Molecular Determinants of Regulated Exocytosis

The constitutive and regulated secretory pathways were described in a classic work by Palade (1) as a series of membrane-trafficking steps. Secretory material, primarily proteins, is synthesized in the endoplasmic reticulum and transported in membrane vesicles via the Golgi apparatus to the plasma membrane. However, in addition to the classic pathway, some secreted small molecules that are released by regulated exocytosis are not synthesized in the endoplasmic reticulum, but are taken up directly into secretory vesicles after these have been generated from the Golgi apparatus. This pathway of regulated exocytosis is exemplified most prominently by neurotransmitters that are transported into synaptic vesicles just before exocytosis. Thus, two pathways of regulated exocytosis can be distinguished (Table 1):

• The first pathway secretes primarily polypeptides (e.g., insulin, glucagon, etc.), which are initially synthesized in the endoplasmic reticulum, modified in the Golgi apparatus, and processed proteolytically in precursor organelles. These organelles mature to secretory granules; stimulation then induces exocytosis of the granules at the plasma membrane. After exocytosis, secretory vesicles in this pathway have to recycle via the Golgi complex in order to be refilled with secretory material.

• The second pathway secretes primarily low-molecular-weight substances (e.g., catecholamines, glutamate, etc.), which are synthesized in the cytosol (and partly in secretory vesicles), taken up into and stored by secretory vesicles, and secreted by exocytosis (3). In contrast to the peptide pathway, secretory vesicles recycle locally and can be reused independently of the Golgi complex (4).

Virtually all regulated endocrine secretion uses the first pathway since most hormones are peptidergic (except for steroid and thyroid hormones), whereas all neurotransmitter release is mediated by the second pathway (Table 1). The two principal pathways of regulated exocytosis differ in the mechanisms by which secretory vesicles are filled with secretory material and by which the vesicles recycle after exocytosis for a new round of secretion. In addition, they exhibit distinct physiological properties (Table 1). Most commonly, regulated exocytosis is stimulated in endocrine glands and presynaptic nerve terminals by Ca²⁺ influx, but the dynamics of secretion are very different. In endocrine glands, Ca²⁺ not only triggers fusion, but also serves to mobilize the vesicles and accelerate their priming for fusion; thus, secretion is relatively slow and sustained (5). Conversely, in nerve terminals Ca²⁺ primarily stimulates exocytosis of predocked vesicles, resulting in very rapid and short-lasting secretory events (6,7).

Synaptic vesicles, the secretory organelle of the synapse, are abundant and relatively easy to purify. As a result, synaptic vesicles have been a rich source of information about the proteins involved in exocytosis (8,9). Although the majority of proteins that function in regulated secretion were discovered at the synapse, virtually all of these proteins also act in endocrine secretion of peptide hormones (see, for example, reference 10). Table 2 presents an overview of the major protein components of the secretory ogranelles in the two pathways of regulated secretion (reviewed in reference 8). It is obvious that the majority of proteins are shared among the pathways; in fact, it is difficult to discern the basis of pathway specificity for the two classes of secretory vesicles (which are often expressed in the same cell) given the large number of shared components. The molecular similarity between the two pathways extends beyond the vesicle components since most of the cytoplasmic and plasma membrane components involved also appear to be identical (e.g., syntaxins, munc18s, N-ethylmaleimide-sensitive factor [NSF], etc; see below). As a consequence, most conclusions derived from studies of synaptic vesicle exocytosis can likely be applied to other regulated (and also constitutive) pathways of exocytosis, and the molecular determinants of the specificities of these pathways remain unknown.

Based on biochemical and electrophysiological studies, regulated exocytosis in nerve terminals and endocrine cells has been divided into three steps (8,11,12): (i) secretory vesicle attachment, which docks or tethers the vesicles at the plasma membrane; (ii) prefusion, which primes the vesicles for exocytosis and may consist of a partial fusion reaction; and (iii) release of the vesicle contents, which requires opening of the fusion pore. Similar steps are likely to occur in all cellular membrane fusion (see, for example, references 13,14). However, some confusion exists about the different stages in various fusion reactions. One problem is that “docking” or membrane attachment is usually defined morphologically. Thus, dependent on how tightly vesicles need to be bound to each other in order to be observed as “docked,” this will require different sets of proteins. Another problem is that the same words are used differently in various fusion reactions. For example, in synaptic and endocrine fusion, “priming” refers to prefusion that is required for docked/tethered vesicles to be stimulated for exocytosis (11,12), whereas, in vacuolar fusion, “priming” refers to preparation of the fusing membranes for subsequent docking/tethering (14).

As implied by its very name, regulated exocytosis is subject to extensive control. The first step in exocytosis, attachment/docking, requires that the secretory vesicles and the plasma membrane “recognize” each other, but the nature of this recognition is unclear. The prefusion/priming step that follows docking probably consists of multiple separate reactions, which, in endocrine exocytosis but probably not in synaptic exocytosis, requires polyphosphatidylinositol phosphates (12,15). Finally, fusion pore opening is probably the most tightly regulated step since it mediates the release of the secretory materials. In the case of neurotransmitter release, acute triggering of release has to act on the very last step of exocytosis simply because Ca²⁺-triggered release is so fast (<100 microseconds in some synapses) (7) that the time is too short to act on earlier steps. Based on this speed, it has been argued that synaptic vesicle exocytosis is arrested at a late step in fusion, i.e., largely completed during the prefusion reaction, and that Ca²⁺ acts at a point in exocytosis at which only fusion pore opening is required for release (8). By analogy to synaptic vesicle exocytosis, it seems likely that the very fast component of Ca²⁺-triggered exocytosis in endocrine cells also acts on a largely prefused vesicle population. However, in endocrine cells this component accounts for only a minor percentage of secretion; most release is delayed, whereas at the synapse delayed release is negligible. Thus, in endocrine exocytosis, acute regulation of exocytosis probably addresses multiple steps. Extensive evidence suggests that other steps in the synaptic vesicle pathway, in addition to the final Ca²⁺-triggered release reaction, are also tightly regulated.

Based on studies of synaptic vesicle exocytosis, membrane traffic in yeast, and in vitro fusion reactions of intracellular organelles, four protein families are probably involved in all eukaryotic fusion reactions: soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs), NSF with adaptor proteins, rabs, and SM proteins (sec1/munc18-like proteins) (reviewed in references 11,16,17).

• SNARE proteins are defined by the presence of a signature sequence of ∼60 residues, the so-called SNARE motif, which is characterized by a high potential for coiled-coil formation. SNARE proteins usually, but not always, contain a transmembrane region, and are present on opposing membranes before fusion. The SNARE motifs in these proteins bind to each other to form an extremely stable complex composed of coiled coils, the so-called core complex (18). A large number of SNARE proteins have been identified that act in different membrane-trafficking reactions all over the cell; some SNAREs are specific for only one fusion reaction (e.g., syntaxin 1 for exocytosis), whereas others are shared among many fusion reactions (e.g., Vti1) (reviewed in references 19–21). Two classes of plasma membrane SNAREs (synaptosomal-associated protein-25 [SNAP-25] and syntaxins) interact with one class of vesicular SNARE (synaptobrevins, also called vesicle-associated membrane proteins [VAMP]) in exocytosis (22). Each exocytotic SNARE is represented by multiple closely related isoforms: SNAP-25 and -23; syntaxins 1, 2, 3, or 4; and synaptobrevins 1 and 2 and cellubrevin. The isoforms are functionally probably slightly different; in regulated exocytosis at the synapse and in endocrine cells, fusion events utilize primarily synaptobrevins 1 or 2, syntaxins 1 or 3, and SNAP-25. In addition to the traditional classification of SNAREs based on their localization on vesicles donor or target acceptor membranes (v- and t-SNAREs), SNAREs can be classified based on the central residue in their SNARE motif that is either Q or R (23). According to this classification, syntaxins and SNAP-25s are Q-SNAREs and synaptobrevins are R-SNAREs. Synaptobrevin/VAMP is a short, very abundant synaptic vesicle protein (∼120 residues) composed of an N-terminal 30-residue proline-rich sequence that is not well conserved between species, a central SNARE motif, and a COOH-terminal transmembrane region (24,25). SNAP-25, or synaptosomal protein of 25 kDa, is composed of two SNARE motifs that are connected by a long linker sequence containing multiple cysteine residues. SNAP-25 lacks a transmembrane region, and is attached to the membrane via multiple palmityl residues that are bound to the cysteine residues in the central region (26). Syntaxin was initially identified and cloned based on a monoclonal antibody named HPC-1 (27,28), and later described as syntaxin in a complex with the synaptic vesicle protein synaptotagmin and Ca²⁺ channels (see below for a discussion of synaptotagmin) (29,30). Syntaxin is similar to synaptobrevin in that it contains a COOH-terminal SNARE motif followed by a single transmembrane region that anchors it in the membrane. Different from synaptobrevin, however, the N-terminal sequence of syntaxin is relatively long (∼180 residues) and forms an independently folded three-helical domain called the Habc domain (31). Interestingly, the Habc domain of syntaxin 1 folds back onto its SNARE motif when syntaxin 1 is not in a complex with SNAP-25 (see Fig. 1) (32). In the folded back stage, syntaxin 1 assumes a “closed” conformation that has to open in order to expose the SNARE motif for SNAP-25 and synaptobrevin binding during core complex formation. The first evidence that SNAREs in general, and the synaptic SNAREs syntaxin, SNAP-25, and synaptobrevin in particular, are involved in membrane fusion was derived from the observation that botulinum and tetanus toxins specifically attack the synaptic SNARE proteins (33–39). Botulinum and tetanus toxins are site-specific proteases that are taken up into nerve terminals where they block synaptic vesicle exocytosis. Because the toxins are catalytic proteases, they are among the most toxic substances known. A single molecule of a toxin is sufficient to poison a whole nerve terminal. Botulinum and tetanus toxins severely inhibit exocytosis in endocrine cells and neurons without causing significant structural changes, suggesting a selective effect on fusion. The fact that different SNARE proteins are targets of different types of botulinum and tetanus toxins suggests that the core complex formed by the SNARE proteins, and not only the individual SNARE proteins, is an essential component of the fusion reaction.

• NSF is an ATPase that binds to SNARE complexes via adaptor proteins called “SNAPs” (soluble N-ethylmaleimide-sensitive adapter protein) (no relation to the synaptic SNARE protein SNAP-25 that was named before the adaptor proteins) (40). NSF functions similar to a chaperone that mediates an ATP-dependent conformational change in its substrate; although specific for core complexes, it exhibits no specificity for SNAREs, but indiscriminately dissociates all core complexes throughout the cell. It seems probable that NSF is necessary as a specific ATPase that recycles core complexes because of the high stability of these complexes (11).

• SM proteins contain ∼600 residues with sequence homology over their entire length, but no special sequence motifs. SM proteins were first discovered in Caenorhabditis elegans in the unc18 mutant (hence the “munc18” label) (41), and later in yeast as the first secrectory mutant (hence sec1) (42). SM proteins have been identified in every intracellular membrane fusion process studied, similar to SNAREs and rab proteins (reviewed in reference 11). Three different SM proteins, munc18-1 (also called munc18a, nsec1, and rbsec1), munc18-2 (also called munc18b), and munc18c, are involved in exocytosis, of which regulated exocytosis primarily utilize munc18-1 (43–47). Munc18-1, the most abundant brain SM protein that functions in synaptic vesicle exocytosis (43), binds directly to syntaxin 1 in a dimeric high-affinity complex and is essential for neurotransmitter release (48). Munc18 binds to the closed conformation of syntaxin 1 in which its N-terminal domain is folded back onto its COOH-terminal SNARE motif (32). The closed conformation may prevent formation of core complexes by syntaxin, which would explain why syntaxin apparently cannot bind munc18 and SNAP-25 simultaneously (45). These biochemical findings are beautifully illustrated in the crystal structure of the munc18-1/syntaxin 1 complex, which confirms the biophysical studies of the closed conformation (32,49). It is unclear, however, if other SM proteins also directly bind to syntaxins, if the complex also involves a closed conformation, and how these SM protein/syntaxin complexes relate to the core complexes formed by the same syntaxin. At least in the vacuolar fusion system, the syntaxin isoform involved (Vam3p) does not fold into a closed conformation, and its N-terminal domain is not required for either SM protein binding or fusion (50,51). One current hypothesis about SM protein function, based on yeast genetics, is that an SM protein bound to a syntaxin SNARE interacts with Rab GTPases and/or their effectors during the initial stages of vesicle target recognition (52). Then the regulated dissociation of the SM protein from the syntaxin SNARE would allow, or perhaps facilitate, an interaction of the SNARE motifs of syntaxin with those of SNAP-25 and synaptobrevin, resulting in formation of the core complex. However, this hypothesis remains to be tested for exocytosis.

• Rab proteins are guanosine triphosphate (GTP)-binding proteins that appear to be associated with specific fusion events (reviewed in reference 53). More than 60 rab proteins are expressed in mammalian cells (54). Rab protein function is probably very diverse. Only one class of rab proteins (rab3A, 3B, 3C, and 3D) with a function in endocrine and synaptic exocytosis has been identified (55). Rab3s are primarily regulatory, and only appear to have a role in regulated exocytosis. Additional rab proteins probably remain to be identified; in view of the incomplete understanding of rab proteins in exocytosis, they will not be discussed further here.

The biochemical properties of SNARE and SM proteins and their tight interactions suggest that these proteins undergo a cycle of association and dissociation during fusion (reviewed in reference 56). A tentative working model for this cycle is shown in Fig. 1. The cycle is based on the fundamental observation that syntaxin engages in sequential interactions during exocytosis, and that these interactions require major conformational changes in syntaxin. It is proposed that syntaxin 1 first binds to munc18-1 and then forms the core complex with the SNAP-25 and synaptobrevin SNAREs. The core complex in turn associates with SNAPs and NSF, which dissociates the core complex under ATP hydrolysis, thereby converting syntaxin 1, SNAP-25, and synaptobrevin into a free form. The free syntaxin 1 binds to munc18-1 again, thereby reinitiating the cycle. In the model, the sequential interactions of syntaxin 1 are hypothesized to be part of the priming/prefusion reaction of exocytosis. Although the model is largely based on the work on the synapse, it probably also applies to endocrine exocytosis. Like all models, it is probably not completely correct, but may provide a useful conceptual framework for future experiments.

The model raises two related questions. First, what are the specific functions of the munc18/syntaxin and the core complex? Second, is another sequence of interactions possible, for example, a role of NSF early in fusion, or of munc18-1 late in fusion? At present, these questions can only be partly addressed. Little is known about the function of the munc18/syntaxin complex, apart from the fact that munc18 is absolutely essential for synaptic exocytosis (48). Initial ideas that the munc18/syntaxin complex is exclusively regulatory are thus unlikely, but no clear alternative model has been proposed. In yeast, SM proteins appear to be essential for a relatively early tethering reaction in membrane fusion before core complexes form, as shown for multiple reactions (13,17,52), and it is possible that munc18 has an analogous role. More precise ideas exist about the function of the core complex. Here an important clue comes from the three-dimensional structure of the complex, which shows that the four SNARE motifs, one each from syntaxin 1 and synaptobrevin and two from SNAP-25, form a four-helical bundle in which the helices are parallel (18). This arrangement implies that formation of the complex will bring the membranes containing the SNAREs close together, perhaps even forcing them to merge, which has been referred as the zipper-model of membrane fusion (57–60). In the four-helical bundle of the core complex, the hydrophobic residues are buried, but in the middle glutamine residues from syntaxin and from SNAP-25 form salt bridges with an arginine residue from synaptobrevin. The resulting structure is very stable, with an unfolding transition over 95°C (61), suggesting that a large free-energy change upon core complex formation could be used to drive membrane fusion. A role for the core complex in driving fusion is a plausible hypothesis that is supported by in vitro fusion reactions (62), which, however, do not require munc18 and thus may not reflect the physiological functions of the proteins involved.

Most but not all exocytotic fusion is Ca²⁺ dependent. This has been best investigated for neurotransmitter release which is triggered by Ca²⁺ extremely rapidly (in as little as 60 microseconds) (7) with a steep Ca²⁺ concentration dependence (63). The Ca²⁺ concentration dependence suggests that at least three or four Ca²⁺ ions must act simultaneously to induce synaptic vesicle exocytosis, and the speed of Ca²⁺-triggering indicates that the Ca²⁺ ions must act locally at the site of exocytosis by effecting a small conformational change or an electrostatic switch. In contrast, Ca²⁺-triggered fusion of peptidergic vesicles in endocrine cells is multiphasic and sustained. Although endocrine cells also exhibit fast exocytosis, this is usually a very minor component of regulated secretion (64). Most Ca²⁺-induced exocytosis in endocrine cells is affected in more sustained phases that correspond to time points at which secretion at the synapse would long have terminated (5). Thus, there are major differences in the time course of Ca²⁺-triggered exocytosis between endocrine and synaptic systems, although the Ca²⁺ dependence and cooperativity appear to be similar.

The simplest way to think about regulated fusion is that it is a variant of constitutive fusion in which a “red light” has been inserted to stop the reaction from completion. This “red light” hypothesis would postulate that an inhibitor freezes the reaction, and that Ca²⁺ binding to a Ca²⁺ sensor abolishes inhibition, allowing the fusion reaction to proceed and the fusion pore to open. This prompts the question, what are the Ca²⁺ sensors involved, and how do they work? Multiple Ca²⁺-binding proteins are known that might be candidates for a role as Ca²⁺ sensors (e.g., calmodulin, rabphilin, and synapsin I), but the best candidates are synaptotagmins. Synaptotagmins constitute a family of at least 13 proteins that are characterized by the same domain organization (reviewed in reference 65), are primarily expressed in neurons, but have distinct subcellular localizations. All synaptotagmins contain an N-terminal transmembrane region, a central linker sequence, and two COOH-terminal C₂-domains referred to as C₂A- and C₂B-domains (Fig. 2). The C₂-domains are the most highly conserved parts of synaptotagmins. In most synaptotagmins, the C₂-domains account for more than half of the protein, suggesting that they constitute the “business ends” of the molecules. C₂-domains represent independently folding domains that bind multiple Ca²⁺ ions and interact with a number of potential targets as a function of Ca²⁺, primarily phospholipids (71,72). Thus, by virtue of their C₂-domains, synaptotagmins are membrane bound Ca²⁺ regulatory proteins.

All synaptotagmins are primarily expressed in brain, although the mRNAs for many synaptotagmins are also synthesized in nonneuronal tissues at low levels (73,74). Synaptotagmins 1 and 2, the first synaptotagmins purified and cloned, constitute abundant synaptic vesicle proteins that are highly homologous. These two synaptotagmins are very similar and are detectable only in brain and endocrine tissues where they are expressed in overlapping but distinct sets of cells (66–68). All synaptic vesicles and secretory granules exocytosis appear to contain either synaptotagmin 1 or 2 or both, suggesting a general function. When the additional synaptotagmins besides synaptotagmins 1 and 2 were identified (74,76), the expectation was that these “other” synaptotagmins would also be synaptic vesicle proteins, or at least vesicular-trafficking proteins. It was thus a great surprise when localizations of three additional synaptotagmins, synaptotagmins 3, 6, and 7, which are most abundant after synaptotagmins 1 and 2, revealed a localization on the plasma membrane (69,70). The precise localizations of other synaptotagmins are currently unknown, although synaptotagmin 4 has also been detected in unidentified intracellular vesicles (77,78).

The best evidence for a function of synaptotagmins in exocytosis was derived in studies of mutants in synaptotagmin 1 in mice, flies, and worms (79–82). The highest resolution in the analysis of these mutants was possible in mutant mice, where sophisticated electrophysiological tools made it possible to define the point of action of synaptotagmin in exocytosis (82). These studies revealed that synaptotagmin is essential only for fast, Ca²⁺-triggered release, but not for other steps in exocytosis, demonstrating a role for synaptotagmin 1 in the Ca²⁺-sensing step of exocytosis. The precise nature of this role was recently examined in knockin mice in which a point mutation that decreases the overall affinity of synaptotagmin 1 for Ca²⁺ approximately twofold was introduced into the endogenous murine synaptotagmin 1 gene (83). This mutation decreased Ca²⁺-dependent neurotransmitter release approximately twofold, and selectively impaired the ability of individual vesicles to be triggered by Ca²⁺ for exocytosis. In agreement with these in vivo studies, the in vitro properties of synaptotagmin 1 provide additional support for a role as exocytotic Ca²⁺sensor. As described above, synaptotagmin 1 binds multiple Ca²⁺ ions via its two C₂-domains with a relatively low affinity and a divalent cation specificity that resembles the properties of the exocytotic Ca²⁺ sensor. As an abundant synaptic vesicle protein, it is located at the right place to function as a Ca²⁺ sensor. Furthermore, synaptotagmin 1 binds to phospholipids, syntaxin, and itself as a function of Ca²⁺ (reviewed in reference 65). This suggests the possibility that Ca²⁺-induced interaction of synaptotagmin with the core complex and lipids in conjunction with its multimerization could trigger fusion-pore opening.

Although these data strongly support a central function for the vesicular synaptotagmins 1 and 2 in triggering exocytosis, the available results also show that these synaptotagmins cannot alone be responsible for Ca²⁺-stimulated exocytosis. Several observations indicate that additional Ca²⁺ sensors must be present in regulated exocytosis. For example, some neurotransmitter release was still present in synaptotagmin 1–deficient flies and worms (79–81). Furthermore, in synaptotagmin 1 knockout mice, residual neurotransmitter release was still Ca²⁺ dependent, although the amount of release was decreased considerably and exhibited a delayed time course (82). Finally, in endocrine PC12 cells, the absence of synaptotagmin 1 appeared to have no deleterious effect on Ca²⁺-triggered release, at least at the slow time points measured in that study (84). The best candidates for other Ca²⁺ sensors in release that could explain these findings are the “other” synaptotagmins, especially the plasma membrane synaptotagmins 3, 6, and 7 (Fig. 2) (74). The following evidence suggests that these synaptotagmins may also play a critical role in Ca²⁺-triggered exocytosis. First, they are strategically located on the plasma membrane, the target for the secretory vesicle during exocytosis. This results in a symmetrical arrangement of synaptotagmins in which distinct isoforms are facing each other when vesicles become docked for exocytosis (Fig. 2) (69,70). However, it should be noted that a possible localization of synaptotagmin 3 on insulin granules has also been reported based on subcellular fractionations (85). Second, synaptotagmins 3, 6, and 7 are also Ca²⁺-binding proteins, and thus could mediate a Ca²⁺-regulatory action (68,74). Third, in permeabilized PC12 cells, the C₂A- and C₂B-domains of synaptotagmin 7 selectively inhibit Ca²⁺-triggered exocytosis, whereas the C₂A- and C₂B-domains of synaptotagmins 1 and 2 had no effect (70). Synaptotagmin 7 C₂-domains were effective at low nanomolar concentrations, and did not impair the secretory apparatus as such, but only Ca²⁺ triggering of exocytosis at Ca²⁺ concentrations above 1 μmol/l. Together these properties indicate that the plasma membrane synaptotagmins could cooperate with the vesicular synaptotagmins in triggering exocytosis. The best current model is that the different synaptotagmins have distinct, but complementary roles in exocytosis, with synaptotagmins 1 and 2 specialized for fast synaptic vesicle exocytosis, whereas synaptotagmins 3, 6, and 7 (and possibly others) may be responsible for slower exocytosis and be more important in endocrine cells. However, this is at present only a model that requires direct testing.

Led by biochemical and genetic studies on synaptic vesicle exocytosis, considerable progress has been made in recent years in our understanding of regulated exocytosis (8). We now have a working model for the molecular machinery that mediates membrane fusion during exocytosis (Fig. 1). The basis of this machinery is a relatively simple cycle of protein associations and dissociations in which syntaxin undergoes sequential interactions with multiple other proteins that are essential for fusion, and experiences major conformational changes during these interactions. Much remains to be clarified in this cycle. For example, the precise function of the SM protein munc18-1 in fusion is still unknown, and the hypothesis that core complex formation from SNAREs drives fusion still needs to be validated. Nevertheless, this cycle provides the basis for additional studies, and its relative simplicity and the conservation of its components in other fusion reactions suggests that it will be paradigmatic. Similarly, elucidation of how Ca²⁺ triggers exocytosis appears to be within reach (Fig. 2). The unexpected complexity of multiple classes of related Ca²⁺ sensors opens up new avenues to explain the characteristics of fusion that were previously unclear. These results point to directions that will be pursued in future studies to test the relative roles of different synaptotagmins in fusion.

In spite of this progress, however, many questions remain unanswered. For example, in view of the widely overlapping protein composition of small synaptic vesicle and large dense core secretory granules (Table 2), how is the identity and nature of these organelles established, and what guides their differential intracellular pathways? Another major question is how secretory vesicles become docked at the plasma membrane for exocytosis. Multiple proteins of unknown function exist in secretory vesicles that could have a role in these processes. Again, future studies, most likely using genetic approaches, will hopefully shed light on these important questions.