|Home | About | Journals | Submit | Contact Us | Français|
Myosin-Va is an actin-based processive motor that conveys intracellular cargoes. Synaptic vesicles are one of the most important cargoes for myosin-Va, but the role of mammalian myosin-Va in secretion is less clear than for its yeast homologue, Myo2p. In the current studies, we show that myosin-Va on synaptic vesicles interacts with syntaxin-1A, a t-SNARE involved in exocytosis, at or above 0.3 μM Ca2+. Interference with formation of the syntaxin-1A–myosin–Va complex reduces the exocytotic frequency in chromaffin cells. Surprisingly, the syntaxin-1A-binding site was not in the tail of myosin-Va but rather in the neck, a region that contains calmodulin-binding IQ-motifs. Furthermore, we found that syntaxin-1A binding by myosin-Va in the presence of Ca2+ depends on the release of calmodulin from the myosin-Va neck, allowing syntaxin-1A to occupy the vacant IQ-motif. Using an anti-myosin-Va neck antibody, which blocks this binding, we demonstrated that the step most important for the antibody's inhibitory activity is the late sustained phase, which is involved in supplying readily releasable vesicles. Our results demonstrate that the interaction between myosin-Va and syntaxin-1A is involved in exocytosis and suggest that the myosin-Va neck contributes not only to the large step size but also to the regulation of exocytosis by Ca2+.
Myosin-V, a processive molecular motor, conveys vesicles and other organelles along F-actin (Mercer et al., 1991 ; Espreafico et al., 1992 ; Cheney et al., 1993 ; Reck-Peterson et al., 2000 ; Vale, 2003 ). This unconventional myosin is a member of the class-V myosins, which are expressed in all eukaryotic species from yeast to mammals (Reck-Peterson et al., 2000 ; Matsui, 2003 ; Vale, 2003 ). Myosin-V is a dimeric protein (Cheney et al., 1993 ). Each monomer is composed of a head region, a long neck domain containing six tandem IQ-motifs, and a tail region (Reck-Peterson et al., 2000 ). The head acts as a plus-end ATPase-dependent molecular motor to move myosin-V along F-actin (Reck-Peterson et al., 2000 ). The long neck region is thought to act as a lever arm to regulate the motor activity of the head and to maintain the large step size of myosin-V via bound light chains. In higher eukaryotes, the light chains consist mainly of calmodulin (CaM) bound to the IQ-motifs in the neck domain (Cheney et al., 1993 ; Vale, 2003 ). In addition, the globular tail of myosin-V interacts with membrane-bound vesicles. In these ways, myosin-V plays a central role in intracellular polarized transport (Reck-Peterson et al., 2000 ; Matsui, 2003 ; Vale, 2003 ).
Among the three isoforms of myosin-V in higher vertebrates, myosin-Va is the most abundant, and it is highly enriched in the brain (Espreafico et al., 1992 ), particularly in the neurons (Tilelli et al., 2003 ). Several lines of evidence indicate that synaptic vesicles, which undergo the Ca2+-regulated exocytosis, are one of the most important cargoes for myosin-Va (Prekeris and Terrian, 1997 ; Bridgman, 1999 ; Tilelli et al., 2003 ). In addition, Myo2p, a yeast homologue of myosin-Va, directs intracellular transport during secretion and budding through interactions with other proteins (Matsui, 2003 ). However, the roles of myosin-Va in secretion and Ca2+-regulated exocytosis are not as clear, probably because the myosin-Va-interacting molecules have not been identified in neurons (Reck-Peterson et al., 2000 ; Matsui, 2003 ).
In the current studies, we found a novel interaction between myosin-Va, which is present on cortical synaptic vesicles (Prekeris and Terrian, 1997 ; Bridgman, 1999 ) and syntaxin-1A, a t-SNARE that participates in exocytosis (Duman and Forte, 2003 ; Li and Chin, 2003 ), in presence of micromolar levels of Ca2+. We also found that this unique interaction, linked to Ca2+-dependent release of CaM from the neck region of myosin-Va, is involved in Ca2+-regulated exocytosis.
Immunoprecipitation using an antibody against the globular tail of myosin-Va (gift of P. C. Bridgman, Washington University School of Medicine, St. Louis, MO; Evans et al., 1998 ) was performed as described previously (Ohyama et al., 2002 ) except that the buffer contained 1 μM CaCl2, 2 mM MgCl2, and 0.5 mM ATP. In some experiments, immunoprecipitation was performed with a pool of myosin-Va and syntaxin-enriched fractions. This material was obtained as fractions 20–26 from a 5–40% sucrose density gradient centrifugation of hypotonically-treated brain P2 fraction, carried out as described by Ohyama et al. (2002 ) for glycerol density gradient centrifugation. The antibody against myosin-Va tail used for immunoblotting was kindly provided by V. I. Gelfand (University of Illinois, Urbana, IL; Karcher et al., 2001 ).
The brain homogenate (S2 fraction) was prepared as described by Fujita et al. (1998 ). Ca2+-dependent syntaxin-1A binding proteins from rat brain were detected using PreScission protease (GE Healthcare, Uppsala, Sweden) as described previously (Ohyama et al., 2002 ). The 190-kDa syntaxin-1A binding protein was digested with trypsin and analyzed by mass spectrometry. This protein contained the sequence YFATVSGSASEANVEEK, which corresponds to amino acids 179–195 of myosin-Va.
Cosedimentation experiments were performed as described by Nascimento et al. (1996 ). Purified brain myosin-Va (50 nM) was mixed with 500 nM of F-actin in the presence of 10–6 M Ca2+. In some experiments, the mixture was added to glutathione S-transferase (GST)-syntaxin-1A (50 nM). Cosedimentation was confirmed by centrifugation of the protein mixture at 100,000 × g for 1 h (Nascimento et al., 1996 ), and the pellet and supernatant were analyzed by SDS-PAGE followed by staining with Coomassie Brilliant Blue.
We also examined whether actin can access the complex between myosin-Va and syntaxin-1A to form a ternary complex. Myosin-Va (5 nM) and either actin (50 nM) or syntaxin-1A (5 nM) were first mixed together and incubated for 1 h at 4°C in the presence of 10–6 M Ca2+. The missing third component (syntaxin-1A or actin, respectively) was then added, and the mixture was incubated for another 1 h. Next, immunoprecipitation was carried out as described above using an anti-myosin-Va antibody (1:200) or an anti-syntaxin-1A antibody (1:200). Myosin-Va, syntaxin-1A, and actin were detected by immunoblotting.
Native myosin-Va was purified from chick brain (Cheney, 1998 ). Recombinant myosin-V (DHM5; [1-1193]) was produced in Sf9 cells as described previously (Homma et al., 2000 ) or by in vitro translation (Promega, Madison, WI) using the mouse dilute cDNA (gift of N. A. Jenkins, University of Sao Paolo, Ribeirao Preto, Brazil; Mercer et al., 1991 ). DHM5 was detected with anti-myosin-Va head antibody (gift of R. E. Larson, National Cancer Institute, Frederick, MD; Nascimento et al., 1996 ; Evans et al., 1998 ). The binding experiments were carried out using GST-syntaxin-1A fusion proteins immobilized on glutathione-Sepharose (Ohyama et al., 2002 ). In some experiments, His6-DHM5 fusion protein (Homma et al., 2000 ) was immobilized on a Ni2+-chelating column. Various concentrations of Ca2+ were generated using an EGTA-Ca2+ buffer with the required amounts of CaCl2 and 4 mM EGTA calculated using Max Chelator or WebMaxC (http://www.stanford.edu/~cpatton/maxc.html) software. In the reconstitution study, the purified myosin-Va and GST-syntaxin 1A [1-262] were incubated together for 1 h, followed by an additional 1 h with recombinant SNAP-25, VAMP-2 [1-96], NSF, and α-SNAP (Hohl et al., 1998 ). The concentration ratio of the proteins (except for DHM5) was determined as described by Hohl et al. (1998 ). Synaptic vesicles were purified from adult rat brain as described previously (Huttner et al., 1983 ). Bacterial two-hybrid experiments were carried out using BacterioMatch (Stratagene, La Jolla, CA) according to the manufacturer's instructions. The rabbit anti-myosin-V neck antibody was generated against the neck domain sequence of mouse myosin-Va and was affinity-purified using protein G-Sepharose (Sigma-Aldrich, St. Louis, MO).
The binding stoichiometry between syntaxin-1A and myosin-Va was measured using a BIAcore3000 (BIAcore, Uppsala, Sweden) by immobilizing myosin-Va on CH5 carboxymethyl chips and adjusting the resonance units (RU) to ~10,000 as described in the manufacturer's instructions. Next, syntaxin-1A [1-262] (0.1–25 μM) in HEPES-buffered saline (10 mM HEPES, pH 7.4, 150 mM NaCl) containing 0.005% Tween 20, 0.1 mM dithiothreitol, and pCa = 5.5 was injected into the flow cells of a BIAcore 3000. The sensorgrams were analyzed using BIA evaluation software version 3.1 (BIAcore). The stoichiometry was calculated from changes in RU at the point between association and dissociation on the compensated sensorgram and using 1200 RU as equal to 1.2 ng of mass per flow cell.
Amperometric measurement of exocytotic catecholamine release was performed as described previously (Ohyama et al., 2002 ; Quetglas et al., 2002 ) except that the chromaffin cells were stimulated with 60 mM KCl. Microinjection was performed using a 6-d-old culture of chromaffin cells on collagen-coated coverslips. In each experiment, the cytosol of 50–150 cells was microinjected using an Eppendorf injection system. The cytosolic concentration of the injected fragments was estimated to be 60–120 μg/ml. For cells injected with the syntaxin-1A fragment [191-240] or its L222E mutant, the microinjected cells were stimulated by 60 mM KCl for 4 s. Cells injected with the anti-myosin-Va neck and normal antibodies were stimulated with KCl for 5 min, and the exocytotic frequency during the initial (0- to 1-min) and sustained (1- to 5-min) phases was compared with determine the step of exocytosis regulated by the interaction between myosin-Va and syntaxin-1A.
Myosin-Va ATPase activity was determined in a reaction mixture containing 50 μg/ml myosin-Va, 420 μg/ml F-actin, 20 mM imidazole-HCl, pH 7.2, 75 mM KCl, 2.5 mM MgCl2, 2 mM ATP, 4 mM EGTA, enough CaCl2 to generate the desired pCa (between 8 and 5) at 25°C, and the presence or absence of 30 μg/ml syntaxin-1A [1-262]. The time course was measured by removing an aliquot every 3 min. The ATPase activity was calculated from the concentration of the released Pi (Chifflet et al., 1988 ) per mole of myosin-Va per second. The assay of myosin-Va motility was carried out using rhodamine-phalloidin-labeled F-actin as described previously (Rock et al., 2000 ). After blocking the flow cells with bovine serum albumin, myosin-Va (20–30 μg/ml) was added and adsorbed to the cells for 2 min at room temperature. The flow buffer contained Ca2+ (pCa = 6) in the presence or absence of 1 μM syntaxin-1A. Similar results were obtained with 1 μM and higher concentrations (e.g., 10 μM) of syntaxin-1A (our unpublished data).
The 80-kDa SNARE complex (i.e., the SDS-resistant complex) was isolated as described previously (Igarashi et al., 1997 ). Briefly, the immobilized 1 μM GST-syntaxin-1A was incubated for 1 h with an equal amount of recombinant SNAP-25 and VAMP-2 and eluted by cleavage with PreScission protease. The eluted proteins were then incubated with or without 1 μM syntaxin-1A [191-240] for 0.5 h and then treated with SDS-sample buffer at 60°C for 5 min (which does not break up the SNARE complex). The 80-kDa protein complex was analyzed by immunoblotting with antibodies specific to syntaxin-1A, SNAP-25, and VAMP-2, which are components of the neuronal SNARE complexes.
AFM was carried out as described previously (Mizuta et al., 2003 ). Myosin-Va was diluted to 5–10 μg/ml in 10 mM HEPES, pH 7.4, containing 2 mM MgCl2. Next, 5 μl of the sample was dripped onto freshly cleaved mica and dried with compressed air. Two minutes later, Milli-Q water (10 μl) was dripped onto the mica surface to remove salts, and the surface was immediately air-dried. The cantilevers (SI-DF40-AL; Seiko Instruments, Neu Isenburg, Germany) used were rectangular, the force constant was 40 Nm–1, and the resonance frequency was 250–390 kHz.
In the current studies, we isolated a synaptosomal fraction from cortex and used it as a source for synaptic vesicle protein complexes (Huttner et al., 1983 ). After solubilization with a nonionic detergent, we performed immunoprecipitation with a myosin-Va antibody. We found that myosin-Va was associated with a 35-kDa protein in the presence of 10–6 M Ca2+ and Mg2+/ATP. This 35-kDa protein was recognized by antibody against syntaxin-1A, a membrane protein that is also known as t-SNARE and is involved in regulated exocytosis (Li and Chin, 2003 ). Association of syntaxin-1A with myosin-Va required Ca2+ and Mg2+/ATP (Figure 1A). After treatment with Ca2+, myosin-Va remained associated with synaptic vesicles purified from the cortex (Figure 1B).
A GST pull-down study using syntaxin-1A mixed with rat brain homogenate in (Ohyama et al., 2002 ) revealed a 190-kDa protein that bound specifically to syntaxin-1A in the presence of Ca2+ and ATP. Using mass spectroscopy, we confirmed that this protein was myosin-Va. Furthermore, binding of brain myosin-Va to syntaxin-1A required the presence of both Ca2+ and ATP (Figure 1C). This association of syntaxin-1A and myosin-Va required at least 10–6 M Ca2+, corresponding to a physiological elevation of Ca2+, whereas two other syntaxin-1A-binding proteins, Munc-18, and tomosyn (Ohyama et al., 2002 ), bound to syntaxin-1A in the absence of Ca2+ (Figure 1D). Although the interaction between myosin-Va and syntaxin-1A required ATP (Figure 1C), nonhydrolyzable analogues of ATP and ADP also enhanced this binding (Figure 1E). Kinetic analysis of this binding using plasmon resonance revealed that the stoichiometry of binding was 0.77 ± 0.12 (mean ± SD; n = 7), implying a 1:1 interaction between syntaxin-1A and myosin-Va dimer. Rat brain contained other myosins, such as myosin-I and -IIB, but these did not bind to syntaxin-1A (Figure 1F).
We next examined whether the properties of myosin-Va are altered by F-actin. Syntaxin-1A cosedimented with both actin and myosin-Va (Figure 2A). Syntaxin-1A could bind to the myosin-Va–actin complex, and actin could associate with the myosin-Va–syntaxin-1A complex (Figure 2B), indicating that these three proteins can form a complex. Myosin-Va ATPase is activated by Ca2+ and actin (Cheney et al., 1993 ), and, interestingly, this enhancement of ATPase was completely inhibited by syntaxin-1A binding at pCa = 6 (Figure 2C). In contrast, the F-actin myosin Va-dependent sliding motility was unchanged under these conditions (Figure 2D). Thus, at pCa = 6, the binding of myosin-Va to syntaxin-1A occurs without a large loss of ATP due to hydrolysis and without an effect on motility.
We found that the binding site of myosin-Va lies between amino acids 191 and 240 of syntaxin-1A, which comprises the first two-thirds of its H3 domain (Figure 3; A; Li and Chin, 2003 ). We also screened the mutated fragments derived from syntaxin-1A [191-240] without the myosin-Va-binding activity and found that the L222E mutant (syntaxin-1A [191-240 L222E]) lacks myosin-Va binding activity (Figure 3, A and B). Specifically, silver staining showed that, in the presence of Ca2+, the only protein that was unique to syntaxin-1A [191-240] was of 190 kDa (Figure 3A). This 190-kDa protein was identified as myosin-Va using mass spectrometry. A separate experiment confirmed that myosin-Va bound specifically to syntaxin-1A [191-240] but not to syntaxin-1A [191-240 L222E] (Figure 3, B and C). Furthermore, complex formation between myosin-Va and syntaxin-1A was inhibited by syntaxin-1A [191-240] but not by syntaxin-1A [191-240 L222E] (Figure 3D). We confirmed that syntaxin-1A [191-240] did not affect the SNARE complex formation (i.e., the SDS-resistant 80-kDa complex; Hayashi et al., 1995 ; Figure 3, E and F). Thus, the myosin-Va-binding fragment [191-240] inhibits the association of myosin-Va and syntaxin-1A.
Chromaffin cells are a typical model system for analyzing exocytosis, and they are more easily studied than other systems such as central neurons (Burgoyne and Morgan, 2003 ). Moreover, it is the most suitable system for examining whether a biochemical interaction plays a physiological role in exocytosis (Fisher et al., 2001 ; Ohyama et al., 2002 ; Quetglas et al., 2002 ). For these reasons, we used chromaffin cell exocytosis to investigate the function of the Ca2+-dependent interaction between syntaxin-1A and myosin-Va. Amperometric measurements were used to examine the physiological role of myosin-Va–syntaxin-1A binding because it is a powerful method not only for quantitative measurement of exocytosis but also for characterizing the mechanism of exocytosis (Segre et al., 2000 ; Fisher et al., 2001 ). We therefore performed an amperometric assay of catecholamine release for dense-core vesicles in chromaffin cells (Ohyama et al., 2002 ; Quetglas et al., 2002 ), which are known to possess myosin-Va (Rosé et al., 2003 ). We found that the syntaxin-1A [191-240] specifically reduced the exocytotic frequency, whereas syntaxin-1A [191-240 L222E] fragment had no effect (Figure 4, A and B).
All of the known membrane-associated myosin-Va-binding proteins bind to its globular tail (Reck-Peterson et al., 2000 ). Therefore, to investigate where on myosin-Va syntaxin-1A binds, we first examined the binding of syntaxin-1A to DHM5, a recombinant myosin-Va protein the lacks the globular tail (Homma et al., 2000 ). DHM5 did not bind VAMP-2 (our unpublished data) because it lacks a VAMP-2-binding site (Ohyama et al., 2001 ). Surprisingly, DHM5 bound syntaxin-1A in the presence of Ca2+ (Figure 5A). The Ca2+ dependence of DHM5 binding by syntaxin-1A was similar to that of native brain myosin-Va (Figures (Figures1C1C and and5B),5B), and the minimal requirement of Ca2+ for binding was pCa = 6.6 (Figure 5C). In addition, using truncated myosin-Va produced by in vitro translation, we confirmed that the whole head-and-neck portion could bind to syntaxin-1A in the presence of Ca2+ and ATP. Removal of the neck region resulted in a substantial loss of syntaxin-1A binding. In addition, a head-and-neck fragment lacking the actin binding site still bound syntaxin-1A, but binding was absent when the ATP binding site was deleted (Figure 5D). Further studies in a bacterial two-hybrid assay, which is a modification of the yeast two-hybrid method, indicated that the six IQ-motifs of the neck bind to syntaxin-1A as well as CaM, although the first IQ alone does not mediate binding (Figure 5E). Collectively, these results demonstrate that the binding site for syntaxin-1A is in the neck of myosin-Va.
CaM is known to be released from the neck of myosin-Va in the presence of micromolar Ca2+ (Cameron et al., 1998 ; Homma et al., 2000 ). We confirmed that preincubation of myosin-Va with Ca2+ released the bound CaM (Figure 5F). In addition, after this treatment, syntaxin-1A could bind myosin-Va in the absence of Ca2+ (Figure 5F). Thus, we suspected that the Ca2+-dependent syntaxin-1A-myosin-Va binding is due to the Ca2+-dependent release of CaM from the myosin-Va neck; in other words, the binding of syntaxin-1A occurs at the same site as CaM.
Next, we directly visualized syntaxin-1A-myosin-Va binding by AFM, a new technology for imaging biological molecules at nanometer resolution (Horber and Miles, 2003 ). The AFM studies show that syntaxin-1A binding occurs between two heads of the myosin dimer and not in the head or tail (Figure 6, A and B; Cheney et al., 1993 ). Similar to these AFM findings, rotary shadowing views of this complex reveal that the binding site was between the two heads and distinct from the head or the tail (Katayama, Watanabe, and Igarashi, unpublished observations). This is also the first report that the IQ-motif binds proteins other than the myosin light chains or CaM family proteins (Cheney et al., 1993 ; Vale, 2003 ). Homma et al. (2000 ) suggested that Ca2+-dependent CaM release most likely occurs at the sixth IQ motif. Our AFM results provide further support for this possibility because they showed that syntaxin-1A binds close to bifurcation of the neck region of myosin-Va (Figure 6, A and B).
Immunoprecipitation from brain homogenate further showed that the myosin-Va–syntaxin-1A complex bound SNAP-25 and VAMP-2, two neuronal SNAREs involved in exocytosis (Figure 7A). Because VAMP-2 binds to the tail of myosin-V (Prekeris and Terrian, 1997 ; Ohyama et al., 2001 ), we examined whether the SNARE complex can be bound by a complex between syntaxin-1A and DHM5, the truncated form of myosin-Va lacking a tail (Figure 5, B and C). We first confirmed that the binding of syntaxin-1A to DHM5 saturated at a 1:1 ratio. At concentrations below saturation, VAMP-2 and SNAP-25 bound to the DHM5–syntaxin-1A complex quantitatively (Figure 7B). Immunoprecipitation further showed that the myosin-Va–syntaxin-1A complex did not associate with NSF or α-SNAP, proteins that dissociate the SNARE complex (Duman and Forte, 2003 ; Figure 5A), and reconstitution studies revealed that, in the presence of VAMP and SNAP-25, syntaxin-1A associates with either α-SNAP/NSF or DHM5 (Figure 7C). Similarly, we found that the SNARE complex interacts with either α-SNAP/NSF or DHM5 (Figure 7D). These results demonstrate that myosin-Va can bind the SNARE complex including VAMP-2 and SNAP-25 and that NSF/α-SNAP can release myosin-Va from the SNARE complex.
We generated an antibody specific to the neck domain of myosin-V (Figure 8A). This antibody inhibits the myosin-Va–syntaxin-1A interaction as effectively as syntaxin-1A [191-240] (Figure 8B). The antibody did not affect formation of the SNARE complex (Figure 8C) nor did it significantly reduce the sliding velocity of myosin-Va (0.24 ± 0.15 μm/s; n = 70; p > 0.1 based on Student's t test; Figure 8D; see also Figure 2D).
Like syntaxin-1A [191-240], the neck-specific antibody reduced the exocytotic frequency as measured by amperometry (Figure 9, A–C). In this particular experiment, we stimulated the cells for 5 min and analyzed the exocytotic frequency in the initial (0–1 min) and sustained phases (1–5 min) to detect in which step the syntaxin-1A–myosin-Va interaction is involved. The anti-myosin-Va neck antibody reduced the total number events during the full 5 min of stimulation (Figure 9B). Interestingly, the reduction of the event frequency was predominant not in the initial phase but in the sustained phase (Figure 9C).
To further define at which step this interaction is involved, chromaffin cells were injected with the anti-myosin-Va neck antibody and stimulated with high K+ for 5 min. This stimulation was for a much longer time than we used for studies of inhibition by syntaxin-1A [191-240] (4 s; Figure 4). The wave patterns reveal that the anti-neck antibody inhibited exocytosis (Figure 9A). Also, the sum of events over the entire 5-min period shows that anti-myosin-Va neck antibody significantly reduced exocytosis compared with the normal IgG (Figure 9B). Interestingly, exocytotic release during the first minute (0–1 min; initial phase) was not affected by the anti-myosin-Va neck antibody, but this antibody severely attenuated exocytosis during the next 4 min (1–5 min; sustained phase) (Figure 9C).
The current model of exocytosis, based on the SNARE mechanism (Duman and Forte, 2003 ), does not completely account for the fact that Ca2+ is required at several steps of vesicular recycling (Burgoyne and Morgan, 2003 ). In this study, we found a submicromolar Ca2+-dependent interaction between myosin-Va, a putative molecular motor for synaptic vesicles, and syntaxin-1A, a neuronal membrane t-SNARE. We also presented evidence that this interaction contributes to the regulation of exocytosis in chromaffin cells. Our current study is the first clear and detailed demonstration that the Ca2+-dependent binding site for syntaxin-1A is neck rather than its tail of myosin-Va.
We applied two probes to inhibit this interaction specifically: the myosin-Va-binding fragment (syntaxin-1A [191-240]), and an anti-myosin-Va neck antibody. We found that the exocytotic frequency was reduced by both probes, indicating that they inhibited the association of myosin-Va and syntaxin-1A. Furthermore, these results confirmed this interaction participates in the regulation of exocytosis.
We next asked in which step of exocytosis this interaction functions. Exocytotic vesicles are classified into readily releasable and reserve pools. The readily released pool is released first, and the reserve pool is released after the former is depleted (Rettig and Neher, 2002 ). In amperometric analysis, the frequency of the exocytotic response in the initial phase corresponds to the number of docked or readily releasable vesicles, and the frequency in the sustained phase represents the release of the newly recruited vesicles (Kumakura et al., 2004 ). The pronounced inhibition of the frequency in the sustained phase by the anti-myosin-Va neck antibody indicates that the interaction between myosin-Va and syntaxin-1A affects the recruitment of vesicles to the readily releasable pool. Therefore, these results, together with the fact that myosin-Va is a cargo-conveying motor molecule (Reck-Peterson et al., 2000 ), suggest that the interaction between myosin-Va and syntaxin-1A affects the process of vesicle mobilization from the reserve pool (i.e., replenishment of the docked vesicle pool).
As in trafficking via the Golgi apparatus, we anticipate that the exocytotic vesicle tethering process is mediated by a long coiled-coil protein that regulates the vesicle-target membrane distance at a point before fusion (Li and Chin, 2003 ; Gillingham and Munro, 2003 ). Myosin-V, which has a long coiled-coil shaft, is likely involved in this process (Cheney et al., 1993 ). Myosin-Va on the vesicles binds to syntaxin-1A at the plasma membrane, and, along with other putative tethering molecules (i.e., Rab proteins and/or the exocyst complex), induces vesicular tethering and exocytosis. It also is thought that myosin-VI, a minus-end motor, plays a role in endocytosis (Hasson, 2003 ). Thus, as depicted in Figure 10, it is plausible that myosin-Va, a plus-directed motor (Cheney et al., 1993 ), is involved in exocytotic events. This possibility is strongly supported by a very recent report that movements of insulin-containing dense-core secretory vesicles along the cortical actin network depend on myosin-Va and are essential for regulated exocytosis (Varadi et al., 2005 ). In addition, syntaxin-1 is localized close to the site of exocytosis (Stanley et al., 2003 ; Ohara-Imaizumi et al., 2004 ), where it could participate in the process of exocytosis by interacting with myosin-Va.
A previous report indicated that hippocampal slices of dilute lethal mice, which lack myosin-Va, do not show a significant defect in glutamate release (Schnell and Nicoll, 2001 ), which seems to contradict our current results. This may be due to differences in the myosin-Va dependence of glutamate and catecholamine release from dense-core vesicles. Despite this contradictory data, our findings are consistent with recent reports that retinal neurotransmitter release is abnormal in dilute lethal mice (Libby et al., 2004 ) and that dilute lethal mutant mice have a defect in basal neurotransmitter release and presynaptic plasticity (Trinchese et al., 2003 ). This latter report used cultured neurons derived from 1-d-old mice, whereas Schnell and Nicoll (2001 ) used tissue slices obtained from 12- to 19-d-old mice, suggesting that the differences may be due to the age of the mice used in the studies. Furthermore, according to a recent detailed immunohistochemical study the brain (Tilelli et al., 2003 ), the hippocampus does not express much myosin-Va, and most of the myosin-Va immunoreactivity is associated with the neuronal cell bodies rather than the neuropila, which are enriched with synaptic terminals. Thus, differences in the levels of myosin-Va in hippocampal synapses mutant between wild-type and the dilute lethal may be subtle. Finally, other proteins, such as additional myosin-V isoforms, may compensate for the loss of myosin-Va function in the dilute lethal hippocampal slices (Schnell and Nicoll, 2001 ; Vale, 2003 ), but not in cell culture.
CaM binds to the myosin-Va neck as a light chain via its IQ-motifs (Cheney et al., 1993 ) but is released when the intracellular Ca2+ rises to micromolar concentrations (Cameron et al., 1998 ; Homma et al., 2000 ). We found that, after Ca2+-dependent release of CaM, syntaxin-1A can bind myosin-Va even in the absence of Ca2+. Thus, we suspected that the apparent Ca2+ dependence of myosin Va-syntaxin-1A binding is due to Ca2+-dependent release of CaM from the myosin-Va neck, and exposure of an otherwise concealed syntaxin-1A binding site. Our results further indicate that CaM can sense submicromolar Ca2+ through the release of CaM from myosin-Va (Cameron et al., 1998 ; Homma et al., 2000 ); the binding between myosin-Va and syntaxin-1A requires at least 0.3 μM intracellular Ca2+ (Figure 4B), which corresponds to the level of Ca2+ when secretory granules enter the readily releasable pool (Burgoyne and Morgan, 2003 ).
CaM, the most abundant Ca2+-sensitive protein, may be widely responsible for micromolar Ca2+ sensitivity (Burgoyne and Clague, 2003 ). We demonstrated previously that exocytosis is regulated by Ca2+/CaM-dependent protein kinase II (CaMKII), which binds to syntaxin-1A in a Ca2+-dependent manner when it is autophosphorylated (Ohyama et al., 2002 ). Although the CaM-binding sites of myosin-Va and CaMKII are distinct (Bähler and Rhoads, 2002 ), their Ca2+-dependencies for syntaxin-1A binding are very similar. Our current studies also could explain the involvement of CaM in exocytosis (Sakaba and Neher, 2001 ) and other CaM-dependent interactions (Junge et al., 2004 ).
Our results demonstrate that submicromolar Ca2+ concentrations induce the 1:1 binding of syntaxin-1A to the myosin-Va neck. This further suggests that syntaxin-1A might mimic CaM for binding at the vacant IQ-motifs. Our results suggest that this complex modulates SNARE-dependent interactions and regulates at least the exocytosis of dense-core vesicles by forming a link between vesicles and their targets; in other words, the complex regulates the recruitment of the vesicles to the readily releasable pool during sustained secretion. Recently, the Ca2+/CaM-dependent interaction between the cargo-conveying tail and the head of myosin-Va was shown to be an important factor regulating its conformation (Krementsov et al., 2004 ; Li et al., 2004 , 2005 ; Wang et al., 2004 ). Ca2+/CaM is an important regulator of both exocytosis and the cargo-conveying activity of myosin-Va, and further studies of this novel interaction should help elucidate the roles of myosin-Va in Ca2+-regulated exocytosis.
We thank all of the donors for the cDNAs and the antibodies; E. Akaishi-Onodera and M. Sato-Igarashi for technical assistance; and T. Abe, T. Ando, O. Arancio, P. C. Bridgman, E. Katayama, K. Hoshino, M. Norita, and M. Takahashi for helpful discussions. This work was supported by Grants-in-Aid from the Ministry of Education, Sciences, Culture, Sports, and Technology of Japan (#15029218, #16015240, #16044216, and #17023019 to M. I.); the Japan Society for the Promotion of Sciences (#15300123 to M. I.); the Life Science Foundation (to M. I.); the Brain Science Foundation (to M. I.); the Yujin Memorial Foundation (to M. I.); the project-promoting grant from Niigata University (to M. I.); and the Tsukada Milk Foundation for Medical Research (to M. W.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E05–03–0252) on July 19, 2005.
Abbreviations used: AFM, atomic force microscopy; CaM, calmodulin; CaMKII, Ca2+/CaM-dependent protein kinase II; GST, glutathione S-transferase; RU, resonance units.