© 2015. Published by The Company of Biologists Ltd | Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

RESEARCH ARTICLE γ-SNAP stimulates disassembly of endosomal SNARE complexes and regulates endocytic trafficking pathways Hiroki Inoue1,*, Yuka Matsuzaki1, Ayaka Tanaka1, Kaori Hosoi1, Kaoru Ichimura2, Kohei Arasaki1, Yuichi Wakana1, Kenichi Asano1, Masato Tanaka1, Daisuke Okuzaki3, Akitsugu Yamamoto2, Katsuko Tani1 and Mitsuo Tagaya1,*

ABSTRACT specific organelles and transport vesicles, and mediates membrane Soluble N-ethylmaleimide-sensitive factor attachment protein transport in specific endocytic and exocytic routes. For example, receptors (SNAREs) that reside in the target membranes and SNAREs such as syntaxin (STX)4, STX7 and STX18 catalyze transport vesicles assemble into specific SNARE complexes to membrane fusion in the plasma membrane, endosomes and drive membrane fusion. N-ethylmaleimide-sensitive factor (NSF) and lysosomes, and the endoplasmic reticulum (ER), respectively its attachment protein, α-SNAP (encoded by NAPA), catalyze (Hatsuzawa et al., 2000; Sumitani et al., 1995; Wong et al., 1998). disassembly of the SNARE complexes in the secretory and SNARE proteins are classified into four groups, Qa, Qb, Qc and R, endocytic pathways to recycle them for the next round of fusion according to the sequence similarity of the SNARE motif and its events. γ-SNAP (encoded by NAPG) is a SNAP isoform, but its flanking regions (Hong, 2005; Jahn and Scheller, 2006). In most function in SNARE-mediated membrane trafficking remains cases, Qa-, Qb- and Qc-SNAREs reside in target membranes and thus unknown. Here, we show that γ-SNAP regulates the endosomal are also called target (t)-SNAREs. By contrast, most R-SNAREs trafficking of epidermal growth factor (EGF) receptor (EGFR) and reside in transport vesicles and thus are also called vesicle (v)- transferrin. Immunoprecipitation and mass spectrometry analyses SNAREs. Three Q-SNAREs and one R-SNARE form a specific four- revealed that γ-SNAP interacts with a limited range of SNAREs, helical bundle complex between opposing membranes of organelles including endosomal ones. γ-SNAP, as well as α-SNAP, mediated and transport vesicles in trans, and drive membrane fusion. The the disassembly of endosomal syntaxin-7-containing SNARE SNARE complex that is formed in a membrane as a result of the complexes. Overexpression and small interfering (si)RNA-mediated fusion event is disassembled and recycled for the next reaction by γ ATPase N-ethylmaleimide sensitive factor (NSF) and its attachment depletion of -SNAP changed the morphologies and intracellular α α distributions of endosomes. Moreover, the depletion partially protein, soluble NSF-attachment protein ( -SNAP; encoded by NAPA). Structural and biochemical analyses have revealed that three suppressed the exit of EGFR and transferrin from EEA1-positive α early endosomes to delay their degradation and uptake. Taken -SNAPs bind to hetero-tetramer SNARE complex and recruit the together, our findings suggest that γ-SNAP is a unique SNAP that NSF hexamer (Antonin et al., 2002; Chang et al., 2012; Söllner et al., – – 1993; Stein et al., 2009; Wimmer et al., 2001). functions in a limited range of organelles including endosomes α β and their trafficking pathways. In mammals, there are three SNAP isoforms: -SNAP, -SNAP (encoded by NAPB)andγ-SNAP (encoded by NAPG). In contrast to KEY WORDS: NAPG, STX7, STX8, Syntaxin 8, Endocytosis ubiquitous expression of α-SNAP, β-SNAP is specifically expressed in brain (Whiteheart et al., 1993). These two isoforms share more INTRODUCTION than 80% amino acid sequence identity, and thus β-SNAP can also Membrane trafficking in eukaryotic cells plays pivotal roles in a bind to and catalyze the disassembly of the SNARE complex, and the wide variety of cellular functions. Transport vesicles containing isoforms act together in regulated exocytosis in neuronal cells cargo molecules – e.g. secretory and membrane proteins – are (Sudlow et al., 1996; Xu et al., 2002). γ-SNAP is as ubiquitously generated from donor membranes, and they are then transported to, expressed as α-SNAP and is only ∼25% identical in its amino acid and become tethered to and fused with target membranes through sequence to both α-SNAP and β-SNAP. In contrast to the numerous cellular machineries. Soluble N-ethylmaleimide sensitive considerable contribution of α-SNAP and β-SNAP to SNARE factor attachment protein receptors (SNAREs) are key elements of complexes in membrane trafficking events, the biochemical and the membrane fusion machinery. At least 38 genes encoding cellular functions of γ-SNAP remain poorly understood. Yeast two- SNAREs exist in the human genome, and all of them have one or hybrid screening experiments have revealed that γ-SNAP interacts two insertions of a characteristic α-helical sequence called the with γ-SNAP-associated factor-1 [Gaf-1; also known as Rab11- SNARE motif (Hong, 2005; Hong and Lev, 2014; Jahn and interacting protein (Rip11) and Rab11-family interacting protein 5 Scheller, 2006). Each SNARE is preferentially distributed in (FIP5)], γ-tubulin and NSF (Chen et al., 2001; Tani et al., 2003). The two-hybrid assay also revealed that γ-SNAP does not bind to at least to six of the SNAREs tested, including STX4 and STX18, although 1 School of Life Sciences, Tokyo University of Pharmacy and Life Sciences, Hachioji, α-SNAP does bind to them (Tani et al., 2003). Moreover, γ-SNAP is Tokyo 192-0392, Japan. 2Faculty of Bioscience, Nagahama Institute of Bio-Science and Technology, Nagahama, Shiga 526-0829, Japan. 3Department of Molecular not required for vesicle-mediated transport of vesicular stomatitis Genetics, Research Institute for Microbial Diseases, Osaka University, Suita, Osaka virus G protein (VSVG) from the ER to the Golgi, but α-SNAP is 565-0871, Japan. essential for this process (Peter et al., 1998). Subramaniam et al., *Authors for correspondence ([email protected]; [email protected]) however, have shown that the endosomal SNARE STX8 is associated with not only α-SNAP and NSF, but also γ-SNAP, as revealed by

Received 24 June 2014; Accepted 17 June 2015 immunoprecipitation of STX8 (Subramaniam et al., 2000). Journal of Cell Science

2781 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

In this study, to determine the biochemical and cellular functions with the SNARE domain of STX8, as α-SNAP did (supplementary of γ-SNAP, we sought to identify binding partners of γ-SNAP by material Fig. S1C). Moreover, the binding of γ-SNAP to STX7 using immunoprecipitation and mass spectrometry analyses. Here, could be reduced upon competition with an excess amount of α- we show that γ-SNAP binds to selected SNAREs and catalyzes the SNAP (supplementary material Fig. S1D,E). These results suggest disassembly of the endosomal STX7–STX8 SNARE complex, but that γ-SNAP directly binds to STX proteins and that it has the same not of the Golgi-associated STX5–Bet1 SNARE complex. We also binding site on SNAREs as α-SNAP. observed that overexpression and small interfering (si)RNA- We next sought to define the residues and/or domain that mediated depletion of γ-SNAP affects the morphology and determine the preference of the interaction of γ-SNAP for intracellular localization of endosomes. Moreover, the depletion endosomal SNAREs. First, we used two chimeras of α-SNAP and delayed the trafficking of EGFR and transferrin (Tfn) from early γ-SNAP to probe their interactions with endogenous STX4 (plasma endosomes, but did not affect the constitutive secretion of membrane SNARE) and STX8 (endosomal SNARE) horseradish peroxidase fused with a secretory signal sequence (supplementary material Fig. S1F). A chimera comprising the N- (ssHRP). Our findings suggest that γ-SNAP promotes endosomal terminal half of bovine α-SNAP (amino acid residues 1–164) and SNARE disassembly and that it is required for normal endosome the C-terminal half of bovine γ-SNAP (amino acid residues 157– function. 312), termed NαCγ, bound well to STX8, as the wild-type proteins do, whereas the chimera still bound to STX4, but the binding was RESULTS much weaker than that of wild-type α-SNAP. By contrast, another γ-SNAP preferentially binds to endosomal SNAREs chimera NγCα, comprising the N-terminal half of γ-SNAP (amino To gain new insights into the function of γ-SNAP, we sought to acid residues 1–156) and the C-terminal half of α-SNAP (amino identify γ-SNAP-interacting proteins by immunoprecipitation and acid residues 165–295) bound to neither STX4 nor STX8. These mass spectrometry analyses. Triple-FLAG-tagged γ-SNAP results suggest that α-SNAP primarily binds to STX4 through its N- (3×FLAG–γ-SNAP) was transiently expressed in 293T cells and terminal half, as proposed previously (Marz et al., 2003; Rice and immunoprecipitated. The precipitated proteins were separated by Brunger, 1999), but that its C-terminal half also makes some using SDS-PAGE, silver-stained and identified mass spectrometry contribution to the binding and that the C-terminal half of γ-SNAP (Fig. 1A,B). In addition to NSF, which we have previously reported plays a more important role in its binding to STX8. (Tani et al., 2003), several SNARE proteins, SNARE-associated Marz et al. (2003) have proposed that eight positively charged proteins (Sly1/Sec1-family domain containing proteins 1 and 2, residues in the concave surface of the N-terminal twisted-sheet SCFD1 and SCFD2, respectively), α-SNAP and other membrane- domain of α-SNAP electrostatically interact with a cluster of trafficking-related proteins – including clathrin heavy chain and negatively charged residues in SNARE complexes that comprise phosphoinositide phosphatase Sac1 – were detected. Among these, STX1. To define the residues that determine the binding preference, we decided to focus on SNAREs because their interactions with γ- we focused on two – Lys94 and Lys163 – because their SNAP have remained poorly characterized (Chen et al., 2001; corresponding residues in γ-SNAP are negatively charged residues Hong, 2005; Tani et al., 2003). (Glu87 and Glu155). We examined the interaction of four mutants, First, to comprehensively identify γ-SNAP-interacting SNARE α-SNAP-K94E and -K163E, and γ-SNAP-E87K and -E155K with proteins, we immunoprecipitated endogenous γ-SNAP with 34 SNAREs (supplementary material Fig. S1G). Both α-SNAP FLAG-tagged SNARE proteins (Fig. 1C). Several SNAREs – mutants, particularly the K163E mutant, exhibited weaker binding including STX6, STX7, STX8, STX10, SNAP29, VAMP3, to STX4 than the wild-type protein, whereas their binding to STX8 VAMP4, Vti1a and Vti1b – precipitated endogenous γ-SNAP to a was comparable to that of wild type, suggesting that the charge of high degree. Interactions of endogenous γ-SNAP were confirmed these residues determine, at least in part, the binding preference of by co-immunoprecipitation with endogenous STX7 and STX8, but SNAPs. However, in contrast to our expectations, neither the E87K not with STX4 (Fig. 1D–F). In addition, weaker interactions with nor the E155K γ-SNAP mutant interacted with STX4. Taken γ-SNAP were observed for some SNAREs, which included STX5, together, these results suggest that the binding preferences of SNAPs STX12, STX16, VAMP5, VAMP7 and p31 (also known as USE1) to SNAREs are determined, at least in part, by the C-terminal half of (Fig. 1C; supplementary material Fig. S1A). In terms of STX16 and SNAPs and the charged residues located in the concave surface of VAMP7, interactions of the endogenous proteins with FLAG- or the N-terminal twisted-sheet domain. glutathione-S-transferase (GST)-tagged γ-SNAP were confirmed In addition to several SNAREs, abundant levels of α-SNAP and (supplementary material Fig. S1B). Most, but not all, of these NSF were also detected as γ-SNAP-interacting proteins in mass γ-SNAP-binding SNAREs have been proposed to function in spectrometry analyses (Fig. 1A,B). Although γ-SNAP has been endosomes and the trans-Golgi network (TGN) (Hong, 2005). shown to directly interact with NSF (Tani et al., 2003), direct interaction with α-SNAP has not been reported so far. Because it is Characterization of the interactions of γ-SNAP with well known that α-SNAP directly binds to NSF (Barnard et al., endosomal SNAREs 1997; Chang et al., 2012), we examined the notion that γ-SNAP α-SNAP binds to the SNARE domains of SNARE proteins and indirectly interacts with α-SNAP through NSF. As expected, in vitro disassembles the SNARE complexes by recruiting NSF (Hanson GST-pulldown experiments showed that γ-SNAP directly bound to et al., 1995; Marz et al., 2003; McMahon and Südhof, 1995). Given NSF, but indirectly to α-SNAP through NSF (supplementary that γ-SNAP interacts with a limited range of SNAREs, we material Fig. S1H). Furthermore, the binding of α-SNAP to NSF questioned whether or not γ-SNAP operates in the same manner as was competed-out by an excess amount of γ-SNAP (supplementary α-SNAP in order to disassemble SNARE complexes. We first material Fig. S1I). Taken together, these results suggest that γ- examined, in vitro, the direct binding of γ-SNAP to SNARE SNAP is incorporated into the same complex as α-SNAP through a proteins and determined its binding site on SNARE proteins by NSF hexamer complex, which has three SNAP-binding sites (Marz using GST-pulldown assays with bacterially expressed and purified et al., 2003), and that their binding sites on the NSF complex are proteins. The experiments revealed that γ-SNAP directly interacted identical, highly overlap or cooperate with each other. Journal of Cell Science

2782 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 1. γ-SNAP potentially interacts with proteins involved in membrane trafficking, including SNAREs. (A) Silver-staining of γ-SNAP interacting proteins that had been isolated by using immunoprecipitation. Transiently expressed 3×FLAG–γ-SNAP in 293T cells was immunoprecipitated, and the precipitated materials were separated by using SDS-PAGE and then silver-stained. The bands that were specific to 3×FLAG–γ-SNAP (arrows) were excised and analyzed by mass spectrometry. Arrowhead, 3×FLAG–γ-SNAP. (B) Proteins identified as γ-SNAP interacting proteins. The MASCOT scores (score) and the number of identified peptides is shown. Asterisks, SNARE proteins. (C) γ-SNAP preferentially binds to endosomal SNAREs. FLAG-tagged SNARE expression vectors (0.8 µg each) were transiently transfected into 293T cells, and immunoprecipitation (IP) was performed using FLAG-M2-conjugated beads. Co-precipitated endogenous γ-SNAP and FLAG–SNARE proteins were detected by immunoblotting (IB). (D–F) Endogenous STX7 and STX8 interact with γ-SNAP. 293T cells were lysed, and endogenous STX7 (D), STX8 (E) and STX4 (F) were immunoprecipitated using specific antibodies. Co-precipitated endogenous γ-SNAP and SNAREs were detected by immunoblotting.

γ-SNAP promotes the disassembly of the STX7–STX8 SNARE endogenous STX8 that formed a complex with GST–STX7 was complex recovered with glutathione-coated beads and detected by The results presented above prompted us to investigate whether or immunoblotting (Fig. 2A). STX8 was efficiently precipitated with not γ-SNAP functions with NSF in the disassembly of the SNARE GST–STX7 when no supplements were present (lane 1). Although complex, as α-SNAP does. For this purpose, we examined the the addition of ATP alone (lane 2), or ATP and NSF (lane 3) ability of γ-SNAP to catalyze the disassembly of the STX7–STX8 decreased the complex of GST–STX7 with STX8, a substantial complex. Lysates of 293T cells that had been transiently transfected amount of the complex still remained. The addition of purified α- with GST–STX7-encoding expression vectors were supplemented SNAP promoted the complete disassembly of the complex (lane 4), with purified recombinant His-tagged α-SNAP or γ-SNAP, NSF, which was blocked by the replacement of ATP with NEM,

N-ethylmaleimide (NEM) and/or ATP. Then after incubation, an inhibitor of NSF (lane 5). The addition of γ-SNAP, instead of Journal of Cell Science

2783 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 2. γ-SNAP stimulates disassembly of the STX7–STX8, but not the STX5-Bet1, complex. (A) STX7–STX8 SNARE complex disassembly assay. GST–STX7 was transiently expressed in 293T cells, and then the cells were lysed in assay buffer as described in Materials and Methods. Each cell lysate was supplemented with reagents – including α-SNAP or γ-SNAP (0.12 µM) – as indicated in each lane, and incubated. Endogenous STX8 that had associated with GST–STX7 was analyzed by immunoblotting. (B) STX5–Bet1 SNARE complex disassembly assay. Disassembly of transiently expressed GST–STX5 and endogenous Bet1 by α-SNAP or γ-SNAP was analyzed as described in A. (C,D) Comparison of the ability of γ-SNAP and α-SNAP to promote disassembly of STX7–STX8. The disassembly assay was performed as described in A, with the exception of the concentrations of SNAPs used. SNAPs were used at concentrations between 0.015 µM and 0.12 µM, as indicated. The levels of STX8 associated with GST–STX7 were quantified. Data are shown as the mean± s.e.m. of five independent experiments. NEM, N-ethylmaleimide.

α-SNAP, also completely disassembled the complex (Fig. 2A, right proteins with the ability to cause a dominant-negative effect. Thus, panels). By contrast, γ-SNAP failed to disassemble the complex of we examined whether or not overexpression of these mutants affects GST–STX5 and endogenous Bet1, a SNARE complex that the morphology of endosomes in HeLa cells (Fig. 3C,D). EEA1-, functions in ER–Golgi vesicle transport (Fig. 2B, right panels), LAMP1- and transferrin receptor (TfR)-positive endosomes although α-SNAP was able to disassemble it (left panels). These (marking early endosomes, late endosomes and/or lysosomes, and results suggest that γ-SNAP promotes the disassembly of recycling endosomes, respectively) in transiently transfected cells endosomal SNARE complexes. were stained with specific antibodies and examined by using Next, we compared the abilities of SNAPs to disassemble the confocal microscopy. In the cells that exhibited overexpression of STX7–STX8 complex. α-SNAP or γ-SNAP was added to lysates of the mutants, enlarged or aggregated EEA1-positive endosomes cells that expressed GST–STX7, which had been supplemented were significantly increased (Fig. 3C,D); however, TfR- or with NSF and ATP, and then the release of STX8 from GST–STX7 LAMP1-positive endosomes and/or lysosomes were little was examined (Fig. 2C,D). γ-SNAP was able to dissociate STX8 affected, if at all (supplementary material Fig. S2A,B). from GST–STX7 to approximately the same extent as α-SNAP, Unexpectedly, overexpression of the wild type also produced a suggesting that γ-SNAP, like α-SNAP, recruits NSF to SNARE similar effect on EEA1-positive endosomes. This swelling of complexes, resulting in their disassembly. EEA1-postive endosomes by overexpressing γ-SNAP was confirmed by using immuno-electron microscopy (immuno-EM) Overexpression of γ-SNAP affects the morphology of EEA1- (supplementary material Fig. S2C,D). These results suggest that the positive endosomes correct level of γ-SNAP activity is necessary to maintain the intact The C-terminal ten residues of α-SNAP are necessary for its morphology of EEA1-positive endosomes and that γ-SNAP interaction with NSF, and replacement of residue Leu294 for Ala participates in trafficking through early endosomes. within the region has a dominant-negative effect on the exocytosis of catecholamine in permeabilized chromaffin cells (Barnard et al., Depletion of γ-SNAP changes cell morphology, and the size 1997). Our previous in vitro GST-pulldown assays have and localization of endosomes in HepG2 cells demonstrated that γ-SNAP mutants with replacement of residue We next examined the effects of RNA interference (RNAi)- Leu311, which corresponds to Leu294 in α-SNAP, or of Cys312 for mediated depletion of γ-SNAP on organelle morphology in HeLa Ala exhibit less binding to NSF (Tani et al., 2003). Their impaired cells. Immunostaining showed no significant change in several binding to NSF was also confirmed upon immunoprecipitation from organelle markers in the cells that had been transfected with siRNAs 293T cell lysates (Fig. 3A). Importantly, the mutants showed against γ-SNAP (supplementary material Fig. S2E). During the weaker, but considerable, binding to STX8 and STX7 than the wild analysis, in contrast to the case of HeLa cells, we realized that type did (Fig. 3B). These properties might provide the mutant depletion of γ-SNAP altered the cellular morphology of human Journal of Cell Science

2784 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 3. Overexpression of γ-SNAP results in enlarged EEA1-positive endosomes. (A,B) γ-SNAP L311A and C312A mutants exhibit no and weaker binding activity toward NSF, and STX7 and STX8, respectively. The wild-type (WT) and mutant FLAG-tagged γ-SNAP proteins were transiently expressed in 293T cells and immunoprecipitated (IP) from the cell lysates using FLAG-M2-conjugated beads. Co-precipitated endogenous NSF (A), STX7 and STX8 (B) were detected by immunoblotting (IB). Asterisk, IgG light chain. (C) Overexpression of the wild-type and mutant (L311A and C312A) γ-SNAP proteins results in enlarged EEA1- positive endosomes. HeLa cells that had been transiently transfected with the FLAG–γ-SNAP wild-type or mutant proteins were immunostained and analyzed by using confocal microscopy. (D) Quantification of cells with enlarged EEA1 endosomes. Cells with enlarged EEA1-positive endosomes were counted. Endosomes with a more than twofold diameter compared with that of regular endosomes seen in non-transfected cells were regarded as enlarged. Data are shown as the mean±s.e.m. of six independent experiments; **P<0.01. Scale bar: 20 µm. hepatocellular carcinoma cell line HepG2 (supplementary material overexpression of γ-SNAP in HeLa cells (Fig. 3C). The swelling of Fig. S3A,B). HepG2 cells have a relatively round cell shape, and EEA1 endosomes upon γ-SNAP knockdown was also observed in tend to stick to each other and to grow in three dimensions 293T cells, as well as in HepG2 cells, but not in HeLa or MCF7 cells (supplementary material Fig. S3B, Mock). By contrast, γ-SNAP- (Fig. 4C). depleted cells had a morphology that made them stick less to each In addition to EEA1 endosomes, another endocytic compartment – other but which meant that they spread more on a substratum comprising TfR-positive recycling endosomes – was redistributed (supplementary material Fig. S3B, γ-SNAP siRNA no. 1 and no. 2). from the perinuclear region and throughout the cells to filopodia- This phenotype was also confirmed in analysis of the kinetics of cell like structures in HepG2 cells upon knockdown of γ-SNAP spreading (supplementary material Fig. S3C). (Fig. 4D). In the early secretory pathway, the perinuclear Golgi – We then examined the morphology of organelles in γ-SNAP- stained for β-COP – was also partially enlarged in γ-SNAP-depleted depleted HepG2 cells. Endocytic compartments – including EEA1-, cells (Fig. 4E). Bap31, an ER marker, was not affected at all cation-independent mannose 6-phosphate receptor (CI-M6PR)- (Fig. 4E). These results suggest that γ-SNAP plays a role in the (marking late endosomes and/or the TGN) and LAMP1-positive endocytic–recycling pathway and the Golgi complex in HepG2 endosomes were significantly enlarged upon knockdown of γ- cells. Immuno-EM analysis confirmed that γ-SNAP depletion SNAP (Fig. 4A). The specificity of this knockdown phenotype to increased the size of EEA1 endosomes (Fig. 5A,B,E). Of note, in EEA1 endosomes was confirmed by performing rescue experiments γ-SNAP-depleted cells, aggregates of small EEA1 endosomes were (Fig. 4B). It is worth noting that the stable expression of γ-SNAP also observed (Fig. 5C,D). These results suggest that γ-SNAP is itself did not induce the enlargement of EEA1 endosomes (Fig. 4B, necessary for the proper turnover of EEA1 endosomes, including mock in parental versus stable), which is in contrast to the transient fusion events between endosomes. Journal of Cell Science

2785 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 4. The size and distribution of endosomes and lysosomes are changed in γ-SNAP- depleted HepG2 cells. (A) HepG2 cells that had been mock transfected or transfected with the indicated siRNA against γ-SNAP (γ-SNAP siRNA #1 or #2) were immunostained for the indicated organelle markers and examined by using confocal microscopy. Left, LAMP1, CI-M6PR and EEA1. Right, quantification of cells with enlarged LAMP1- and EEA1-positive vesicles. Data are shown as the mean±s.e.m. of three independent experiments. (B) siRNA-resistant γ-SNAP rescues the EEA1 endosomal phenotype caused by knockdown of γ-SNAP. HepG2 cells that stably expressed bovine γ-SNAP (siRNA-resistant γ- SNAP Stable), which is resistant to siRNAs that target human γ-SNAP, were transfected with the siRNAs, immunostained for EEA1 and analyzed by using confocal microscopy. Left, representative images of staining of EEA1; right, quantification of the average size of EEA1 endosomes. (C) γ-SNAP depletion results in the enlargement of EEA1 endosomes in 293T cells, as in HepG2 cells, but not in HeLa or MCF7 cells. The cell lines were treated and analyzed as described for B. (D,E) HepG2 cells were transfected and analyzed as described in A except for immunostaining for endogenous TfR and γ-SNAP (D), and β-COP and Bap31 (E). *P<0.05, **P<0.01. Scale bars: 20 µm. Journal of Cell Science

2786 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 5. γ-SNAP depletion induces enlargement of EEA1 endosomes in HepG2 cells. Mock-transfected (A) or γ-SNAP-siRNA#1-transfected (B–D) HepG2 cells were fixed with 4% paraformaldehyde and then processed for immuno-EM analysis with an antibody against EEA1 and gold-particle-labeled secondary antibodies. B–D show several fields from the same condition. Arrows indicate EEA1-positive endosomes. (E) Quantification of the size of EEA1 endosomes in immuno-EM images. Left, frequency distribution. Right, the average size of EEA1 endosomes. The average data are shown as the mean±s.e.m. (the number of the analyzed endosomes from five cells, n=38 for mock and n=95 for those transfected with siRNA against γ-SNAP). **P<0.01. Scale bars: 2 µm (A–C); 1 µm (D).

Knockdown of γ-SNAP delays degradation of EGFR percentage, was also determined in the same dataset. There was less Knockdown of γ-SNAP affected the morphology of early and late of an increase in the percentage of the γ-SNAP-depleted cell endosomes, and lysosomes in HepG2 cells. EGFR is one of the most population that exhibited a lower level of fluorescence after treatment characterized endocytic cargos that are targeted to and degraded in with EGF than that of the mock cells (Fig. 6D). These phenotypes lysosomes. We therefore analyzed the effect of knocking down were reversed upon expression of siRNA-resistant wild-type γ- γ-SNAP over a time course of EGFR degradation after stimulation SNAP, but not by that of the Leu311Ala γ-SNAP mutant or wild- with the ligand in HepG2 cells that stably expressed green fluorescent type α-SNAP (Fig. 6E–G). These results suggest that γ-SNAP protein (GFP)-tagged EGFR (HepG2-EGFR–GFP cells) because the functions in the trafficking of EGFR during ligand-induced expression level of endogenous EGFR in parental HepG2 cells was degradation in a NSF-interaction-dependent manner and, at least in too low to be determined quantitatively. Over time, after treatment part, in a non-redundant manner with α-SNAP. with EGF, the number of EGFR–GFP-positive intracellular vesicles decreased in both mock and γ-SNAP-depleted cells, but more Knockdown of γ-SNAP partially inhibits Tfn uptake and its vesicles remained in γ-SNAP-depleted cells at 60 and 180 min after exit from early endosomes the treatment (Fig. 6A). The endocytosed EGFR–GFP was retained We next examined endocytosis and recycling of Tfn, which is in EEA1-positive endosomes, most of which were enlarged, as shown another well-characterized cargo protein for endocytosis and in Fig. 4, in γ-SNAP-depleted cells at 60 min; conversely, only a recycling. For the endocytosis assay, serum-starved HepG2 cells small fraction of EGFR–GFP was observed in EEA1-positive were allowed to continuously endocytose fluorescence-labeled Tfn endosomes in mock cells (Fig. 6A, Zoom). (Alexa–Tfn) for the indicated periods of time. In both mock cells To evaluate the delay of the degradation of EGFR–GFP in cells and cells that had been transfected with siRNA against γ-SNAP, it that had been transfected with siRNA against γ-SNAP, the intensity was observed by confocal microscopy that endocytosed Tfn of fluorescence was quantified by using flow cytometry (Fig. 6B). consecutively increased up to 1 h after the start of Tfn uptake The mean intensity of fluorescence (MIF) of EGFR–GFP decreased (Fig. 7A). To compare more precisely, the levels of endocytosed Tfn over time upon EGF stimulation in both mock and γ-SNAP-depleted were quantified by using flow cytometry. γ-SNAP depletion cells, as expected, but the decrease in γ-SNAP-depleted cells was resulted in a small but significant decrease in the level of significantly slower than that in mock cells (Fig. 6C). The population endocytosis in comparison with that of mock cells (Fig. 7B). For 3 that exhibited less EGFR–GFP fluorescence (MIF<1×10 ), as a recycling analysis, HepG2 cells that had been saturated with Alexa– Journal of Cell Science

2787 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 6. See next page for legend. Journal of Cell Science

2788 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 6. γ-SNAP depletion delays EGFR degradation in HepG2 cells. (A) Time 3×FLAG-γ-SNAP and mass spectrometry. The interactions were course of endocytosis and degradation of EGFR–GFP. Mock-transfected confirmed by immunoprecipitation of endogenous γ-SNAP with HepG2 cells and HepG2 cells transfected with siRNA against γ-SNAP (γ-SNAP – – – FLAG SNAREs, and endogenous STX7 and STX8. In addition to siRNA) stably expressing EGFR GFP (HepG2-EGFR GFP) cells were α stimulated with EGF for the indicated periods of time, fixed, stained with Hoechst these SNAREs, NSF and -SNAP were also identified as interacting 33342 and visualized by using confocal microscopy. The areas enclosed by proteins by using mass spectrometry, and GST-pulldown white boxes in the images at 60 min are enlarged and shown with EEA1 staining experiments showed that γ-SNAP was indirectly associated with α- (red) on the far right. (B–D) Flow cytometry analysis of EGF-induced EGFR– SNAP through NSF and that their binding sites on SNAREs and NSF GFP degradation in mock- and γ-SNAP-siRNA-transfected HepG2 cells. are seemingly identical. Moreover, γ-SNAP stimulated the (B) Representative histograms of the EGFR–GFP fluorescence level. The – 3 disassembly of the STX7 STX8 endosomal SNARE complex. number in each histogram indicates the percentage of cells exhibiting <1×10 α fluorescence intensity. (C) Time course of the EGFR–GFP level (relative Several lines of evidence indicate that three -SNAPs interact with geometric mean intensity fluorescence). (D) Time course of the percentage of the cis-SNARE complex to recruit the NSF hexamer to the complex cells that exhibited degradation of EGFR–GFP. (E–G) Rescue experiments. and to drive disassembly of the complex by inducing the ATPase HepG2-EGFR–GFP cells that stably expressed bovine wild-type (WT) or L311A activity of NSF (Barnard et al., 1997; Marz et al., 2003). From these mutant γ-SNAP proteins, or wild-type α-SNAP (bovine, siRNA-resistant) were observations, we propose a model in which α-SNAP and γ-SNAP mock transfected or with siRNA against γ-SNAP, and then EGFR–GFP partly replace each other in the NSF–SNAP–SNARE complex in endocytosis and degradation assays were performed as described in A and endocytic organelles and function to optimise the disassembly of the B. Confocal microscopy pictures at 60 min after treatment with EGF (E). The initial levels of EGFR–GFP fluorescence decrease for the first 30 min endosomal SNARE complex (Fig. 8). Alternatively, it is also (F). Percentages of cells without EGFR–GFP at 180 min (G), as analyzed by conceivable that γ-SNAP operates as a regulatory factor for α-SNAP- using flow cytometry. Data are shown as the mean±s.e.m. of three to five dependent disassembly of the SNARE complex. Although we were independent experiments; *P<0.05, **P<0.01. Scale bars: 20 µm. unable to distinguish the disassembly activities of α-SNAP and γ-SNAP, at least in vitro, both overexpression and depletion of Tfn were chased with Alexa–Tfn-free 10%-FBS-containing γ-SNAP clearly affected the endosomal system in cultured cells, medium, and the decrease of the fluorescence was analyzed. γ- suggesting that γ-SNAP has a unique function in the system. SNAP depletion caused no apparent change in the cell-associated Sec22b and Sly1, which we identified by using mass Tfn level at the endpoint (60 min) (Fig. 7C,D). However, because spectrometry as candidates of γ-SNAP binding proteins with the initial levels of cell-associated Tfn were different between mock higher MASCOT scores but which were excluded from further and γ-SNAP-depleted cells, the rate of the decreases from 0 min to analyses in this study, are a SNARE and a SNARE-associated 30 min were statistically significant (mock −22.4±1.0 MIF/30 min protein, respectively, that function in the trafficking between the ER vs γ-SNAP siRNA no. 1−17.9±1.0 and siRNA no. 2−16.8±0.8, and the Golgi (Dascher and Balch, 1996; Zhang et al., 1999). STX6, P<0.05). Vti1a, VAMP4 and SNAP29 also showed strong interactions with γ- As a fraction of the endocytosed EGFR–GFP accumulated in SNAP in FLAG–SNARE immunoprecipitation experiments. They EEA1-positive early endosomes in γ-SNAP-depleted cells (Fig. 6A), operate in the retrograde trafficking from endosomes to the TGN we also evaluated the colocalization of endocytosed Tfn and EEA1 to (Mallard et al., 2002). It is plausible that γ-SNAP is involved in the elucidate the underlying mechanism of the delay of endocytosis. In retrograde trafficking from endosomes to the ER through the Golgi. mock cells, a substantial amount of Tfn passed through early Indeed, the morphology of the Golgi was partially affected by γ- endosomes and was visualized in EEA1-negative vesicles at 40 min SNAP depletion in HepG2 cells. Moreover, γ-SNAP binds to a after the start of Alexa–Tfn uptake (Fig. 7E, upper row). By contrast, purified Golgi membrane (Whiteheart et al., 1992), although γ- the majority of Tfn remained in EEA1-positive endosomes in SNAP is involved in neither the anterograde trafficking of the γ-SNAP-depleted cells (Fig. 7E, lower row). Quantitative analyses VSVG protein from the ER to the Golgi (Peter et al., 1998) nor the showed that γ-SNAP depletion resulted in a significant increase in constitutive secretion of ssHRP. Whether γ-SNAP functions in the the colocalization of Tfn and EEA1 (Fig. 7F). retrograde trafficking through the Golgi in addition to the endocytic Finally, we also examined whether γ-SNAP is involved in the pathway should be evaluated in the near future, and thus we are now constitutive secretion of ssHRP. The secretion of ssHRP was studying these points. comparable in mock and γ-SNAP-depleted HepG2 cells, whereas it RNAi-mediated depletion of γ-SNAP affected the morphology was significantly impaired in HepG2 cells that had been depleted and localization of multiple organelles in the endocytic pathway. of α-SNAP (supplementary material Fig. S4), which is known to Early endosomes and lysosomes were modestly but significantly be an essential SNAP in the early secretory pathway (Peter et al., enlarged by the depletion. Furthermore, overexpression of wild- 1998). These results suggest that γ-SNAP is specifically required type γ-SNAP and its mutants also enlarged early endosomes. These for the efficient endocytosis of Tfn and trafficking from EEA1- observations suggest that γ-SNAP plays a role in early endosomes positive early endosomes, in addition to the trafficking of EGFR– and that an appropriate level of γ-SNAP is required to maintain the GFP. normal size of the endosomes. In addition to the change in the organelle size, the depletion of γ-SNAP partially inhibited ligand- DISCUSSION induced EGFR degradation and Tfn uptake, but not ssHRP In this study, we reveal that γ-SNAP interacts with endosomal secretion. These endocytosed EGFR–GFP and Alexa–Tfn proteins SNAREs and stimulates the disassembly of the endosomal SNARE colocalized more with EEA1, and their exit from early endosomes STX7–STX8 complex, but not of the Golgi-associated SNARE was delayed, suggesting that γ-SNAP is also required for efficient STX5–Bet1 complex. Moreover, our data suggest that γ-SNAP trafficking of endocytic cargos from early endosomes. Similar positively regulates the trafficking of EGFR and Tfn from early swelling of early endosomes and delay of the exit of Tfn from the endosomes in HepG2 cells. These findings are summarized in the enlarged early endosomes is observed in cells that have been model presented in Fig. 8. depleted of FIP5, which is a Rab11- and γ-SNAP-binding protein We identified several SNARE proteins as candidates for γ-SNAP- (Schonteich et al., 2008). Similarly, the partial inhibition of EGFR interacting proteins by using immunoprecipitation analyses with degradation by depletion of γ-SNAP remarkably resembled the Journal of Cell Science

2789 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 7. Endocytosis of Tfn and its exit from early endosomes are affected by γ-SNAP depletion. (A) Time course of endocytosis of Tfn. Mock-transfected HepG2 cells and HepG2 cells transfected with siRNA against γ-SNAP (γ-SNAP siRNA) were allowed to take up Alexa546–Tfn for the indicated periods of time and then analyzed by using confocal microscopy. (B) Quantification of Tfn uptake. The cells were treated as described in the Materials and Methods and analyzed by using flow cytometry. (C) Time course of recycling of Tfn. The transfected cells were saturated with Alexa546–Tfn for 1 h and then allowed to recycle Alexa546– Tfn out for the indicated periods of time. (D) Quantification of Tfn recycling. The cell-associated Alexa–Tfn was analyzed by using flow cytometry as described in Materials and Methods. (E) The transfected cells were allowed to endocytose Alexa488–Tfn for 40 min, washed, fixed, stained for EEA1 and visualized by using confocal microscopy. (F) Quantification of the fraction of Tfn that colocalized with EEA1. The colocalization (Manders’ coefficients) was calculated from the datasets – including panel E – by using the JACoP plugin in ImageJ. Data are shown as the mean±s.e.m. of three independent experiments; *P<0.05, **P<0.01. results of inhibition of STX8 using antibodies against STX8 could cooperate with STX8 and FIP5 in the trafficking of cargos (Prekeris et al., 1999; Antonin et al., 2000) and of STX8- from early endosomes to late endosomes and/or lysosomes, and interacting protein Mig-6 using siRNA-mediated depletion (Ying recycling endosomes, respectively. et al., 2010). STX8 operates in the trafficking from early In HepG2 cells, γ-SNAP depletion caused an enlargement of endosomes to lysosomes and is primarily enriched in early EEA1-positive endosomes. A similar morphological defect of endosomes (Subramaniam et al., 2000). Collectively, γ-SNAP endosomes caused by γ-SNAP depletion was also observed in 293T Journal of Cell Science

2790 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Fig. 8. A proposed model for the function of γ-SNAP. (A) Interaction of α-SNAP and γ-SNAP with the SNARE complex comprising NSF. γ-SNAP directly interacts with selected SNAREs, especially endosomal SNAREs – including STX7 and STX8 – through their SNARE domains, as α-SNAP does (Fig. 1; supplementary material Fig. S1). γ-SNAP also directly interacts with NSF, and indirectly with α-SNAP through NSF (Figs 1, 3; supplementary material Fig. S1). α-SNAP and γ-SNAP share the same binding sites on SNARE and NSF (supplementary material Fig. S1). γ-SNAP, probably together with α-SNAP, promotes disassembly of the SNARE complex in a NSF-dependent manner (Fig. 2). (B) The trafficking pathway in which γ- and α-SNAP are involved. γ-SNAP maintains the morphologies of early and late endosomes and lysosomes (Figs 4–7). Moreover, γ-SNAP is required for efficient endocytosis of EGFR and Tfn, and their exit from early endosomes (Figs 6, 7). γ-SNAP might be involved in these events through disassembly of the endocytic SNARE complex and/or regulation of the trafficking itself of SNARE proteins.

cells, but not in HeLa, MDA-MB-231 or MCF7 cells. Although induced endosomal abnormality and the delay of EGFR degradation γ-SNAP and α-SNAP are ubiquitously expressed in all cell lines and by stably expressing α-SNAP in HepG2 cells (Fig. 6E–G; tissues tested, their expression levels vary in tissues and cell lines. In supplementary material Fig. S3D), suggesting that the expression mouse tissues, mRNA encoding γ-SNAP is most highly expressed level of α-SNAP is not the major reason for the cell line difference in heart, liver and kidney, whereas mRNAs encoding α-SNAP and that both SNAPs have non-redundant functions in EEA1 shows the lowest expression in these tissues (Whiteheart et al., endosomes. 1993). Interestingly, HepG2 and 293T cells are derived from liver The gene encoding γ-SNAP is evolutionarily conserved in and kidney, respectively. The differing endosomal phenotype upon multicellular organisms; it does not exist in unicellular yeasts, γ-SNAP knockdown, therefore, might be related to tissue- or including Saccharomyces cerevisiae and Schizosaccharomyces cell-type-specific endogenous factor(s). One of the possibilities is pombe, but does exist in the social amoeba Dictyostelium, the fruit that α-SNAP compensates for the defect caused by γ-SNAP fly, nematodes, higher plants and vertebrates – including humans. depletion. However, we failed to rescue the γ-SNAP-depletion- The distribution of the gene implies that γ-SNAP might be associated Journal of Cell Science

2791 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634 with biological processes that are unique to multicellular organisms – STX5 and STX7, their full-length ORFs were subcloned into pEBG (Mayer such as morphogenesis and intercellular communication. Because et al., 1995). For protein purification, STX7 and STX8 cDNAs with their these processes are highly reliant on endocytic recycling and transmembrane domains deleted were amplified by PCR and subcloned into degradation of adhesion molecules and receptors for growth factors pGEX-4T-1 (GE Healthcare, Little Chalfont, UK) and pQE30 (Qiagen, and cytokines, it might explain the phenotypes caused by γ-SNAP Venlo, The Netherlands). cDNA encoding human EGFR that was C- terminally tagged with GFP was amplified by using PCR and subcloned into knockdown of cell spreading, the redistribution of TfR-positive the pCI-IRES-bsr vector (Promega, Madison, WI) (Yonekawa et al., 2011). recycling endosomes and EGFR degradation in HepG2 cells, which The sequences of all PCR-amplified DNAs were verified by DNA exhibit tighter cell–cell adhesion than HeLa cells. Interestingly, a sequencing using an automatic DNA sequencer, ABI PRISM 3100 recent study has revealed a genetic interaction between γ-SNAP and (Applied Biosystems, Foster City, CA). the endosomal sorting complex required for transport (ESCRT), which is involved in the biogenesis of multivesicular bodies and RNA interference other processes, in Drosophila eyes (Lu et al., 2013). The expression siRNAs against human γ-SNAP were purchased from Life Technologies. of a mutant of the ESCRT-III subunit CHMP2B causes spotted The sequences of siRNA no. 1 and siRNA no. 2 were 5′-GCCAGAUUA- melanization and the rough eye phenotype, which represents UGACAGUGCCGCUUCU-3′ (HSS112897) and 5′-CAAGUACAUGG- disruption of the regularly packed ommatidia. Lu et al. have shown ACAAUGAUUAUGCU-3′ (HSS112898), respectively, based on the that the phenotypes caused by the mutation of CHMP2B are human γ-SNAP sequence. Both the siRNAs efficiently suppressed the enhanced by deletion of the Nsf2 gene, which encodes one of the two expression of the endogenous γ-SNAP protein (∼80% knockdown, Drosophila supplementary material Fig. S3A). siRNAs were transfected according to NSF proteins in . In addition, deletion or RNAi-mediated ’ ‘ ’ depletion of α-SNAP, γ-SNAP or the endosomal SNARE STX12 the manufacturer s reverse transfection protocol. Briefly, siRNAs were applied to 1×105 of trypsinized and suspended cells in a well of 12-well plate also enhances the phenotypes, similar to those seen upon deletion of at 30 nM, final concentration, in culture medium with 2 µl of Lipofectamine NSF2. These observations support our results and the hypothesis that 2000 (Life Technologies). Cells were examined 72 h after transfection. γ-SNAP functions in the endosomal system to modulate receptor trafficking and biological processes specific to multicellular Protein purification and in vitro pulldown assay organisms. GST- and His-tag proteins were expressed in BL21-CodonPlus (Agilent Technologies, Santa Clara, CA) harboring pREP4 (Qiagen) and then MATERIALS AND METHODS purified using glutathione–Sepharose (GE Healthcare) and Ni-NTA-agarose Cell culture and plasmid transfection (Qiagen), respectively, according to the manufacturers’ instructions. For Cells were maintained and transfected with plasmids as described only His–NSF purification, 1 mM ATP was included in all buffers to previously (Inoue et al., 2008). Cells were used for assays 18 h after prevent dissociation of the homohexamer. transfection. HepG2 cells stably expressing EGFR–GFP, and bovine γ- To examine in vitro interactions of proteins, the GST-pulldown assay was SNAP – wild type, Leu311Ala mutant – or α-SNAP were isolated and performed as described previously (Inoue et al., 2008). Briefly, GST-tagged maintained in the presence of 5 µg/ml blasticidin S (Kaken Pharmaceutical and His-tagged proteins were mixed in reconstitution buffer [0.1% Triton X- Co., Tokyo, Japan) and 1 µg/ml puromycin (Wako Pure Chemical, Osaka, 100, 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA and 1 mM Japan), respectively, in Dulbecco’s modified Eagle’s medium (DMEM) dithiothreitol (DTT)] at the concentrations indicated in each figure and then containing 10% FBS and penicillin–streptomycin. incubated for 1 h at 4°C with gentle agitation. Glutathione-coated beads were then added to the mixture, followed by incubation at 4°C for 30 more minutes. Antibodies After extensive washing with reconstitution buffer, materials bound to the beads The commercial mouse monoclonal antibodies used here were anti-FLAG were eluted with elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM reduced M2 (Sigma-Aldrich, St. Louis, MO), anti-STX8 (clone 48; BD Biosciences, glutathione) and detected by immunoblotting using appropriate antibodies. Franklin Lakes, NJ), anti-Bet1 (clone 17; BD Biosciences), anti-α-tubulin (clone MD1A; Sigma), anti-Penta-His (Qiagen, Valencia, CA), anti-EEA1 Immunoprecipitation and mass spectrometry (clone 14; BD Biosciences), anti-LAMP1 (clone H4A3; Developmental Subconfluent 293T cells from a well of 12-well plate and from one-third of a Studies Hybridoma Bank, Iowa City, IA), anti-CI-M6PR (clone 2G11; 10-cm dish per sample were used for immunoprecipitation of transiently Affinity BioReagents, Golden, CO) and anti-TfR (Life Technologies, transfected FLAG-tagged SNAREs and endogenous STXs, respectively. Carlsbad, CA). The commercial rabbit polyclonal antibodies used were anti- The cells were lysed in 1% Triton lysis buffer (1% Triton X-100, 25 mM FLAG (Sigma), anti-GST (Santa Cruz Biotechnology, Santa Cruz, CA), Hepes-KOH, pH 7.2, 150 mM KCl, 2 mM EDTA and 1 mM DTT) anti-HA (Santa Cruz Biotechnology) and anti-EGFR (Santa Cruz containing 1 µg/ml aprotinin, 0.5 µg/ml leupeptin, 1 µM pepstatin and Biotechnology). The rabbit polyclonal antibodies to β-COP, BAP31 and 1 mM phenylmethanesulfonyl fluoride (PMSF), and then the lysates were γ-SNAP have been described previously (Kawase et al., 2003; Tani et al., cleared by centrifuging. FLAG-M2-conjugated beads (Sigma) for FLAG- 2003; Wakana et al., 2008). Anti-STX4 and anti-STX7 polyclonal tagged proteins, or specific antibodies and Protein-G-beads (Jackson antibodies were raised by immunizing rabbits with the purified ImmunoResearch, West Grove, PA) for endogenous proteins were added recombinant His-tagged cytoplasmic domains of human STX4 (residues to the lysate, followed by incubation for 1 h at 4°C with gentle agitation. 1–273) and STX7 (residues 1–237), respectively, and then the antisera After washing the beads extensively with 1% Triton lysis buffer, the were affinity-purified. Secondary antibodies labeled with horseradish precipitated proteins were eluted with SDS-sample buffer and analyzed by peroxidase and fluorochrome were purchased from Bio-Rad Laboratories immunoblotting. (Hercules, CA) and Life Technologies, respectively. For mass spectrometry analysis, subconfluent 293T cells in a 10-cm dish were transfected with the 3×FLAG–γ-SNAP expression vector and DNA constructs immunoprecipitated with FLAG-M2 beads as described above. The Bacterial expression vectors of γ-SNAP and NSF have been described precipitated materials were eluted with 25 µg/ml 3×FLAG peptide previously (Tani et al., 2003). Triple-FLAG-tagged γ-SNAP (3×FLAG–γ- (Sigma), separated by SDS-PAGE and stained with Protein Silver Stain SNAP) was constructed by inserting the human γ-SNAP open reading frame Kit KANTO III (Kanto Chemical Co., Tokyo, Japan). Protein bands specific (ORF) into pcDNA4/TO-3×FLAG in-frame (Ingmundson et al., 2007). to 3×FLAG-γ-SNAP immunoprecipitation were cut out, and the proteins cDNAs encoding SNARE proteins were reverse-transcribed from HeLa, were trypsinized. The resultant peptides were analyzed by nanocapillary 293T and MDA-MB-231 cells, amplified by PCR and then cloned into reversed-phase LC-MS/MS using a C18 column (f 0.1×150 mm) on a pFLAG6 (Sigma). To obtain mammalian expression vectors of GST-tagged nanoLC system (Advance, Michrom BioResources, Auburn, CA) coupled Journal of Cell Science

2792 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634 to an LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, HRP secretion assay Waltham, MA). Proteins were identified by database searching in-house The HRP secretion assay was performed as described previously (Bossard MASCOT Server (Matrix Science, London, UK). The proteins for which the et al., 2007) with some modifications. Briefly, HepG2 cells were transfected MASCOT score was more than 100 are listed in Fig. 1B. Keratins, heat with siRNAs against γ-SNAP or α-SNAP as described above. After 48 h for shock proteins, ribosomal proteins, tRNA-related proteins, translation γ-SNAP knockdown and 24 h for α-SNAP knockdown, the cells were initiation and elongation factors, dehydrogenases (L-lactate, G3P), RNA transfected again with an expression plasmid encoding FLAG-tagged splicing factors, nuclear ribonucleoproteins, hemoglobin subunits, ssHRP. Eighteen hours after plasmid transfection, the cells were washed mitochondrial inner membrane proteins and γ-SNAP itself were excluded twice with Opti-MEM (Life Technologies), cultured in Opti-MEM for 2 or from the list of the interacting protein candidates because they are highly 4 h, and then the conditioned medium containing secreted HRP was abundant in cells and are routinely detected as contaminants. harvested. After being cleared by centrifugation, 10 µl of the conditioned medium was mixed with 200 µl of ABTS solution (Roche) – a colorimetric SNARE disassembly assay substrate for HRP – and then the mixture was incubated at 37°C for 1 h. The Subconfluent 293T cells transiently expressing GST-tagged full-length absorbance of the mixture was determined at 405 nm. At the same time, the STXs in a 10-cm dish were washed with ice-cold PBS twice and then lysed cell layer was washed with PBS thoroughly, lysed in 1% Triton lysis buffer, in 1 ml of 1% Triton lysis buffer containing the protease inhibitors. The cleared by centrifugation, mixed with ABTS and then the absorbance at lysates were cleared by centrifuging, divided into five tubes (200 µl each) 405 nm was determined as for the conditioned medium. To normalize the and then supplemented with 8 mM MgCl2, 0.5 mM ATP, 10 µg/ml His– secreted HRP level with the expression level of HRP in the cells, the NSF, 4 µg/ml His-tagged α-orγ-SNAP and/or 10 µg/ml NEM, as indicated absorbance from the medium was divided by that from the cell lysate. The in each figure. After incubation at 16°C for 1 h, 10 µl of glutathione beads data are shown as the relative values versus mock control. was added to the reactions, followed by incubation at 4°C for 1 h with gentle agitation and washing with 1% Triton lysis buffer three times. Co- Statistical analysis precipitated SNARE proteins were then eluted with SDS-sample buffer and The results from at least three experiments are expressed as the mean±s.e.m. detected by immunoblotting. Statistical analysis was performed using a two-tailed, unpaired Student’s t-test with P<0.05 considered to be statistically significant. Immunofluorescence and microscopy Immunostaining and confocal microscopy were performed as described Acknowledgements previously (Inoue et al., 2012). For visualization of trafficking and We thank Drs Craig R. Roy (Yale University, New Haven, CT) and Vivek Malhotra degradation of EGFR–GFP, cells were starved overnight in serum-free (Centro De Regulación Genomica, Barcelona, Spain) for providing pcDNA4/TO- DMEM and then treated for 1 h with 10 mg/ml cycloheximide (CHX) in 3×FLAG and ssHRP-FLAG vectors, respectively, and Dr Kazunobu Saito (Osaka University, Suita, Japan) for assistance with mass spectrometry. We also thank Yoko 0.02% BSA-containing DMEM (DMEM/BSA) at 37°C under 5% CO2 to block de novo synthesis of EGFR–GFP. The cells were then stimulated with Mizuya for secretarial assistance and Takuya Watanabe, Yuta Tateno, Mika Hiratsuka and Asami Tsuchida for experimental support. 100 ng/ml recombinant human EGF (rhEGF; Promega) in DMEM/BSA containing CHX for the indicated periods of time, fixed and visualized as Competing interests described above. For analysis of trafficking of Tfn, HepG2 cells were The authors declare no competing or financial interests. starved for 30 min in serum-free DMEM and then incubated with 10 µg/ml Alexa-Fluor-594-conjugated human Tfn (Life Technologies) in DMEM/ Author contributions – BSA for 30 min on ice to allow Alexa Tfn to bind to the cell surface. To This study was designed and directed by H.I. and M. Tagaya; H.I., Y.M., A.T., K.H., assess the uptake of Tfn, the cells were warmed at 37°C under 5% CO2 for K. Arasaki, and Y.W. performed experiments; K.I. and A.Y. performed the electron the indicated periods of time in the Alexa–Tfn-containing DMEM/BSA. To microscopic experiments; K. Asano and M. Tanaka contributed with expertise in flow assess the recycling, the cells were saturated with Alexa–Tfn by incubating cytometric analysis; D.O. performed mass spectrometry; K.T. contributed with the cells for 30 min in the Alexa–Tfn-containing medium, washed with reagents and expertise concerning endosomal trafficking analysis. DMEM/BSA to remove unbound Tfn and then incubated in DMEM with 10% FBS for the indicated periods of time. At the end of uptake or recycling, Funding the cells were extensively washed with ice-cold PBS, fixed and visualized as This work was supported in part by Grants-in-Aid for Scientific Research from the described above. The captured images were analyzed for colocalization Ministry of Education, Culture, Sports, Science and Technology of Japan [project ̀ numbers 23570174 to H.I.; 24790425 to K.A.; 25840042 to Y.W. and 25291029 to using the JACoP plug-in for ImageJ software (Bolte and Cordelieres, 2006). M.T.].

Immuno-EM Supplementary material Immuno-EM was performed as described previously (Iinuma et al., 2009), Supplementary material available online at except that cells were fixed with 4% PFA for 30 min to retain the http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.158634/-/DC1 immunoreactivity of EEA1. References Flow cytometry Antonin, W., Holroyd, C., Fasshauer, D., Pabst, S., von Mollard, G. F. and Jahn, R. siRNA-transfected cells were harvested with 5 mM EDTA in PBS at 37°C (2000). A SNARE complex mediating fusion of late endosomes defines conserved for 15 min. To analyze EGFR–GFP fluorescence, HepG2-EGFR–GFP cells properties of SNARE structure and function. EMBO J. 19, 6453-6464. were washed with DMEM/BSA and then chilled on ice for 30 min with Antonin, W., Fasshauer, D., Becker, S., Jahn, R. and Schneider, T. R. (2002). 10 µg/ml CHX in DMEM/BSA. rhEGF was applied at 100 ng/ml to the Crystal structure of the endosomal SNARE complex reveals common structural cells, and then the cells were incubated for the indicated periods of time at principles of all SNAREs. Nat. Struct. Biol. 9, 107-111. Barnard, R. J. O., Morgan, A. and Burgoyne, R. D. (1997). Stimulation of NSF 37°C under 5% CO . The cells were then fixed with 2% paraformaldehyde 2 ATPase activity by alpha-SNAP is required for SNARE complex disassembly and in PBS, washed with PBS containing 0.5% BSA and 5 mM EDTA, and then exocytosis. J. Cell Biol. 139, 875-883. analyzed with a FACSVerse Flow Cytometer (BD Biosciences). To analyze Bolte, S. and Cordelieres,̀ F. P. (2006). A guided tour into subcellular colocalization – Alexa Tfn fluorescence, HepG2 cells were treated as described in the analysis in light microscopy. J. Microsc. 224, 213-232. Immunofluorescence and microscopy section, except that all reactions were Bossard, C., Bresson, D., Polishchuk, R. S. and Malhotra, V. (2007). Dimeric performed with suspended cells and that Alexa-Fluor-488-labeled Tfn (Life PKD regulates membrane fission to form transport carriers at the TGN. J. Cell Biol. Technologies) was used instead of the Alexa-Fluor-594-labeled one. The 179, 1123-1131. cell-associated Tfn fluorescence was determined using a FACSVerse Flow Chang, L.-F., Chen, S., Liu, C.-C., Pan, X., Jiang, J., Bai, X.-C., Xie, X., Wang, H.-W. Cytometer as for EGFR–GFP, the data are shown as the relative values of and Sui, S.-F. (2012). Structural characterization of full-length NSF and 20S

MFI versus the initial value of each sample. particles. Nat. Struct. Mol. Biol. 19, 268-275. Journal of Cell Science

2793 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 2781-2794 doi:10.1242/jcs.158634

Chen, D., Xu, W., He, P., Medrano, E. E. and Whiteheart, S. W. (2001). Gaf-1, a γ- Prekeris, R., Yang, B., Oorschot, V., Klumperman, J. and Scheller, R. H. (1999). SNAP-binding protein associated with the mitochondria. J. Biol. Chem. 276, Differential roles of syntaxin 7 and syntaxin 8 in endosomal trafficking. Mol. Biol. 13127-13135. Cell 10, 3891-3908. Dascher, C. and Balch, W. E. (1996). Mammalian Sly1 regulates syntaxin 5 Rice, L. M. and Brunger, A. T. (1999). Crystal structure of the vesicular transport function in endoplasmic reticulum to Golgi transport. J. Biol. Chem. 271, protein Sec17: implications for SNAP function in SNARE complex disassembly. 15866-15869. Mol. Cell 4, 85-95. Hanson, P. I., Otto, H., Barton, N. and Jahn, R. (1995). The N-ethylmaleimide- Schonteich, E., Wilson, G. M., Burden, J., Hopkins, C. R., Anderson, K., sensitive fusion protein and alpha-SNAP induce a conformational change in Goldenring, J. R. and Prekeris, R. (2008). The Rip11/Rab11-FIP5 and kinesin II syntaxin. J. Biol. Chem. 270, 16955-16961. complex regulates endocytic protein recycling. J. Cell Sci. 121, 3824-3833. ̈ Hatsuzawa, K., Hirose, H., Tani, K., Yamamoto, A., Scheller, R. H. and Tagaya, M. Sollner, T., Whiteheart, S. W., Brunner, M., Erdjument-Bromage, H., (2000). Syntaxin 18, a SNAP receptor that functions in the endoplasmic reticulum, Geromanos, S., Tempst, P. and Rothman, J. E. (1993). SNAP receptors implicated in vesicle targeting and fusion. Nature 362, 318-324. intermediate compartment, and cis-Golgi vesicle trafficking. J. Biol. Chem. 275, Stein, A., Weber, G., Wahl, M. C. and Jahn, R. (2009). Helical extension of the 13713-13720. neuronal SNARE complex into the membrane. Nature 460, 525-528. Hong, W. (2005). SNAREs and traffic. Biochim. Biophys. Acta 1744, 120-144. Subramaniam, V. N., Loh, E., Horstmann, H., Habermann, A., Xu, Y., Coe, J., Hong, W. and Lev, S. (2014). Tethering the assembly of SNARE complexes. Trends Griffiths, G. and Hong, W. (2000). Preferential association of syntaxin 8 with the Cell Biol. 24, 35-43. early endosome. J. Cell Sci. 113, 997-1008. Iinuma, T., Aoki, T., Arasaki, K., Hirose, H., Yamamoto, A., Samata, R., Sudlow, A. W., McFerran, B. W., Bodill, H., Barnard, R. J. O., Morgan, A. Hauri, H.-P., Arimitsu, N., Tagaya, M. and Tani, K. (2009). Role of syntaxin and Burgoyne, R. D. (1996). Similar effects of alpha- and beta-SNAP on Ca(2+)- 18 in the organization of endoplasmic reticulum subdomains. J. Cell Sci. 122, regulated exocytosis. FEBS Lett. 393, 185-188. 1680-1690. Sumitani, S., Ramlal, T., Liu, Z. and Klip, A. (1995). Expression of syntaxin 4 in rat Ingmundson, A., Delprato, A., Lambright, D. G. and Roy, C. R. (2007). Legionella skeletal muscle and rat skeletal muscle cells in culture. Biochem. Biophys. Res. pneumophila proteins that regulate Rab1 membrane cycling. Nature 450, Commun. 213, 462-468. 365-369. Tani, K., Shibata, M., Kawase, K., Kawashima, H., Hatsuzawa, K., Nagahama, M. Inoue, H., Ha, V. L., Prekeris, R. and Randazzo, P. A. (2008). Arf GTPase- and Tagaya, M. (2003). Mapping of functional domains of γ-SNAP. J. Biol. Chem. activating protein ASAP1 interacts with Rab11 effector FIP3 and regulates 278, 13531-13538. pericentrosomal localization of transferrin receptor-positive recycling endosome. Wakana, Y., Takai, S., Nakajima, K.-I., Tani, K., Yamamoto, A., Watson, P., Mol. Biol. Cell 19, 4224-4237. Stephens, D. J., Hauri, H.-P. and Tagaya, M. (2008). Bap31 is an itinerant Inoue, H., Baba, T., Sato, S., Ohtsuki, R., Takemori, A., Watanabe, T., Tagaya, M. protein that moves between the peripheral endoplasmic reticulum (ER) and a and Tani, K. (2012). Roles of SAM and DDHD domains in mammalian juxtanuclear compartment related to ER-associated degradation. Mol. Biol. Cell intracellular phospholipase A1 KIAA0725p. Biochim. Biophys. Acta 1823, 19, 1825-1836. 930-939. Whiteheart, S. W., Brunner, M., Wilson, D. W., Wiedmann, M. and Rothman, J. E. Jahn, R. and Scheller, R. H. (2006). SNAREs – engines for membrane fusion. Nat. (1992). Soluble N-ethylmaleimide-sensitive fusion attachment proteins (SNAPs) Rev. Mol. Cell Biol. 7, 631-643. bind to a multi-SNAP receptor complex in Golgi membranes. J. Biol. Chem. 267, Kawase, K., Shibata, M., Kawashima, H., Hatsuzawa, K., Nagahama, M., 12239-12243. Tagaya, M. and Tani, K. (2003). Gaf-1b is an alternative splice variant of Gaf-1/ Whiteheart, S. W., Griff, I. C., Brunner, M., Clary, D. O., Mayer, T., Buhrow, S. A. Rip11. Biochem. Biophys. Res. Commun. 303, 1042-1046. and Rothman, J. E. (1993). SNAP family of NSF attachment proteins includes a Lu, Y., Zhang, Z., Sun, D., Sweeney, S. T. and Gao, F.-B. (2013). Syntaxin 13, a brain-specific isoform. Nature 362, 353-355. ̈ ̈ genetic modifier of mutant CHMP2B in frontotemporal dementia, is required for Wimmer, C., Hohl, T. M., Hughes, C. A., Muller, S. A., Sollner, T. H., Engel, A. and autophagosome maturation. Mol. Cell 52, 264-271. Rothman, J. E. (2001). Molecular mass, stoichiometry, and assembly of 20 S Mallard, F., Tang, B. L., Galli, T., Tenza, D., Saint-Pol, A., Yue, X., Antony, C., particles. J. Biol. Chem. 276, 29091-29097. Wong, S. H., Xu, Y., Zhang, T. and Hong, W. (1998). Syntaxin 7, a novel syntaxin Hong, W., Goud, B. and Johannes, L. (2002). Early/recycling endosomes-to- member associated with the early endosomal compartment. J. Biol. Chem. 273, TGN transport involves two SNARE complexes and a Rab6 isoform. J. Cell Biol. 375-380. 156, 653-664. Xu, J., Xu, Y., Ellis-Davies, G. C., Augustine, G. J. and Tse, F. W. (2002). Marz, K. E., Lauer, J. M. and Hanson, P. I. (2003). Defining the SNARE complex Differential regulation of exocytosis by alpha- and beta-SNAPs. J. Neurosci. 22, binding surface of alpha-SNAP: implications for SNARE complex disassembly. 53-61. J. Biol. Chem. 278, 27000-27008. Ying, H., Zheng, H., Scott, K., Wiedemeyer, R., Yan, H., Lim, C., Huang, J., Mayer, B. J., Hirai, H. and Sakai, R. (1995). Evidence that SH2 domains promote Dhakal, S., Ivanova, E., Xiao, Y. et al. (2010). Mig-6 controls EGFR trafficking processive phosphorylation by protein-tyrosine kinases. Curr. Biol. 5, 296-305. and suppresses gliomagenesis. Proc. Natl. Acad. Sci. USA 107, 6912-6917. ̈ McMahon, H. T. and Sudhof, T. C. (1995). Synaptic core complex of synaptobrevin, Yonekawa, S., Furuno, A., Baba, T., Fujiki, Y., Ogasawara, Y., Yamamoto, A., syntaxin, and SNAP25 forms high affinity alpha-SNAP binding site. J. Biol. Chem. Tagaya, M. and Tani, K. (2011). Sec16B is involved in the endoplasmic reticulum 270, 2213-2217. export of the peroxisomal membrane biogenesis factor peroxin 16 (Pex16) in Peter, F., Wong, S. H., Subramaniam, V. N., Tang, B. L. and Hong, W. (1998). mammalian cells. Proc. Natl. Acad. Sci. USA 108, 12746-12751. Alpha-SNAP but not gamma-SNAP is required for ER-Golgi transport after vesicle Zhang, T., Wong, S. H., Tang, B. L., Xu, Y. and Hong, W. (1999). Morphological budding and the Rab1-requiring step but before the EGTA-sensitive step. J. Cell. and functional association of Sec22b/ERS-24 with the pre-Golgi intermediate Sci. 111, 2625-2633. compartment. Mol. Biol. Cell 10, 435-453. Journal of Cell Science

2794