of ATG9A from a peri-nuclear pool to forming autophagosomes via P38IP, a protein that binds to the C-terminal region of ATG9A [8]. It has also been documented that phosphorylation of ATG9A on a serine located in its C-terminal region (S761), which is mediated by ULK1 and the AMP-activated protein kinase, regulates association of ATG9A with 14-3-3 z, and thereby the transport of ATG9A to the

autophagosomal membrane [25]. In the future, it will be interesting to make use of the ER-sorting deficient ATG9A mutants identified in present and previous studies [14] to investigate which ATG9A binding partner(s), known or yet to be found, control(s) the first steps of its intracellular trafficking. Acknowledgements Part of this work was supported by F.R.S. - FNRS grant CDR J.0055.13 to Michel Jadot. Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.bbrc.2016.09.097. References [1] C.A. Lamb, T. Yoshimori, S.A. Tooze, The autophagosome: origins unknown, biogenesis complex, Nat. Rev. Mol. Cell Biol. 14 (2013) 759e774. [2] Y. Xie, R. Kang, X. Sun, M. Zhong, J. Huang, D.J. Klionsky, D. Tang, Posttranslational modification of autophagy-related proteins in macroautophagy, Autophagy 11 (2015) 28e45. €chl, S.G. Crawshaw, S. High, D.W. Hailey, [3] A.R. Young, E.Y. Chan, X.W. Hu, R. Ko J. Lippincott-Schwartz, S.A. Tooze, Starvation and ULK1-dependent cycling of mammalian Atg9 between the TGN and endosomes, J. Cell Sci. 119 (2006) 3888e3900. [4] A. Orsi, M. Razi, H.C. Dooley, D. Robinson, A.E. Weston, L.M. Collinson, S.A. Tooze, Dynamic and transient interactions of Atg9 with autophagosomes, but not membrane integration, are required for autophagy, Mol. Biol. Cell 23 (2012) 1860e1873. [5] E. Zavodszky, M. Vicinanza, D.C. Rubinsztein, Biology and trafficking of ATG9 and ATG16L1, two proteins that regulate autophagosome formation, FEBS Lett. 587 (2013) 1988e1996. [6] T. Yamada, A.R. Carson, I. Caniggia, K. Umebayashi, T. Yoshimori, K. Nakabayashi, S.W. Scherer, Endothelial nitric-oxide synthase antisense (NOS3AS) gene encodes an autophagy-related protein (APG9-like2) highly expressed in trophoblast, J. Biol. Chem. 280 (2005) 18283e18290. [7] T. Saitoh, N. Fujita, T. Hayashi, K. Takahara, T. Satoh, H. Lee, K. Matsunaga, S. Kageyama, H. Omori, T. Noda, N. Yamamoto, T. Kawai, K. Ishii, O. Takeuchi, T. Yoshimori, S. Akira, Atg9a controls dsDNA-driven dynamic translocation of STING and the innate immune response, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 20842e20846. [8] J.L. Webber, S.A. Tooze, Coordinated regulation of autophagy by p38alpha MAPK through mAtg9 and p38IP, EMBO J. 29 (2010) 27e40. [9] Y. Takahashi, C.L. Meyerkord, T. Hori, K. Runkle, T.E. Fox, M. Kester, T.P. Loughran, H.G. Wang, Bif-1 regulates Atg9 trafficking by mediating the fission of Golgi membranes during autophagy, Autophagy 7 (2011) 61e73. [10] S. He, D. Ni, B. Ma, J.H. Lee, T. Zhang, I. Ghozalli, S.D. Pirooz, Z. Zhao, N. Bharatham, B. Li, S. Oh, W.H. Lee, Y. Takahashi, H.G. Wang, A. Minassian, P. Feng, V. Deretic, R. Pepperkok, M. Tagaya, H.S. Yoon, C. Liang, PtdIns(3)Pbound UVRAG coordinates Golgi-ER retrograde and Atg9 transport by differential interactions with the ER tether and the beclin 1 complex, Nat. Cell Biol. 15 (2013) 1206e1219. [11] D. Popovic, I. Dikic, TBC1D5 and the AP2 complex regulate ATG9 trafficking and initiation of autophagy, EMBO Rep. 15 (2014) 392e401. [12] C.A. Lamb, S. Nühlen, D. Judith, D. Frith, A.P. Snijders, C. Behrends, S.A. Tooze, TBC1D14 regulates autophagy via the TRAPP complex and ATG9 traffic, EMBO J. 35 (2016) 281e301. [13] Y. Takahashi, N. Tsotakos, Y. Liu, M.M. Young, J. Serfass, Z. Tang, T. Abraham,

C. Staudt et al. / Biochemical and Biophysical Research Communications 479 (2016) 404e409

[14]

[15]

[16]

[17]

[18]

[19]

H.G. Wang, The Bif-1-Dynamin 2 membrane fission machinery regulates Atg9-containing vesicle generation at the Rab11-positive reservoirs, Oncotarget 7 (2016) 20855e20868. C. Staudt, F. Gilis, M. Boonen, M. Jadot, Molecular determinants that mediate the sorting of human ATG9A from the endoplasmic reticulum, Biochim. Biophys. Acta 1863 (2016) 2299e2310. C. Puri, M. Renna, C.F. Bento, K. Moreau, D.C. Rubinsztein, Diverse autophagosome membrane sources coalesce in recycling endosomes, Cell 154 (2013) 1285e1299. K. Imai, F. Hao, N. Fujita, Y. Tsuji, Y. Oe, Y. Araki, M. Hamasaki, T. Noda, T. Yoshimori, Atg9A trafficking through the recycling endosomes is required for autophagosome formation, J. Cell Sci. (2016). Sep 1. pii: jcs.196196, Epub ahead of print. T. Chang, L.A. Schroder, J.M. Thomson, A.S. Klocman, A.J. Tomasini, P.E. Strømhaug, W.A. Dunn Jr., PpATG9 encodes a novel membrane protein that traffics to vacuolar membranes, which sequester peroxisomes during pexophagy in Pichia pastoris, Mol. Biol. Cell 16 (2005) 4941e4953. C. He, M. Baba, Y. Cao, D.J. Klionsky, Self-interaction is critical for Atg9 transport and function at the phagophore assembly site during autophagy, Mol. Biol. Cell 19 (2008) 5506e5516. C. Aicart-Ramos, R.A. Valero, I. Rodriguez-Crespo, Protein palmitoylation and subcellular trafficking, Biochim. Biophys. Acta 1808 (2011) 2981e2994.

409

[20] F.H. Fenteany, K.J. Colley, Multiple signals are required for alpha2,6sialyltransferase (ST6Gal I) oligomerization and Golgi localization, J. Biol. Chem. 280 (2005) 5423e5429. [21] M. Raja, The potassium channel KcsA: a model protein in studying membrane protein oligomerization and stability of oligomeric assembly? Arch. Biochem. Biophys. 510 (2011) 1e10. [22] R.T. Youker, J.R. Bruns, S.A. Costa, Y. Rbaibi, F. Lanni, O.B. Kashlan, H. Teng, O.A. Weisz, Multiple motifs regulate apical sorting of p75 via a mechanism that involves dimerization and higher-order oligomerization, Mol. Biol. Cell 24 (2013) 1996e2007. [23] D.B. Scott, T.A. Blanpied, M.D. Ehlers, Coordinated PKA and PKC phosphorylation suppresses RXR-mediated ER retention and regulates the surface delivery of NMDA receptors, Neuropharmacology 45 (2003) 755e767. [24] M. Gassmann, C. Haller, Y. Stoll, S. Abdel Aziz, B. Biermann, J. Mosbacher, K. Kaupmann, B. Bettler, The RXR-type endoplasmic reticulum-retention/ retrieval signal of GABAB1 requires distant spacing from the membrane to function, Mol. Pharmacol. 68 (2005) 137e144. [25] V.K. Weerasekara, D.J. Panek, D.G. Broadbent, J.B. Mortenson, A.D. Mathis, G.N. Logan, J.T. Prince, D.M. Thomson, J.W. Thompson, J.L. Andersen, Metabolic-stress-induced rearrangement of the 14-3-3z interactome promotes autophagy via a ULK1- and AMPK-regulated 14-3-3z interaction with phosphorylated Atg9, Mol. Cell. Biol. 34 (2014) 4379e4388.