Modulation of N-Ethylmaleimide-sensitive Factor Activity upon Amino Acid Deprivation*

Adaptation of eukaryotic cells to changing environmental conditions entails rapid regulation of protein targeting and transport to specific organelles. Such adaptation is well exemplified in mammalian cells exposed to nitrogen starvation that are triggered to form and transport autophagosomes to lysosomes, thus constituting an inducible intracellular trafficking pathway. Here we investigated the relationship between the general secretory machinery and the autophagic pathway in Chinese hamster ovary cells grown in the absence of amino acid. Utilizing VSVG-YFP (vesicular stomatitis virus G protein fused to yellow fluorescent protein) and norepinephrine as markers for constitutive and regulated exocytosis, respectively, we found that secretion is attenuated in cells grown in media lacking amino acid. Such decrease in exocytosis stems from partial inhibition of N-ethylmaleimide-sensitive factor ATPase activity, which in turn causes an accumulation of SNARE complexes at both the Golgi apparatus and the plasma membrane of the starved cells. These findings expose a novel cellular strategy to attenuate secretion of proteins under conditions of limited amino acid supply.

Adaptation of eukaryotic cells to changing environmental conditions entails rapid regulation of protein targeting and transport to specific organelles. Such adaptation is well exemplified in mammalian cells exposed to nitrogen starvation that are triggered to form and transport autophagosomes to lysosomes, thus constituting an inducible intracellular trafficking pathway. Here we investigated the relationship between the general secretory machinery and the autophagic pathway in Chinese hamster ovary cells grown in the absence of amino acid. Utilizing VSVG-YFP (vesicular stomatitis virus G protein fused to yellow fluorescent protein) and norepinephrine as markers for constitutive and regulated exocytosis, respectively, we found that secretion is attenuated in cells grown in media lacking amino acid. Such decrease in exocytosis stems from partial inhibition of N-ethylmaleimide-sensitive factor ATPase activity, which in turn causes an accumulation of SNARE complexes at both the Golgi apparatus and the plasma membrane of the starved cells. These findings expose a novel cellular strategy to attenuate secretion of proteins under conditions of limited amino acid supply.
Specific recognition between an intracellular vesicle carrying cargo molecules and its appropriate target membrane involves the interaction between v-SNAREs, 1 integral membrane proteins located on the vesicle, and t-SNAREs, located at the target membrane (1). These interactions form a SNARE "core complex," which consists of four entwined ␣-helix bundles of typically three Q-SNARE helices and one R-SNARE helix, a classification based on conserved glutamine or arginine residues at the center of their SNARE-binding domain. The SNARE core complex is stabilized mainly by hydrophobic interactions between the four helices and by a central ionic layer consisting of one arginine and three glutamine residues contributed by each of the four ␣-helices (2). Formation of trans-complexes of SNAREs from opposing membranes yields a close, stable proximity between the two membranes, which facilitates overcoming the energy barrier required for membrane fusion (3)(4)(5). Furthermore, using three sets of functionally identified yeast t-SNAREs to mediate the fusion of ER-derived transport vesicles with the Golgi, the homotypic fusion of vacuoles, and the fusion with the plasma membrane (PM), it was demonstrated that isolated SNARE proteins encode compartmental specificity and mediate the actual fusion event (6 -8). Additional proteins are required for targeting and tethering of these transport vesicles with their specific targets (9).
The hexameric ATPase N-ethylmaleimide-sensitive factor (NSF) utilizes ATP hydrolysis to dissociate cis-SNARE complexes after membrane fusion, allowing the individual SNARE proteins to be recycled for subsequent rounds of fusion (10,11). Whereas specific v-and t-SNAREs are associated with each intercompartmental transport step, NSF is a general cytosolic factor that can disassemble SNARE complexes from most intracellular transport steps. The ATPase activity of NSF is enhanced by ␣-SNAP (soluble NSF attachment protein), which mediates NSF binding to different SNARE complexes (12), and possibly by other factors such as GATE-16 (13,14) and rab-6 (15). Notably, evidence for direct negative regulation of NSF activity in vivo has been suggested by Matsushita et al. (16), whereby S-nitrosylation of specific cysteine residues on NSF leads to attenuation of triggered exocytosis of endothelial granules. However, it is not known yet whether NSF S-nitrosylation controls other cellular membrane fusion processes such as constitutive exocytosis or vesicular transport along the secretory pathway.
Direct regulation of vesicular trafficking is crucial not only for the spatially and temporally controlled secretion of bioactive molecules but also for controlling the levels of various molecules (e.g. receptors, transporters) on the PM (17). A variety of regulatory events that operate at different steps of vesicular transport control the trafficking along specific cellular pathways in response to different signals. Developmental, environmental, and cell cycle-related signals have differential effects on different vesicular transport pathways. For example, transport along the secretory pathway in Xenopus oocytes is blocked between the trans-Golgi and the PM during meiotic maturation (18). On the other hand, transport of special vesicles to lysosomes is enhanced upon amino acid deprivation in a process known as autophagy (19).
Autophagy is a bulk protein degradation process in which newly formed double membrane vesicles, termed autophagosomes, deliver cytoplasmic contents and organelles for lysosomal degradation (19). There are at least three types of autophagy: macroautophagy (referred to as autophagy hereafter), microautophagy (20), and chaperon-mediated autophagy (21). Autophagy, a relatively understudied membrane-trafficking * This work was supported in part by the Israel Science Foundation and the Weizmann Institute Minerva Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. process, may be regarded as a unique vesicular transport pathway that is triggered in response to stress conditions such as nitrogen starvation (19). Recent studies implicated transport factors such as Sec18p (NSF yeast homologue), SNAREs, Rabs, and members of GATE-16 family in fusion steps associated with autophagy (22)(23)(24)(25)(26). In the present study we have investigated whether deprivation of amino acid also affects other vesicular transport pathways such as constitutive and regulated secretion. By comparing exocytosis in starved versus nonstarved control cells, we conclude that the Golgi to PM transport is significantly attenuated under amino acid starvation conditions. Structural and functional assays suggest that this attenuation is directly caused by inhibition of NSF ATPase activity during starvation. Our data provide, for the first time, a molecular explanation for the attenuation of exocytosis under these physiological stress conditions.

EXPERIMENTAL PROCEDURES
Cell Culture and Induction of Amino Acid Starvation-CHO cells were grown in ␣-MEM containing 10% FCS, 100 units/ml penicillin, and 100 g/ml streptomycin, either on plates or in suspension. PC12 cells were grown in DMEM containing 6% FCS, 6% horse serum, 100 units/ml penicillin, and 100 g/ml streptomycin. To induce autophagy by amino acid deprivation, cells were washed three times with Earl's balanced salt solution (EBSS) and then incubated in EBSS (supplemented with vitamins and pyruvate to the same concentrations present in the rich media) for the indicated time periods at 37°C. As a control, washed cells were incubated with ␣-MEM or DMEM in the absence of FCS for the indicated starvation periods.
Degradation and Secretion Assays-Measurement of long-lived protein degradation in CHO cells was based on an assay previously described by Ogier et al. (27). Briefly, CHO cells plated on 6-well plates were prelabeled with [ 14 C]valine (Amersham Biosciences) for 24 h. Cells were then washed three times with phosphate-buffered saline and then preincubated for 1 h in either ␣-MEM or EBSS, both containing 0.1% bovine serum albumin and 10 mM cold valine. After 1 h of incubation, the culture media were replaced with identical fresh media, and the cells were incubated for an additional 2 or 4 h. The media were collected and centrifuged for 3 min at 2200 rpm, and the supernatant was precipitated in 10% trichloroacetic acid. Total cells radioactivity was measured after lysing the cells in 0.1 M NaOH. Degradation and secretion of prelabeled proteins were measured as the ratio between the radioactivity found in the trichloroacetic acid-soluble fraction (free [ 14 C]valine released to the medium) or in the trichloroacetic acid-insoluble fraction (incorporated [ 14 C]valine released to the medium) and the total cells radioactivity, respectively.
VSVG Trafficking-CHO cells cultured in Lab-Tek chambered coverglass system (Nunc) were transfected at 40°C with VSVG tsO45-YFP (45) by Lipofectamine according to manufacturer's protocol. After 24 h of incubation, cells were washed three times with prewarmed EBSS and further incubated at 40°C for 3 h in control or starvation media. Cells were then shifted to 32°C to trigger transport at t ϭ 0. A series of fluorescent microscopy images was made at the time points indicated. To analyze the progression of VSVG along the secretory pathway under favorable versus starvation conditions, cells were scored as showing Golgi localization of VSVG when the Golgi staining intensity was at least strong as that of the ER. Cells were scored as showing PM localization of VSVG in this analysis when PM staining was clearly visualized. Representatives for each case are shown in Fig. 2.
Norepinephrine Release Assay-PC12 cells were grown in poly-Llysine-coated 6-well plate for 2-3 days until they reached 50 -70% confluence. Cells were loaded at 37°C for 16 h in DMEM containing 6% FCS, 6% horse serum and 1.5 Ci of [ 3 H]NE (ARC Inc.)/well, rinsed with phosphate-buffered saline, and then chased in control or starvation medium for additional 3.5 h. To prevent spontaneous leakage of [ 3 H]norepinephrine (NE) from starved cells resulting from decreased ATP cellular level, extra glucose was added to the EBSS medium to the same level found in DMEM. PC12 cells were then incubated in low K ϩ medium (10 mM Hepes, 5 mM KCl, 145 mM NaCl, 2 mM CaCl 2 , 10 mM glucose, pH 7.4) for 10 min at 37°C prior to the induction of secretion with high K ϩ medium (10 mM Hepes, 55 mM KCl, 95 mM NaCl, 2 mM CaCl 2 , 10 mM glucose, pH 7.4). [ 3 H]NE release was assessed in time course experiments in which the medium was collected and replaced every 2 min. Media samples were centrifuged at 3000 rpm for 3 min, and supernatant radioactivity was determined by liquid scintillation counting. At the end of the experiment, the cells were solubilized in 1 ml of 0.25 N NaOH, and the lysates were carefully collected. [ 3 H]NE release was calculated as the ratio between the medium radioactivity and the total cells radioactivity (sum of all media counts and the radioactivity remaining in the cells at the end of the experiment).
Preparation of Cytosol and Membrane Extracts-Cells were washed twice with phosphate-buffered saline and then twice in a homogenization buffer containing 0.25 M sucrose, 25 mM Tris-HCl, pH 7.4, and 50 mM KCl. The cells were then homogenized in the same buffer containing 1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 2 g/ml aprotinin, and 2 M pepstatin A using a Balch homogenizer (for cells grown in suspension) or Dounce homogenizer (for cells grown on plates). The homogenates were centrifuged for 5 min at 2500 rpm, and the supernatant was further centrifuged for 30 min at 200,000 ϫ g. The supernatant containing the cytosol was collected. The pellet, containing total membranes, was resuspended in extraction buffer containing 20 mM Hepes, pH 7.0, 20 mM KCl, and 0.5% Triton X-100.
To obtain cytosol and membrane extracts in the presence of ATP, CHO cells were washed twice with phosphate-buffered saline, then twice in a homogenization buffer containing 20 mM PIPES, pH 7.2, 10 mM MgCl 2 , 5 mM ATP, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 0.1 M KCl, and homogenized and fractionated as described above.
SNARE Complex Disassembly Assays-To compare SNARE complex levels in control versus starved cells, membrane extracts from CHO or PC12 cells were incubated in SDS-sample buffer at either 30°C or 100°C for 5 min and then subjected to immunoblotting. To examine the capacity of cytosolic fractions to dissociate SNARE complexes, rat brain membranes containing SNARE complexes were pretreated on ice with 1 mM N-ethylmaleimide (NEM) for 10 min followed by treatment with 2 mM dithiothreitol for 15 min. NEM-treated membranes were incubated with 50 g of cytosol obtained from control or starved CHO cells in a reaction buffer containing 0.5 g of ␣-SNAP, 0.5 mM ATP, 2 mM MgCl 2 , and ATP regeneration buffer at 30°C for 30 min. In the indicated reactions, 1.2 g of recombinant NSF was added to the disassembly assay. The membranes were then isolated by centrifugation (14,000 rpm, 10 min at 4°C), resuspended in SDS-sample buffer, and processed for immunoblotting without prior boiling unless mentioned.
Proteolytic Digestion-Recombinant NSF proteins and cytosolic fractions obtained from starved or non-starved CHO cells were incubated at 30°C with trypsin (ratio of 1:20 and 1:200 trypsin/protein (w/w), respectively). Progress of the proteolysis was assessed by removing aliquots at the indicated time points and quenching the digestion with a 5:1 (w/w) excess of soybean trypsin inhibitor over trypsin. The aliquots were subjected to 12% SDS-PAGE and analyzed by Western blot using a mixture of 2E5, 2C8, and 6E6 anti-NSF monoclonal antibodies.
Immunoprecipitation-Monoclonal antibodies (anti-NSF) and polyclonal antibodies (anti-Gos-28 or anti-syntaxin-5) were covalently coupled by dimethyl pimelimidate (Sigma) to protein G-and protein Aagarose beads (Santa Cruz Biotechnology), respectively. The coupled antibodies were then incubated with either recombinant His-NSF-Myc protein (0.5 g) or cytosolic (200 g) and membrane extract fractions (200 g) obtained from control or starved cells for at least 2 h at 4°C. Beads were washed four times in phosphate-buffered saline, the bound material was eluted by 2% SDS at 95°C, and the eluates were subjected to Western blot analysis. In some experiments, recombinant His-NSF-Myc fractions (0.5 g) were preincubated with cytosolic fractions (50 g) obtained from control or starved CHO cells at 30°C for 30 min prior to the immunoprecipitation.
ATPase Activity Assay-Prior to the ATPase activity assay, NSF (6 g) was preincubated with cytosolic fractions (40 g) obtained from control or starved CHO cells at 30°C for 20 min in a total volume of 100 l of 25 mM Tris-HCl, pH 7.4, 50 mM KCl, 2 mM MgCl 2 and 0.5 mM ATP. The assay was initiated by adding residual amounts of [␥-32 P]ATP (0.25 Ci/sample), carried out for 30 min at 30°C, and stopped by adding 100 l of ice-cold perchloric acid to a final concentration of 10%. Nucleotides were bound to 500 l of ice-cold charcoal, and samples were centrifuged at 14,000 rpm for 15 min at 4°C. The resulting supernatants containing released 32 P i were quantified by liquid scintillation counting. Background activities of control and starved cytosol alone were subtracted (0.08 and 0.079 mol of P i /mg of protein/h for cytosolic fractions obtained from control or starved cells, respectively).

RESULTS
Secretion Is Inhibited in Response to Amino Acid Deprivation-Autophagy triggered by starvation may be regarded as a special case of membrane-trafficking process that shares some of the components utilized by other intracellular membranetrafficking pathways. In an attempt to study the relationship between the autophagic and the secretory pathways, we utilized [ 14 C]valine-prelabeled tissue culture cells to examine the effects of amino acid starvation on intracellular protein degradation and simultaneously on protein secretion. In this system the autophagic activity was determined by monitoring the bulk protein degradation of long-lived proteins (27), whereas exocytosis was determined by monitoring the release of secreted proteins from the cells. For this purpose, prelabeled CHO cells were incubated for different time periods in either ␣-MEM (without FCS) or EBSS (starvation medium), and the levels of secreted labeled proteins and amino acid into the media were determined. As depicted in Fig. 1A, bulk degradation of longlived proteins, measured by the release of trichloroacetic acidsoluble [ 14 C]valine from the cells, was about 50% higher in cells incubated in medium lacking amino acid compared with control medium, indicating significant enhancement of autophagy in starved cells. Concomitantly, the radioactivity measured in the trichloroacetic acid-insoluble pellet was significantly reduced (ϳ30%) in cells deprived of amino acid (Fig. 1B). Similar results were obtained when CHO cells were prelabeled with [ 35 S]methionine, as well as in different cell lines such as COS-7 and PC12 cells that were treated similarly (data not shown). Taken together, these results indicate that during amino acid starvation protein degradation is stimulated, whereas the constitutive secretion of newly synthesized proteins is largely inhibited.
Vesicular Transport to the Plasma Membrane Is Inhibited during Starvation-To identify the step along the secretory pathway that is most affected by amino acid starvation, we utilized the well studied vesicular stomatitis virus ts045 G protein fused to yellow fluorescent protein (VSVG-YFP) to monitor trafficking through the various compartments of the secretory pathway in living cells. We took advantage of the fact that at 40°C the ts045 VSVG mutant protein is retained within the ER, whereas upon a shift to 32°C it moves as a synchronous population to the Golgi complex before being transported to the PM (28). Hence, CHO cells were transfected with VSVG-YFP, incubated in normal or starvation medium for 3 h at 40°C, and then shifted to 32°C to trigger transport. The transport of the VSVG-YFP ts045 protein was monitored by confocal microscopy. At each of several selected time points, the percentage of cells in which the VSVG-YFP protein was localized to the ER, Golgi, or the PM was calculated. As shown in Fig. 2, the VSVG protein accumulated in the Golgi apparatus in about 60% of the control cells within a 5-min incubation period at 32°C and in 100% of these cells within 20 min. During this period, no significant difference in the rate of transport of VSVG protein from the ER to the Golgi was detected in the starved cells. However, within 60 min the VSVG protein was detected in the PM of more than 60% of the control cells, whereas only 25% of the starved cells exhibited PM labeling. From this analysis we conclude that although ER-to-Golgi transport is only marginally affected by amino acid starvation, the transport from the Golgi apparatus to the PM is significantly attenuated in starved cells.
The effect of amino acid deprivation on exocytosis was studied further in PC12 cells, a rat adrenal phaeochromocytoma cell line, which expresses several well defined neuronal properties in culture, including the regulated secretion of neurotransmitters. To examine the effects of amino acid starvation on exocytosis in this cell line, release of [ 3 H]norepinephrine from prelabeled cells was measured in control versus starved cells. Hence, prelabeled cells were incubated for 3.5 h in either DMEM or EBSS supplemented with glucose (starvation media) and then subjected to time course secretion analysis. As depicted in Fig. 3, the basal rates of NE release were low and similar for both control and starved cells. However, when NE secretion was stimulated by directly depolarizing the PC12  (starvation medium, black). The culture media were collected and subjected to trichloroacetic acid precipitation. A, degradation of prelabeled proteins was measured as the ratio between the radioactivity found in the trichloroacetic acid-soluble fraction (free [ 14 C]valine released to the medium) and total cells radioactivity. B, secretion of prelabeled proteins was measured as the ratio between the radioactivity found in the trichloroacetic acid-insoluble fraction and total cells radioactivity. Data are the means Ϯ S.D. of triplicates from a representative experiment repeated three times. cells with elevated extracellular K ϩ ions (55 mM KCl), 14% of the [ 3 H]NE content was secreted within the first 2 min and cumulatively up to 31% within 12 min. Under these conditions, however, secretion of NE from the starved cells was inhibited by about 32%, indicating that triggered exocytosis too is sensitive to amino acid starvation.
Disassembly of SNARE Complexes Is Inhibited during Amino Acid Starvation-A normal vesicular transport cycle requires rapid and regulated disassembly of SNARE complexes, mediated by the ATPase activity of NSF and its co-factor ␣-SNAP (reviewed in Ref. 29). Exocytosis in PC12 cells, for example, requires the formation and subsequent dissociation of a specific SNARE complex, including the vesicle-associated membrane protein (VAMP) and the PM proteins syntaxin and SNAP-25. To examine whether lack of amino acid affected the dissociation of these complexes, we relied on the fact that SNARE complexes are SDS-resistant at 37°C and thus can be detected as high molecular weight complexes by immunoblotting (30). As shown in Fig. 4A, the level of SNARE complexes found in the starved cells was much higher than in the control cells, indicating that disassembly of SNARE complexes involved in exocytosis is inhibited under amino acid starvation conditions.
To further examine the effect of amino acid deprivation on the oligomeric state of SNARE molecules, we isolated membrane extracts from control and starved CHO cells and analyzed the oligomeric states of GOS-28, a v-SNARE participating in intra-Golgi transport, and its cognate t-SNARE, syntaxin-5. Consistent with the results obtained with the PC12 cells (Fig. 4A), we observed significantly more GOS-28/syntaxin-5 SDS-resistant complexes in the starved cells (Fig. 4B), indicating that the disassembly of SNARE complexes containing GOS-28 and syntaxin-5 was impaired under these conditions. The effect of starvation on the interaction between GOS-28 and syntaxin-5 was further analyzed by co-immunoprecipitation experiments. For this purpose, agarose-protein A beads coupled to anti-syntaxin-5 or anti-GOS-28 antibodies, as indicated, were mixed with membrane extracts obtained from control or starved CHO cells, and the eluted material was subjected to Western blot analysis. The results presented in Fig. 4C demonstrate that the interaction between GOS-28 and syntaxin-5 in a complex containing NSF is increased during starvation.
NSF Activity Is Inhibited in Starved Cells-The accumulation of different SNARE complexes in the starved cells may best be explained by a reduction in NSF activity. To test this hypothesis, we determined the ability of cytosolic fractions obtained from control or starved cells to dissociate endogenous SNARE complexes. Rat brain membranes were first treated with NEM to abolish intrinsic NSF activity. Next, the treated membranes were incubated with cytosol obtained from starved or control CHO cells, and the level of SNARE complexes was analyzed by Western blotting. As shown in Fig. 5, incubation of NEM-treated membranes with cytosol obtained from control cells resulted in a significant dissociation of high molecular weight SNARE complexes detected in this system. In contrast, no substantial dissociation of SNARE complexes was observed upon incubation with cytosol obtained from starved cells, suggesting that NSF activity was inhibited in this fraction. Moreover, the addition of recombinant NSF to the starved cytosol recovered most of the SNARE complex disassembly activity of this fraction, indicating that the inhibition of NSF in starved cells led to the accumulation of SNARE complexes in these cells. Notably, the recovery of the disassembly activity was not complete as compared with the disassembly activity found in the control cytosol, suggesting partial inactivation of the recombinant NSF proteins by the starved cytosolic fraction (see below).
NSF Undergoes a Conformational Change in Response to Starvation-NSF is a homohexamer in which each of the protomers consists of three domains: an N-terminal domain, NSF-N, which is responsible for the interaction with the ␣-SNAP-SNARE complex, and two homologous ATP-binding domains, NSF-D1 and NSF-D2 (31,32). ATP binding to the D1 domain is crucial for NSF ATPase activity, leading to dissociation of the SNARE complex. In response to ATP binding to its D1 domain, NSF undergoes a major conformational change (33). Thus, in the presence of ATP, recombinant NSF is relatively resistant to limited proteolysis by trypsin, whereas in a buffer lacking ATP it is rapidly degraded (Fig. 6A and Ref. 33).
To determine whether the inability of NSF to disassemble SNARE complexes was correlated with structural changes induced by starvation conditions, cytosols obtained from control and from starved CHO cells were treated with a mild trypsin concentration for different time periods, and the proteolytic profile of NSF was analyzed by Western blot using anti-NSF antibodies. It turned out that relative to NSF from starved cells, NSF obtained from cytosol of non-starved control cells was more sensitive to the mild trypsin treatment (Fig. 6B). In addition, the proteolytic products of ϳ35 kDa appeared much more resistant in cytosolic NSF from starved cells. Accordingly, the proteolytic profile of NSF obtained from starved cells was similar to that obtained from NSF in ATP buffer, whereas the profile of NSF obtained from the control cells was similar to that of NSF in a solution of low ATP concentration. We therefore propose that amino acid starvation induces changes in cytosolic NSF, stabilizing the ATP-bound form of the protein. Notably, in different cytosols NSF is probably found in equilibrium between its two states, thus obliterating the differences in the trypsinolysis patterns of the crude cytosols ( Fig. 6B) relative to the pure recombinant protein (Fig. 6A).
We also tested whether anti-NSF monoclonal antibodies could distinguish between the different conformations of NSF. Hence, we examined the ability of monoclonal anti-NSF antibodies to immunoprecipitate recombinant NSF in a nucleotide dependent manner. As shown in Fig. 7A, recombinant NSF immunoprecipitated more efficiently when incubated with ATP. When cytosolic fractions obtained from starved and control CHO cells were tested in this system, significantly more NSF was precipitated in the cytosol obtained from the starved cells, although no difference was detected in the total level of NSF found in the different cytosolic fractions (Fig. 7B, left  panel). Similar results were obtained with cytosolic fractions from control versus starved PC12 cells (data not shown). To further determine whether NSF conformational changes triggered by amino acid starvation were related to the nucleotidebound state of NSF, cytosolic fractions obtained from starved and control cells were prepared in the presence of 5 mM ATP and then subjected to immunoprecipitation by anti-NSF antibodies. In the presence of high ATP concentration, no difference in the amount of immunoprecipitated NSF from the different cytosolic fractions was observed (Fig. 7B, right panel). Notably, the amounts of NSF precipitated in ATP-rich cytosol from control and starved cells were similar to the amount precipitated from starved cells in the absence of ATP, suggesting that under starvation conditions NSF is mostly stabilized in a nucleotide-bound state (Fig. 7B).
Next we utilized the immunoprecipitation assay to study the conformational changes of NSF in a cell-free system. To that end, recombinant NSF fused at its C terminus to the Myc epitope tag was incubated with cytosolic fractions obtained from control or starved CHO cells for 30 min at 30°C followed by immunoprecipitation with anti-NSF monoclonal antibodies. As shown in Fig. 7C, significantly more recombinant NSF immunoprecipitated after treatment with the cytosol ob- FIG. 5. NSF activity is inhibited in starved cells. NEM-treated rat brain membranes were incubated with CHO cytosol obtained from starved (starv.) or non-starved (cont., control) cells at 30°C for 30 min. In the indicated reactions, 1.2 g of recombinant NSF was added to the disassembly assay. Membranes were then isolated by centrifugation, and SNARE complex disassembly was analyzed by Western blotting using anti-Syntaxin antibodies. Immunoprecipitates were analyzed by Western blotting using anti-NSF monoclonal antibodies. NSF was not detected in immunoprecipitates obtained by monoclonal anti-VSVG antibodies. Total represents 25% of each fraction used in these experiments. C, recombinant NSF undergoes a conformational change in vitro. His-NSF-Myc was preincubated with cytosolic fractions obtained from control or starved CHO cells for 30 min at 30°C and then subjected to immunoprecipitation with anti-NSF antibodies. Immunoprecipitates were analyzed by Western blotting using anti-NSF and anti-Myc monoclonal antibodies. The asterisk represents the endogenous NSF precipitated from starved cells. IB, immunoblot (Western). tained from starved CHO cells, indicating that this cytosol maintained its ability to induce changes in NSF in vitro. This cytosolic activity was abolished upon heat inactivation, implying that a soluble protein(s) was responsible for the apparent NSF conformational change. Notably, these results explain the inability of recombinant NSF to fully recover the defective disassembly activity of endogenous NSF in starved cytosol shown in Fig. 5, as the freshly added enzyme may be partially modified in vitro.
The finding that recombinant NSF can be altered in vitro upon incubation with cytosol obtained from starved cells was further utilized to determine whether the ATPase activity of NSF was inhibited under these conditions. Accordingly, recombinant NSF was preincubated with cytosolic fractions from control versus starved cells and then subjected to ATPase activity assay. Notably, the endogenous ATPase activities (0.08 and 0.079 mol of P i /mg of protein/h for the cytosolic fractions obtained from control or starved cells, respectively) were subtracted. As shown in Fig. 8A, only cytosol obtained from starved cells, but not from control cells, reduced the NSF ATPase by 50%. Finally, we tested the hypothesis that NSF was blocked in an ATP-bound conformation during starvation by directly analyzing the levels of the nucleotides bound to NSF from control versus starved cells. CHO cells were first metabolically labeled with [ 32 P]orthophosphate and then incubated in ␣-MEM or EBSS medium for 2.5 h. NSF from these cells was then immunoprecipitated by monoclonal anti-NSF antibodies, and the eluted nucleotides were separated on TLC. When precipitated with specific antibodies, NSF eluted from control and starved cells was predominantly bound to ATP (Fig. 8B). Using this method, cell lysates from control and starved cells yielded little if any labeled adenosine nucleotides after immunoprecipitation with anti-VSVG antibodies, the control assay in this experiment. Taken together, our results are consistent with the notion that NSF ATPase activity is partially inhibited during starvation and therefore that the protein is found in an ATPbound state. DISCUSSION The basic mechanism of intracellular membrane trafficking is not only evolutionary conserved between yeast and man but is also remarkably similar in different cellular membrane transport pathways. Autophagy is believed to represent a special membrane-trafficking route for the delivery of cytosolic proteins and organelles for degradation in lysosomes (34). Gathering data indicate that many of the factors involved in vesicular transport also play an important role in formation and subsequent transport of autophagosomes to their target membranes (22)(23)(24)(25)(26). However, the relationships between vesicular transport and autophagy are yet unknown. In the present study we show that amino acid starvation induces autophagy while attenuating exocytosis. This attenuation is accompanied by the accumulation of SNARE complexes at both the Golgi apparatus and the PM, caused by inhibition of NSF ATPase activity. Together, these findings reveal a novel strategy utilized by cells to attenuate secretion of proteins under conditions in which amino acid supply is limited. Considering the changes in the equilibrium between biosynthesis and catabolism of proteins during nitrogen starvation, the controlled inhibition of different steps along the secretory pathway may be utilized by cells to adapt to the shortage in amino acid by rendering the secreted proteins available for lysosomal degradation mediated by autophagy.
Exocytosis of bioactive molecules is essential for many physiological processes including neurotransmission, immune response to infections, wound healing, and control of blood glucose levels. These processes require controlled fusion between transport vesicles or storage granules with the PM. Such fusion events are basically mediated by the assembly of v-SNAREs on the vesicle with their cognate t-SNAREs on the target membrane (35). This process involves a large set of additional proteins that provide spatial and temporal control to assure the correct response for exocytic signals. Once fusion is completed, the v-and t-SNAREs are found in an extremely stable fourhelix bundle on the target membrane. The physiological role of NSF is to allow the dissociation of such cis-SNARE complexes and thereby subsequent membrane fusion events. This process occurs not only at the PM but also during other intracellular membrane fusion events. The unexpected identification of NSF as a target for regulation of vesicular transport during starvation suggests that other membrane-trafficking processes may be attenuated in starved cells. Of much interest is the mechanism underlying formation of autophagosomes and their subsequent fusion with lysosomes under nitrogen starvation. At least three different fusion steps may take place during this process (19). First, autophagosomal formation requires homotypic fusion of the cup-shaped isolation membrane in order to form the double membrane structures. Second, autophagosomes mature by fusing with endosomes or endosome-derived vesicles to form amphisomes (AVi) (36). Finally, these AVi fuse with lysosomes. The mechanism involved in these fusion processes in mammalian cells remains unclear. Studies in yeast FIG. 8. NSF ATPase activity is inhibited by amino acid deprivation. A, His-NSF-Myc was preincubated with cytosolic fractions from control or starved cells and analyzed for its ATPase activity by measuring the release of 32 P from [␥-32 P]ATP. The ATPase of the cytosolic fractions alone (0.08 and 0.079 mol of P i /mg of protein/h for cytosolic fractions obtained from control or starved cells, respectively) was subtracted. Plotted values represent NEM-sensitive activities. B, CHO cells were metabolically labeled with [ 32 P]orthophosphate and then incubated in ␣-MEM or EBSS for 2.5 h. NSF proteins with tightly bound nucleotides were immunoprecipitated (IP) from these cytosolic fractions using anti-NSF antibodies, and the eluted nucleotides were separated on TLC.
indicated that Sec-18 activity is not required for autophagosome formation, although it is essential for their subsequent fusion with the vacuole (24). Moreover, the v-SNARE Vti1p, localized on autophagosomes outer membrane, and the t-SNAREs Vam3p and Vam7p, localized on the vacuole membrane, were suggested to be involved in this fusion process in yeast (23,37,38). The only mammalian factor known to reside on autophagosomes membrane is LC-3, which is likely to take part in one or more of these fusion steps (26). LC-3 shares high similarity with GATE-16, a low molecular weight factor involved in intra-Golgi transport (14). GATE-16 function is coupled with NSF ATPase-independent activity, shown to be involved in homotypic fusion between fragmented Golgi membranes (13). Taking these findings together, it is tempting to speculate that similarly to its function in reassembly of mitotic Golgi membranes, the ATPase activity of NSF is not required for autophagosomes/lysosomes fusion processes. Alternatively, under conditions of reduced NSF activity, these processes may involve other AAA ATPase fusion factors. For example, P97, a close homologue of NSF that mediates several homotypic membrane fusion events (39,40), may mediate homotypic fusion of autophagosomes. Another AAA ATPase protein that may be involved in this process is SKD1, an essential factor for endosomal trafficking (41). Thus, overexpression of a dominant negative SKD1 mutant defective in its ATP hydrolysis results in accumulation of autophagosomes that cannot fuse with the lysosomes (42).
NSF forms homohexamers in which each subunit is composed of three domains (12): an N-terminal domain that interacts with the SNAP family, found associated with the SNARE complexes; a D1 domain that hydrolyzes ATP and provides the mechanical force for SNARE complex disassembly; and a D2 domain that tightly binds ATP and is responsible for the hexamerization of NSF subunits. Nucleotide binding to the D1 domain induces dramatic conformational changes in the Nterminal domain. These changes are assumed to be the driving force for the capability of NSF to stimulate the dissociation of cis-SNARE complexes. In the present study, we show that amino acid starvation stabilizes the ATP-bound conformation of NSF by inhibiting its ATPase activity. We propose that this effect may block the ability of NSF to change further its conformation upon ATP hydrolysis and therefore may hinder its ability to dissociate SNARE complexes. The exact mechanism by which starvation induces such an effect on NSF is yet unclear. The conformational change in NSF probably does not simply stem from changes in the levels of available ATP, as no significant differences between ATP levels were found in cytosols obtained from control versus starved CHO cells. 2 Interestingly, Matsushita et al. (16) have recently reported for aortic endothelial cells that S-nitrosylation of NSF induced by nitric oxide inhibits the exocytosis of Weibel-Palade bodies. The effects reported here for amino acid starvation, including inhibition of exocytosis, accumulation of SNARE complexes, and inhibition of NSF, resemble those reported for nitric oxide. It is therefore conceivable that similar molecular mechanisms may regulate both processes. Indeed, it has been reported that nitrogen starvation is accompanied by changes in the redox state (43), which in turn may affect essential cysteine residues on NSF. Further study is required to determine whether nitric oxide or other reactive oxygen species molecules are involved in the regulation of NSF activity during starvation.
The fact that NSF, being a general fusion factor, is a target for regulating exocytosis under starvation conditions is rather surprising. It should be noted, however, that amino acid starvation induces only partial inhibition of the secretory pathway, predominantly affecting transport to the PM. Moreover, it is possible that the affected steps are those that have a particularly high demand for NSF, and partial inhibition of NSF activity may therefore result in a selective effect. In accord with this suggestion, reduced levels of NSF expression have been detected in schizophrenic patients (44). Alternatively, NSF inactivation under amino acid starvation may occur locally, at the vicinity of the affected SNARE complexes. Such a scenario may require the involvement of lipids and/or other components with a confined partition to specific microdomains within the membrane. In summary, our results describe for the first time a mechanism by which environmental conditions directly control the activity of NSF, a key factor required for most intracellular trafficking processes.