(Received for publication, January 23, 1997, and in revised form, March 21, 1997)
From the Department of Biochemistry, University of Wisconsin, Madison, Wisconsin 53706
The Ca2+-activated fusion of secretory vesicles with the plasma membrane responsible for regulated neurotransmitter and hormone secretion has previously been studied in permeable neuroendocrine cells, where requirements for ATP and cytosolic proteins were identified. As reported here, Ca2+-activated fusion mechanisms are also preserved following cell homogenization. The release of norepinephrine (NE) and other vesicle constituents from a PC12 cell membrane fraction was activated by micromolar Ca2+ (EC50 ~ 3 µM) and exhibited a dependence upon MgATP and cytosol. Ca2+-dependent NE release was inhibited by botulinum neurotoxins and by CAPS (Ca2+-dependent activator protein for secretion) antibody implying that syntaxin, synaptobrevin, SNAP-25 (synaptosomal-associated protein of 25 kDa), and CAPS are required for regulated exocytosis in this system. The exocytosis-competent membrane fraction consisted of rapidly sedimenting dense core vesicles associated with plasma membrane fragments. Free vesicles did not release NE either in the absence or presence of plasma membranes, indicating that only docked vesicles were competent for exocytosis under the reconstitution conditions used. A cell-free system for Ca2+-activated fusion will facilitate studies on the roles of essential proteins such as syntaxin, synaptobrevin, SNAP-25, and CAPS that act at post-docking steps in the regulated exocytotic pathway.
Cellular compartments within eukaryotic cells maintain discrete identities despite the continuous disassembly and reassembly required for the vectorial transport of lumenal and membrane proteins to specific destinations. This requires a molecular basis for membrane identity and strict regulation of the membrane fusion process (1). While recent progress toward understanding the molecular basis of vesicle targeting and fusion has been dramatic (2), the biochemical reactions responsible for membrane fusion remain poorly understood. Cell-free systems that reconstitute in vitro aspects of vesicle formation, translocation, targeting, and fusion have provided insights on the mechanisms of vesicle fission and fusion at several stages of the protein trafficking pathway (2).
A tightly regulated membrane fusion process underlies the secretion of peptide hormones and neurotransmitters from endocrine and neural cells. Exocytotic fusion of secretory vesicles with the plasma membrane is triggered by intracellular Ca2+ concentration increases into the 1-300 µM range (3). Recent studies have identified some of the proteins required for regulated fusion (see "Discussion"), but an understanding of the events in which these proteins participate and their relationship to membrane bilayer fusion remains limited due to the lack of functional cell-free in vitro assays for Ca2+-regulated fusion.
Studies of regulated fusion have instead utilized permeable cell preparations, most of which exhibit limited permeability or possess limited stability. Exceptions where exocytotic fusion reactions have been studied in vitro in cell-free systems are reconstituted sea urchin egg cortices (4) and acinar pancreatic zymogen granule-plasma membrane preparations (5); however, these systems exhibit a poorly understood partial or complete loss of Ca2+ regulation. An extension of previous studies (6, 7) of Ca2+-activated exocytosis in mechanically permeabilized PC12 neuroendocrine cells led to the development of a simple purified membrane preparation consisting of large dense core vesicles (LDCVs)1 associated with plasma membrane fragments. This preparation may be enriched for a very late stage intermediate of the regulated exocytotic pathway in which LDCVs are stably docked at the plasma membrane and in which the final steps of Ca2+-regulated fusion can be studied in vitro.
PC12 cells were cultured in 10-cm dishes as described previously (8) and incubated with 0.5 µC/ml [3H]norepinephrine (NE; 45 Ci/mmol; Amersham Corp.) for 16 h in the presence of 0.5 mM ascorbic acid. Labeled cells were washed extensively with and incubated in Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin to remove unincorporated [3H]NE, and the cells were detached from culture dishes by pipetting in cold Kglu buffer (0.02 M HEPES, pH 7.2, 0.12 M potassium glutamate, 0.02 M potassium acetate, 0.002 M EGTA, 0.1% bovine serum albumin). For plasma membrane labeling, cells were incubated with 0.5 mg/ml fluorescein isothiocyanate (FITC) in 0.1 M sodium borate pH 8.0, 0.15 M NaCl or with 20 µg/ml FITC-wheat germ agglutinin (FITC-WGA) in Dulbecco's modified Eagle's medium for 60 min at 0 °C.
Cell Homogenization and Sucrose GradientsCell homogenates
were prepared by passing cell suspensions 15 times through a narrow
clearance (10 µm) stainless steel ball homogenizer (9). Homogenates
were clarified by centrifugation at 800 × g for 10 min
twice in succession to rigorously remove broken cells (monitored
microscopically), and the postnuclear supernatant was centrifuged at
5000 × g for 10 min to recover rapidly sedimenting
LDCVs. The 5000 × g membrane pellet was washed by
resuspension in Kglu buffer and centrifugation at 10,000 × g for 10 min. Approximately 28% of the [3H]NE
in post-nuclear supernatants was rapidly sedimenting and recovered in
5000 × g pellets. The remaining free LDCVs were
quantitatively isolated from a 10,000 × g supernatant
by centrifugation at 40,000 × g for 15 min. Whereas
the majority of synaptic vesicle-like microvesicles containing
synaptophysin was present in the supernatant of the 40,000 × g centrifugation, LDCVs in PC12 cells also contain synaptophysin (see Fig. 2). For sucrose gradient purification, 5000 × g pellet fractions were collected on a pad of 2 M sucrose in 0.02 M Tris, pH 7.8, 0.15 M NaCl, 0.002 M EGTA (TNE buffer), diluted
5-fold, and loaded onto either continuous or discontinuous sucrose
gradients. Continuous gradients consisting of 9 ml of a linear gradient
of 0.25-2 M sucrose and discontinuous gradients consisting
of 2 ml each of 0.25, 0.5, 1.0, 1.5, and 2 M sucrose in TNE
were centrifuged in a Beckman SW 41 rotor at 20,000 × g for 30 min and collected from the bottom. For Western
blotting, membranes were recovered from gradient fractions by 8-fold
dilution in TNE buffer and centrifugation at 150,000 × g for 16 h. Pellets were solubilized in sample buffer
for electrophoresis on SDS-polyacrylamide gels and transferred to
nitrocellulose for immunoblotting with antibodies to chromogranin B
(Pel-Freez, Rogers, AR), synaptophysin (Boehringer Mannheim), SV2 (R.H.
Scheller, Stanford), syntaxin (Sigma), and SNAP-25 (Sternberger
Monoclonals Inc., Baltimore, MD). Rabbit polyclonal antibodies to
synaptotagmin I were generated using a glutathione
S-transferase-cytoplasmic domain fusion protein of
synaptotagmin I. Rabbit polyclonal antibodies to synaptobrevin-2 were
generated using an N-terminal peptide (SATAATVPPAAPAGEGGPPC) conjugated
to bovine serum albumin. Western blots were processed with
125I-protein A or 125I-labeled goat anti-mouse
antibody (DuPont NEN) and subjected to autoradiography.
Exocytosis Assays
Secretion assays to monitor the release of [3H]NE were conducted in 0.2-ml reactions that contained Kglu buffer adjusted with CaCl2 to the indicated [Ca2+]f, 0.002 M MgCl2, 0.002 M ATP, and the resuspended 5000 × g pellet. Incubations at 30 °C were terminated by chilling on ice, followed by centrifugation at 100,000 × g for 90 min. [3H]NE in supernatant and in 1% Triton X-100-solubilized pellet fractions were determined by liquid scintillation counting and used to calculate the percentage of NE release as [dpm supernatant]/[dpm total] × 100%. Supernatant values were corrected by subtraction of zero time incubation values determined for each experiment. Antibody to rat CAPS was generated by immunizing rabbits with a glutathione S-transferase fusion protein representing amino acids 264-364 (accession no. U16802[GenBank]) and an IgG fraction was purified by chromatography on protein A-Sepharose (Pharmacia Biotech). The recombinant botulinum neurotoxin E light chain was kindly provided by A. Zdanovsky (Promega Corp., Madison, WI).
The
Ca2+-regulated secretion of NE is faithfully reconstituted
in PC12 cells permeabilized by single passage through a narrow clearance ball homogenizer (6, 7). The
Ca2+-dependent release of NE and other
vesicle constituents by exocytotic fusion of LDCVs with the
plasma membrane is ATP- and cytosol protein-dependent and
inhibited by clostridial neurotoxins that specifically proteolyze synaptobrevin, SNAP-25, and syntaxin (6-10). Multiple passes of PC12
cells through a wider clearance ball homogenizer generated a cell
homogenate consisting of small membrane fragments and intact LDCVs.
Surprisingly, ATP- and cytosol-dependent,
Ca2+-activated release of NE was preserved in membranes
recovered from the homogenate (Fig. 1A). The
Ca2+-dependent release of NE was not affected
by dilution of the membranes, suggesting that the responsive
subcellular fraction contained all of the membranes that participate in
exocytotic fusion. The responsive fraction was recovered by
centrifugation at 5000 × g and was found to contain
~28% (12-44%, n = 8) of the cellular LDCVs present
in post-nuclear supernatants based on the recovery of
[3H]NE or vesicle-specific antigens (see below). Free
LDCVs in the homogenate fail to sediment at 5000 × g
but are fully recovered by centrifugation at 40,000 × g consistent with the size (~75 nm) of LDCVs in PC12 cells
(11). Comparison of the membranes sedimented at 5000 × g and 40,000 × g showed that only the
former exhibited Ca2+-activated NE release (Fig.
1B), indicating that free LDCVs do not release their
contents under standard incubation conditions. Because the LDCVs in the
5000 × g pellet fraction exhibited
Ca2+-dependent NE release and sedimented more
rapidly than free LDCVs, it was likely that these LDCVs were associated
with plasma membranes.
Regulated Exocytosis in a LDCV-Plasma Membrane Complex
To
characterize the membranes present in the 5000 × g
pellet fraction, analysis by velocity gradient centrifugation was
conducted (Fig. 2). LDCVs in the 5000 × g pellet fraction sedimented rapidly into the gradient (Fig.
2A, arrow) as indicated by the co-sedimentation of LDCV contents markers [3H]NE, chromogranin B (Fig.
2A), and dopamine -hydroxylase (data not shown). The
vesicle membrane proteins synaptotagmin I, synaptophysin (Fig. 2,
A and B), SV2, and synaptobrevin (data not shown)
co-sedimented with the contents markers. Rapidly sedimenting LDCVs also
exhibited reserpine-sensitive amine transporter activity characteristic of NE-containing LDCVs (Fig. 2C). Variable amounts of LDCVs
were dislodged during preparation and sedimented nearer the top of the
gradients at a position characteristic of free LDCVs (e.g. fraction 12 in Fig. 2A). Incubation of the 5000 × g pellet fraction under exocytosis conditions prior to
gradient analysis revealed that the release of NE occurred from the
rapidly sedimenting but not free LDCVs (Fig. 2D).
Plasma membranes were monitored by surface labeling of cells and by immunoblotting with SNAP-25 and syntaxin antibodies. The majority of plasma membrane in the PC12 cell homogenate sedimented at 800 × g. About 23% of the plasma membrane was recovered in the 5000 × g pellet fraction and the majority of this co-sedimented with LDCVs in sucrose gradients (Fig. 2, B and E, arrow). Co-sedimentation of LDCVs and plasma membranes was also observed in isopycnic gradient centrifugation in sucrose, Percoll and Ficoll (data not shown). Incubation of 5000 × g pellet fractions under exocytosis conditions prior to gradient analysis revealed a redistribution of plasma membranes following membrane fusion (Fig. 2E). Collectively, the results indicated that the rapidly sedimenting, exocytosis-competent subcellular fraction consisted of LDCVs associated with plasma membrane fragments.
Electron microscopy of the membrane complex purified on sucrose
gradients was consistent with this conclusion revealing the presence of
LDCVs on membrane fragments (Fig. 3, left
panel). LDCVs were not evident in membrane fractions that were
incubated with Ca2+ in the presence of ATP and cytosol
(Fig. 3, right panel) as anticipated if exocytotic fusion
and the release of LDCV constituents had occurred.
The majority of LDCVs in PC12 cells are morphologically docked at the plasma membrane (12), but only about one-third of the LDCVs in a post-nuclear supernatant fraction were isolated in a homogenization-resistant complex with plasma membranes. To evaluate the possibility that LDCV-plasma membrane complexes were disrupted during homogenization, the conditions of homogenization were systematically varied. The results (data not shown) indicated that an optimal yield of rapidly sedimenting LDCVs required 5-10 passes through the ball homogenizer, whereas 15-20 passes did not further increase or decrease the number of LDCVs isolated in rapidly sedimenting complexes. These results suggest that a fraction of the LDCVs in PC12 cells exist in an unusually stable docked state, which could represent a very late stage intermediate in the LDCV exocytotic pathway.
Characterization of Ca2+-activated Exocytosis from LDCV-Plasma Membrane ComplexThe release of NE from LDCV-plasma
membrane preparations exhibited characteristics similar to those
described for permeable PC12 cells prepared by a single pass through
the ball homogenizer (6-8). NE release was entirely dependent on
Ca2+, largely dependent on cytosol, and partially dependent
on MgATP (Fig. 4A). Ca2+
activation was half-maximal at 2-5 µM and optimal at 10 µM (Fig. 4B). Ca2+ stimulated the
concomitant release of several LDCV constituents in addition to NE
including preloaded quinacrine (Fig. 4C) and chromogranin B
(Fig. 4D). The Ca2+-activated release of LDCV
constituents was entirely abolished by treatment with type B, type C
(data not shown), and type E (Fig. 4E) botulinum
neurotoxins, indicating a requirement for functional synaptobrevin,
syntaxin, and SNAP-25 as anticipated for
Ca2+-dependent exocytotic fusion (13).
Botulinum neurotoxin-sensitive release of NE from LDCV-plasma membrane
preparations was alternatively activated by nonhydrolyzable GTP analogs
such as GMPPNP in the absence of Ca2+ (Fig. 1B),
although the rate of stimulated secretion was lower than with
Ca2+ even at maximally effective concentrations (3 µM) of MgGMPPNP.
Previous studies with permeable PC12 cells had shown that the
requirement for MgATP in regulated exocytosis was for priming steps
that precede the final step of Ca2+-triggered fusion (7).
The MgATP requirement for NE release from LDCV-plasma membrane
preparations was similarly found to be largely met in preincubations
with MgATP lacking Ca2+ (Fig. 5), indicating
that a priming process was required for optimal exocytosis of docked
LDCVs. In permeable cells, the Ca2+-triggered fusion step
that follows priming was shown to require a 145-kDa cytosolic protein
CAPS (6, 7).2 CAPS was also found to be
necessary for Ca2+-triggered fusion in the LDCV-plasma
membrane preparation, as indicated by the specific inhibition observed
with IgGs from CAPS immune serum but not control IgGs (Fig.
4A). A nearly complete (90%) inhibition of
cytosol-dependent NE release was observed in incubations
with a different CAPS antibody (data not shown), indicating that the
incomplete inhibition shown in Fig. 4A was attributable to
properties of the antibody.
Docked but Not Undocked LDCVs Undergo Regulated Exocytosis
The results of Figs. 1B and 2D
indicated that only LDCVs associated with plasma membranes were
triggered to exocytose by Ca2+. It was of interest to
determine whether free LDCVs could undergo exocytosis in the presence
of plasma membranes, and whether LDCV-plasma membrane complexes were
competent to dock and fuse additional LDCVs. Mixing experiments were
undertaken in which free LDCVs were incubated with LDCV-plasma membrane
preparations under exocytosis conditions. Incubating unlabeled
LDCV-plasma membrane preparations with [3H]NE-loaded free
LDCVs in various proportions under exocytosis conditions failed to
elicit release of [3H]NE (data not shown). Similarly,
co-incubating [3H]NE-loaded LDCV-plasma membrane
preparations with [3H]NE-loaded free LDCVs in the
presence of Ca2+, MgATP and cytosol failed to elicit more
[3H]NE release than that observed for the LDCV-plasma
membrane complexes alone (Fig. 6). Similar studies with
GMPPNP to trigger fusion at low Ca2+ concentrations also
failed to provide evidence for fusion of free LDCVs with the plasma
membrane (data not shown). The results indicated that exocytosis was
restricted to LDCVs already docked at the plasma membrane and that LDCV
recruitment to and functional docking at the plasma membrane was not
reconstituted in the standard exocytosis assay.
Elucidation of the process of Ca2+-regulated membrane fusion will require identification of proteins that mediate secretory vesicle transport to, docking at, and fusion with the plasma membrane as well as those that mediate the Ca2+ regulation of steps in this sequence. Some of the proteins required for regulated exocytosis have recently been identified, but their precise roles are poorly understood. The SNARE proteins (for soluble NSF attachment protein receptors) synaptobrevin, syntaxin, and SNAP-25, which are the targets of clostridial neurotoxin proteolytic activity (14) and the receptors for the NSF/SNAP proteins required for constitutive fusion (2, 15), form oligomeric complexes that are proposed to function in vesicle docking and/or fusion (2, 16). Studies on the Ca2+-binding vesicle protein synaptotagmin suggest its possible role as a Ca2+-dependent regulator of exocytosis (13). Many additional proteins potentially involved in regulated secretion have been identified genetically or biochemically but remain poorly understood (13, 17). Physiological studies in neuroendocrine cells indicate that secretory granules are staged at numerous points in the exocytotic pathway in transit to the plasma membrane for fusion (18) and that Ca2+ accelerates granule transit through several stages that include recruitment to the plasma membrane for docking, post-docking priming steps, fusion pore formation, and fusion pore expansion (17). It will be essential to reconstitute these stages along the exocytotic route in vitro as distinct biochemical steps in which the role of specific proteins can be assessed and their mechanisms determined.
Some progress toward elucidating partial reactions in the exocytotic pathway has been achieved with permeable neuroendocrine cells. In permeable PC12 cells where the majority of LDCVs are morphologically docked, ATP-dependent priming steps precede and are required for Ca2+-activated fusion steps (7, 12). Ca2+-activated secretion in LDCV-plasma membrane preparations similarly exhibited a requirement for ATP. Because the LDCVs are stably docked at the plasma membrane, the results indicate that priming is distinct from vesicle recruitment or docking and that it occurs in vitro following docking and prior to fusion.
Two distinct ATP-dependent priming reactions have been characterized in permeable PC12 cells. In the first of these, PI transfer protein and phosphoinositide kinases catalyze synthesis of PI(4,5)P2, which is required for Ca2+-triggered fusion (19, 20). In a second priming process, NSF and SNAP proteins catalyze a rearrangement of SNARE protein complexes (12). It remains to be determined whether both of these priming reactions are required for exocytosis in the LDCV-plasma membrane preparation or whether the stably docked LDCVs are at a stage beyond one of these priming steps. The one third of the LDCVs that are consistently recovered in rapidly sedimenting complexes represent a subset of morphologically docked LDCVs that are in a stable, homogenization-resistant state of association with the plasma membrane. It is possible that this unusually stable docked state represents a late prefusion intermediate in which LDCVs intimately associate with the plasma membrane prior to fusion. This intermediate might result from a specific late post-docking reaction, possibly one of the ATP-dependent priming steps, having been completed in situ prior to homogenization. It will be important to determine the molecular basis for the stable association of LDCVs with the plasma membrane. This stable association might be maintained by protein-protein interactions such as those among the SNARE proteins (16, 21), by protein-membrane interactions as proposed for synaptotagmin and phospholipids (22), or by partial membrane fusion as proposed for a stalk intermediate in a hemi-fusion mechanism (23).
A related issue concerns the mechanisms responsible for LDCV docking at the plasma membrane. Under standard incubation conditions, free LDCVs were unable to engage in fusion with the plasma membrane. This may be due to saturation of plasma membrane docking sites by the stably bound LDCVs, to the absence of critical factors required for docking, or to the presence of barriers to docking such as cytoskeletal elements. MacLean and Edwards (5) reported that zymogen granules and pancreatic plasma membranes apparently engage in docking that is followed by fusion. However, fusion in that preparation was not activated by Ca2+ but by nonhydrolyzable guanine nucleotides. Because GTP analogs elicit NE release by a Ca2+-independent, botulinum neurotoxin-sensitive mechanism in the LDCV-plasma membrane preparation, GMPPNP was also tested in mixed incubations of LDCVs with LDCV-plasma membrane complexes, but fusion was not detected. Hence, the process of functional LDCV docking was not reconstituted under the present incubation conditions and will need further investigation.
Studies with permeable PC12 cells indicated that the late Ca2+-dependent step beyond docking and ATP-dependent priming is sensitive to inhibition by clostridial neurotoxins, implying roles for SNARE proteins proximal to Ca2+-dependent membrane fusion steps (11). Studies in other systems have suggested a post-docking role for SNARE proteins (14, 18, 25, 26). The present results further confirm that SNARE proteins have a function beyond docking because Ca2+-activated fusion in the LDCV-plasma membrane preparation was completely inhibited by neurotoxins that proteolytically cleave synaptobrevin, syntaxin, and SNAP-25. Because SNARE proteins in hetero-oligomeric complexes are insensitive to neurotoxin proteolysis (27), the inhibition of regulated secretion in LDCV-plasma membrane preparations by neurotoxins implies that the SNARE proteins are beyond the point of oligomerization that confers toxin-insensitivity. Studies in permeable PC12 cells showed that SNARE proteins become accessible to neurotoxin proteolysis following the NSF-catalyzed rearrangement that occurs during priming (13). It will be of interest to determine whether NSF-dependent priming events occur in the LDCV-plasma membrane preparation in vitro or whether they were completed in situ prior to the isolation of the stably docked LDCVs.
Previous studies in permeable PC12 cells revealed a requirement for the 145-kDa CAPS protein in Ca2+-activated LDCV fusion at a step beyond docking and priming (7).2 Similarly, a requirement for CAPS for regulated fusion in the LDCV-plasma membrane preparation was shown by specific antibody inhibition confirming a role for CAPS beyond docking. Recent studies have shown that CAPS is a PI(4,5)P2-binding protein consistent with a role beyond priming during which PI(4,5)P2 synthesis occurs.3 While regulated exocytosis in the LDCV-plasma membrane preparation exhibits some of the same properties as that in permeable cells, this preparation provides simplifications over permeable cells in consisting of a structurally less complex, exocytosis-competent fraction that will facilitate mechanistic studies on the roles of specific proteins at a post-docking step in the exocytotic pathway.