1To investigate the role of actin in vertebrate nerve terminals, nerve-muscle preparations from garter snake (Thamnophis sirtalis) were treated with the actin-depolymerizing agent latrunculin A. Immunostaining revealed that actin filaments within presynaptic motor terminal boutons were disrupted by the drug.
2In preparations loaded with the optical probe FM1-43, destaining was reduced by latrunculin treatment, suggesting that transmitter release was partially blocked.
3Latrunculin treatment did not influence the amplitude or time course of spontaneous miniature endplate potentials (MEPPs). Similarly, endplate potentials (EPPs) evoked at low frequency were comparable in control and latrunculin-treated curarized preparations.
4Brief tetanic stimulation of the muscle nerve (25 Hz, 90 s) depressed EPP amplitudes in both control and latrunculin-treated preparations. After tetanus, EPPs elicited at 0.2 Hz in control preparations recovered rapidly (0–5 min) and completely (usually potentiating to above pre-tetanus levels; 130 ± 11%, mean ±s.e.m.). In contrast, EPPs evoked in latrunculin-treated preparations recovered slowly (8–10 min) and incompletely (84 ± 8%).
5The influence of latrunculin on post-tetanic EPPs depended on its concentration in the bath (KD= 3.1 μm) and on time of incubation.
6These observations argue that actin filaments facilitate transmitter release rather than impede it. Specifically, actin may facilitate mobilization of vesicles towards the releasable pools.
Actin filaments are ubiquitous in vertebrate nerve terminals during development, where they contribute to growth cone motility via at least two mechanisms: polymerization and interaction with myosin motors (Purves & Lichtman, 1985). After development, actin continues to be expressed even though motility largely ceases. Indeed, it remains a significant cytoskeletal element in adult terminals, along with microtubules (Drenckhahn & Kaiser, 1983; Hirokawa et al. 1989). Although details of its function in adults remain unclear, actin has been implicated in certain steps of the vesicle processing cycle. One view is that an actin meshwork sequesters synaptic vesicles in the ‘reserve’ pool, keeping them from reaching release sites prematurely (Wang et al. 1996). In this model, actin binds synapsin I, which in turn binds transmitter-containing vesicles (Llinás et al. 1985; Hirokawa et al. 1989; Greengard et al. 1993). Consistent with the model, synapsin I knockout mice have smaller reserve pools and more rapid depletion of vesicles (Chi et al. 1999). In contrast, disruption of filamentous actin with cytochalasin D treatment eliminates refilling of the reserve pool but not the local ‘exo/endo cycling pool’ in Drosophila mutant shibire nerve terminals (Kuromi & Kidokoro, 1998). This observation implicates actin as a component in vesicle mobilization pathways, not in sequestration of vesicles. Myosin motors are also reported to participate in the mobilization of vesicles to release sites (Ryan, 1999), perhaps interacting with actin in the process. Finally, there is evidence that actin is required for endocytosis (Gustafsson et al. 1999) and initial movement of vesicles away from endocytic sites (Merrifield et al. 1999). Given these disparate (though not mutually exclusive) putative roles for filamentous actin, it is difficult to predict the dominant net effect(s) of its disruption. For example, if actin served to sequester vesicles as a negative regulator, one might expect enhanced release with disruption, at least until the reserve pool is depleted. Conversely, if actin helps myosin to mobilize vesicles, actin disruption should have the same effect as myosin blockade, namely diminution in the number of vesicles available for release (Ryan, 1999). And if actin is required for endocytosis, one would expect its disruption to eventuate in total depletion of vesicles and cessation of synaptic activity.
To assess the dominant role of actin in transmitter release, we used the actin depolymerizing agent latrunculin A (latrunculin) in experiments designed to measure nerve terminal activity in two ways: activity-dependent staining and destaining of terminals and electrophysiological recording from the postsynaptic cell. Latrunculin treatment significantly reduced evoked release both during and following a short tetanus, but did not affect spontaneous or low-frequency evoked release. Consistent with this, the agent also dramatically impaired destaining of nerve terminals loaded with the activity-dependent probe FM1-43, while having a minor effect on dye loading, and this only when nerve terminals were subjected to very high stimulation levels. Taken together, our results suggest that actin positively affects vesicle mobilization, probably between reserve and releasable vesicle pools, during high levels of nerve terminal activity. Lesser roles may also be indicated in vesicle mobilization to the reserve pool and endocytosis. We see no evidence for vesicle sequestration by actin. Parts of this paper have been published in abstract form (Cole et al. 1999).
Garter snakes (Thamnophis sirtalis) were immersed in cold water for 10 min and killed by rapid decapitation. Several contiguous segments of the transversus abdominis muscle were dissected from the animal, placed in reptilian saline solution, and divided as needed to provide individual two to three segment experimental and control nerve-muscle preparations. Details of the muscle anatomy, dissection procedure and saline composition are described elsewhere (Wilkinson & Lichtman, 1985). All experiments were carried out according to the guidelines established by the Animal Studies Committee of Washington University.
Visualization and disruption of actin
Actin was visualized in nerve terminals using a commercial fluorescein-conjugated monoclonal antibody (FITC-anti β-actin; Sigma, St Louis, MO, USA). This reagent cross-reacted with neuronal but not muscle actin in snake, thereby permitting visualization of terminals without muscle actin background. Preparations were fixed with ice-cold methanol for 5 min, then incubated in the antibody solution (1:200 in PBS) at room temperature (RT) for 30 min.
Actin was disrupted using latrunculin A (Biomol Research Laboratories, Inc., Plymouth Meeting, PA, USA), a sea sponge toxin that inhibits actin polymerization and is 10- to 100-fold more potent than cytochalasins in cultured cell assays (Spector et al. 1983, 1989). The drug was dissolved and stored in DMSO (1 mM stock concentration). Aliquots of drug stock were diluted in reptilian saline to a final working concentration of 15 μM (1.5 % DMSO) unless otherwise stated. Initially, we found that activity varied among aliquots, perhaps due to precipitation. For this reason, every aliquot of latrunculin was tested on cultured fibroblasts to confirm that the drug was in solution and active. The normally flat, fibroblasts ‘balled up’ when bathed in active drug solution for < 10 min, indicating disruption of filamentous actin. Control preparations were incubated in reptilian saline containing equimolar amounts of DMSO only. Separate experiments confirmed that DMSO (≤ 2.5 %) had no independent effect. Unless otherwise noted, all preparations were incubated at RT for 2 h.
Preparations were placed in a dish on the stage of an inverted microscope equipped with differential interference contrast optics. For electrical stimulation the cut end of the muscle nerve was drawn into a hook-in-oil electrode and negative rectangular pulses (200 μs) were delivered. Stimulus amplitude was set supramaximal (2-5 V) as judged by visible contraction of the muscle. Curare (≤ 35 μM) was then applied to prevent motion artifact during recording. Stimuli were delivered in a fixed sequence: 0.2 Hz for 30–60 s, 25 Hz for 90 s, and 0.2 Hz for several minutes. The length and frequency of the tetanic portion of the stimulus protocol (25 Hz, 90 s) were shown in initial experiments to be the minimum required to consistently reveal the effect of latrunculin. All preparations were rested for 15 min between recording sessions. miniature endplate potentials (MEPPs) and endplate potentials (EPPs) were recorded from visualized endplates with microelectrodes filled with 3 M KCl (20-50 MΩ); records were analysed using pCLAMP 7.0 software (Axon Instruments, Inc., Foster City, CA, USA).
Activity-dependent staining and destaining
An equimolar substitution of KCl (60 mM) for NaCl was made to provide high-K+ reptilian saline solution for chemical depolarization of terminals. For activity-dependent staining, FM1-43 (6-13 μM; Molecular Probes, Eugene, OR, USA) was dissolved in the high-K+ reptilian saline and applied to preparations (3 min each) which had been pre-incubated in either latrunculin or control solutions. Each preparation was washed quickly (15 s) with normal reptilian saline and then rinsed with fixative (1 % formaldehyde in 100 mM sodium phosphate buffer) until clear. Fixation was for 30 min at RT, followed by a 30 min wash in PBS.
In destaining experiments, the above protocol was modified. First, FM1-43 application preceded latrunculin treatment to eliminate the latter's effects on dye loading as a confounding variable when quantifying destaining (see Results). Second, the FM1-43 application time was reduced to 1 min so that the initial staining was optimized to reveal changes in destaining levels. Third, preparations were destained using dye-free high-K+ reptilian saline for 45 min before fixation.
Light microscopy and data analysis
To quantify FM1-43 staining and destaining, preparations were imaged with an upright microscope equipped with a Zeiss ×63, 1.4 NA objective and a confocal adaptor (fluorescein optics; Bio-Rad MRC 1024; Hercules, CA, USA). Collected image stacks (8-25 images at planes separated by 0.5 μm) were analysed as described elsewhere (Teng et al. 1999) using software provided by Bio-Rad. Briefly, the brightness averages of four to five terminals from each preparation were measured, corrected for background, and averaged together.
Our initial approach towards relating actin and transmitter release was to visualize actin in nerve terminals and determine if its location or appearance changed with activity. Living terminals were removed from their endplates (to eliminate background staining from muscle actin; see Wilkinson & Lunin, 1994) and treated with the actin-binding drug fluorescein-phalloidin. Actin was present within each bouton, but stimulation of the muscle nerve had no reproducible effect on its distribution. Similar negative results were obtained using an antibody against mammalian β-actin (see below). We therefore undertook the alternative course described here, namely to examine the effect of actin disruption on transmitter release. The agent latrunculin A was chosen because (at least at the concentrations and incubation times studied) it appeared to disrupt actin filaments in terminal boutons without noticeably disrupting muscle fibres. Cytochalasin B was used initially with results similar to those described below, except that some disruption of muscle fibres occurred (n= 3 muscles from 3 snakes).
Immuno- and activity-dependent staining
The bouton's actin filaments were found predominantly in the cortex (just inside the plasma membrane) and were brightly labelled by phalloidin (see above) and by anti-β-actin (Fig. 1A and B; n= 16 terminals from 4 snakes). After latrunculin treatment, the label became barely visible; moreover, it was no longer localized to the cortex but was distributed throughout the bouton (Fig. 1C and D; n= 16 terminals from 4 snakes). This redistribution of actin staining was the same whether or not the terminal was stimulated (data not shown).
The effect of actin disruption on activity-dependent uptake of FM1-43 is shown in Fig. 1E and F. We examined uptake with both brief and prolonged stimulation. In snake, brief electrical stimulation at reduced temperature (5 Hz for 5–20 s at ∼7°C) results in a punctate distribution of dye uptake due to clathrin-mediated endocytosis near active zones (Teng et al. 1999). Latrunculin treatment had no effect on this form of dye uptake (data not shown; control, n= 6 terminals from 2 snakes; latrunculin, n= 5 terminals from 2 snakes). In contrast, KCl depolarization (or more intense electrical stimulation) recruits a second endocytic process which involves macropinocytosis. Dye uptake induced by this more intense stimulation was slightly inhibited in terminals which had been previously treated with latrunculin (Fig. 1F; 82 ± 13 % of control staining, mean ±s.d., n= 9 terminals from 2 snakes, P < 0.02).
Destaining of terminals loaded with FM1-43 was more sensitive to latrunculin treatment than was staining. However, the order of application of the two agents, latrunculin and FM1-43, was crucial. In initial experiments, terminals already treated with latrunculin were stained as in Fig. 1F (KCl depolarization with FM1-43) and then destained (KCl depolarization). With this protocol destaining was not systematically affected by latrunculin. In two experiments latrunculin treatment slightly enhanced destaining, while in three others destaining was slightly diminished (data not shown; destaining protocol as in Fig. 1 legend). In contrast, terminals which were first loaded with FM1-43 and then treated with latrunculin behaved differently from control terminals loaded with FM1-43 only (Fig. 1E). Staining intensity after latrunculin treatment (n= 4 terminals from 1 snake) was unchanged from that of controls (n= 8 terminals from 2 snakes), indicating that latrunculin did not in itself initiate vesicle fusion or some other process which would destain the terminal. Destaining was partially blocked, however, indicating that fewer vesicles underwent exocytosis (Fig. 1G; expressed as a percentage of control staining: control, 25 ± 7 %; mean ±s.d., n= 8 terminals from 2 snakes; latrunculin-treated, 45 ± 17 %; mean ±s.d., n= 8 terminals from 2 snakes; P < 0.005). Thus the most pronounced effect of actin disruption was a decrease in activity-dependent destaining, which is associated directly with transmitter release (bright spots in Fig. 1E and G are probably endosomes; see Teng et al. 1999). Activity-dependent staining, which is associated indirectly with transmitter release due to putative ‘matching’ of exo- and endocytosis (see Betz & Bewick, 1993) was demonstrable but less pronounced.
Because of the time required for latrunculin treatment (>1 h; see below), it was not feasible to record EPPs from single endplates before and after exposure to the drug. Rather, two to three recordings from randomly selected endplates in control and latrunculin-treated muscles were compared. Care was taken to ensure that experimental conditions were as similar as possible: paired control and treated preparations comprised adjacent segmental components of one contiguous muscle, while incubation times, saline composition and amount of added curare were identical.
Latrunculin treatment had no discernable effect on the amplitude, rise time, or frequency of MEPPs (data not shown), nor did it affect the amplitude or rise time of EPPs elicited by low frequency stimulation (< 2 Hz; see Fig. 2G). In contrast, latrunculin clearly influenced EPPs elicited at high frequency. The following protocol was therefore adopted. After delivery of low frequency stimuli (0.2 Hz) to establish a baseline, a brief tetanus was delivered, during which both control and latrunculin-treated neuromuscular junctions (NMJs) exhibited depression of EPP amplitudes. Following the tetanus, the stimulus was returned to its previous low frequency and continued for several minutes. Latrunculin-treated preparations differed significantly from controls in two ways: EPP amplitudes were more severely depressed during the tetanus, and recovery from tetanus was delayed or incomplete. Some representative recordings are shown in Fig. 2. Depression at the end of tetanus was severe in both preparations, but significantly more so with latrunculin treatment (compare EPPs B and E in Fig. 2). Moreover, the control NMJ potentiated (see below) and then returned to pre-tetanus EPP amplitudes while the latrunculin-treated NMJ recovered slowly, requiring several minutes to reach pre-tetanus values. Figure 3A summarizes the results from all NMJs examined using the protocol of Fig. 2. Diminution of EPP amplitudes by latrunculin was significant both at the end of tetanus (34 ± 19 % of control end-tetanus amplitudes; mean ±s.d., n= 6 terminals from 3 snakes; P < 0.003) and for several minutes after it (n= 6 terminals from 3 snakes; P < 0.02 at 2 min; P < 0.2 at 5 min). Interestingly, for control NMJs the amount of depression evident during the tetanus predicted the amount of post-tetanus recovery. As is shown in Fig. 3B, EPP amplitude at the end of tetanus was positively correlated with the maximum EPP amplitude observed immediately post-tetanus (both expressed as a percentage of pre-tetanus amplitude). This correlation was not apparent in NMJs treated with latrunculin (Fig. 3B). As can be seen from the range of post-tetanus EPP amplitudes in Fig. 3B, some but not all control NMJs potentiated following tetanus. The mechanism of latrunculin action, however, appeared unrelated to potentiation; its effect on recovery was robust whether or not the corresponding control synapse exhibited post-tetanic potentiation.
The action of latrunculin depended on its bath concentration, and on the time of incubation. As is shown in Fig. 4, the concentration chosen for the experiments described above (15 μM) was near saturation, but the incubation time (2 h) was probably insufficient for maximal effect. This shorter time was chosen in part for convenience and in part to promote the stability of the preparations.
Actin filaments were found present in snake motor terminals and were disrupted by the agent latrunculin A. Contrast enhancement of weak staining in latrunculin-treated terminals revealed a diffuse pattern throughout the terminal boutons, not the localized pattern of bright staining seen in the cortex of untreated boutons. This could have been due to immunolabelling of actin fragments or monomers, perhaps with decreased affinity, or to other non-specific staining that had been masked by the staining of filamentous actin in control preparations.
As mentioned above, actin has putative roles in endocytosis, vesicle sequestration, and vesicle mobilization, any of which might underlie the observed depression of transmitter release. Results from activity-dependent staining and destaining experiments were most consistent with mobilization for the following reasons. Both control and latrunculin-treated terminals took up FM1-43 from the surrounding bath in an activity-dependent manner. There was no difference in uptake at lower activity levels, and only slightly less uptake relative to controls in latrunculin-treated terminals with high activity. Thus actin disruption appeared to have only a minor negative effect on endocytosis. In contrast, destaining was significantly blocked by latrunculin. With the protocol used, the reserve pools of control and latrunculin-treated terminals were identically loaded with FM1-43, as evidenced by equal staining intensity just prior to destaining. Upon depolarization in a bath free of FM1-43, a reduction, minor or otherwise, of endocytosis of unstained vesicles into an already stained reserve pool would seem unlikely to block that pool's destaining. Similarly, vesicles putatively sequestered by actin would be freed by latrunculin; this would enhance release and destaining, not block it. Thus the observed blockade of destaining by latrunculin argues in favour of the third option – decrease in vesicle mobilization, probably from the reserve pool to release sites.
In electrophysiological experiments, latrunculin acted in a dose- and time-dependent manner to depress end-tetanus and post-tetanus EPPs in snake motor terminals. Control post-tetanus EPPs exhibited rapid full (or potentiated) recovery while the corresponding EPPs in latrunculin preparations were significantly depressed for long periods (∼5 min). This post-tetanic effect was predictable, because latrunculin end-tetanus EPPs were smaller than control EPPs. The question arises as to whether these phenomena could be postsynaptic. Receptors can be responsible for short-term synaptic depression (Magleby & Pallotta, 1981), and actin influences receptor localization while its disruption causes changes in receptor clustering (Hall et al. 1981; Bloch, 1986; Luther et al. 1996). Latrunculin had no effect on MEPPs or on EPPs elicited by low-frequency stimulation, observations which would ordinarily rule out postsynaptic mechanisms. However, we saw no presynaptic effects of latrunculin under these conditions either. Thus the strongest evidence of a presynaptic locus for latrunculin action was its robust blockade of FM1-43 destaining, but a concomitant postsynaptic effect cannot be ruled out.
Our electrophysiological results confirmed those from destaining in demonstrating that transmitter release is diminished by latrunculin. Latrunculin's effects may have resulted directly from actin disruption or indirectly via removal of some other component(s) which normally interacts with actin. This occurred at higher stimulus intensities: low frequency release was not affected, suggesting that the releasable pool functioned normally. One mechanism consistent with both optical and electrophysiological observations is normal vesicle depletion in releasable pools and the failure to refill them with reserve pool vesicles. We therefore propose that actin filaments interact with myosin or other motors to help mobilize vesicles from reserve to releasable pools. A similar conclusion may be drawn from experiments in which myosin light chain kinase activity was blocked in frog nerve terminals (Ryan, 1999). Similar mechanisms have been implicated in a variety of vesicle transport systems found in different cell types, including transport of endoplasmic reticulum vesicles in neurons, pigment granules in melanocytes and Golgi-derived vesicles in epithelial cells (for review see DePina & Langford, 1999).
We thank R. Roberts for help with actin staining, D. Schafer for helpful discussions about actin and latrunculin, and H. Teng for a critical reading of the manuscript. This work was supported by United States Public Health Service Grant NS-24752.