The Timing of Synaptic Vesicle Endocytosis (FM 1-43/Confocal Microscopy/Recycling) TIMOTHY A

Proc. Natl. Acad. Sci. USA Vol. 93, pp. 5567-5571, May 1996 Neurobiology The timing of synaptic vesicle endocytosis (FM 1-43/confocal microscopy/recycling) TIMOTHY A. RYAN*t, STEPHEN J SMITH*, AND HARALD REUTERt *Department of Molecular and Cellular Physiology, Stanford University, Stanford, CA 94305; and tDepartment of Pharmacology, University of Bern, CH-3010 Bern, Switzerland Communicated by Charles F. Stevens, The Salk Institute for Biological Studies, San Diego, CA, January 17, 1996 (received for review October 2, 1995)

ABSTRACT Alternative models to describe the endocyto- longed stimulation and that this slowing results from a main- sis phase of synaptic vesicle recycling are associated with time tained increase in intracellular Ca2l ([Ca2+]). scales of vesicle recovery ranging from milliseconds to tens of Details of the physiology and biochemistry of endocytosis seconds. There have been suggestions that one of the major remain scarce, because direct measurements of endocytosis models, envisioned as a slow process that occurs only after have proven more difficult to obtain than those of exocytosis. complete fusion of the vesicle membrane with the neuro- Unfortunately, although measurements of membrane capaci- lemma, might be applicable only under conditions of heavy, tance have been a useful probe of endocytosis in many systems nonphysiological stimulation. Using FM 1-43 and similar (9, 10), they are not readily applicable to typical fast synapses; fluorescent probes to label recycling synaptic vesicles in rat the distance of the release site from the cell body, the size of hippocampal neurons, we have measured the kinetics of the small clear synaptic vesicles (equivalent capacitance of 0.07 endocytosis with a wide range of action-potential-driven exo- fF per vesicle), as well as the low average number of vesicles cytotic loads. Our results indicate that when either 5% or 25% per bouton ("200) (11) make such measurements at these of the vesicle is vesicles are recovered with a nerve terminals problematic. Here we make use of the optical pool used, tracer method, pioneered in elegant imaging studies by Betz and half-time on the order of20 s (24°C). This endocytosis rate was Bewick (12, 13), to measure the residence time ofsynapticvesicles not influenced by operations designed to alter intracellular in the plasma membrane after action-potential-stimulated exo- Ca2l during membrane retrieval, suggesting that residual cytosis in synapses of cultured hippocampal neurons. Ca2+ after strong stimuli probably does not greatly retard In previous studies (14, 15), we obtained estimates of the endocytosis. Finally, we have shown that vesicle-destaining time course of endocytosis after large, potentially nonphysi- kinetics are not strongly influenced by the substantially ological stimuli. The results presented here extend these differing rates at which two marker dyes tested dissociate from measurements to stimuli one order of magnitude smaller than membranes. This observation suggests that vesicles remain those previously measured. We investigated the kinetics of open long enough for essentially complete dissociation of even membrane recovery under stimulation conditions that use as the slower dye (a few seconds) or, alternatively, that both dyes little as 5% of the available vesicle pool and compared it with readily escape vesicle membrane by lateral diffusion through the recovery of heavier exocytotic loads. Our data indicate that any exocytotic opening. These data seem most consistent with even after such stimuli, endocytosis proceeds with a slow applicability of the slow-endocytosis, complete-fusion model half-time (20 s). We have found also that endocytosis at these at low as well as high levels of exocytosis. synapses is independent of extracellular Ca2+ over a wide concentration range. Finally, we have measured the impact of The recapture of synaptic vesicle membrane by endocytosis is the tracer's dissociation rate upon the tracer's ability to escape an important step in the recycling of synaptic vesicles and is from the vesicle membrane after exocytosis and before recap- necessary, along with vesicle repriming, for the maintenance of ture. All of these measurements seem to favor the slower, a releasable pool during physiological neurotransmitter re- complete-fusion model of endocytosis at hippocampal syn- lease. The mechanisms responsible for this recapture are apses, even at action potential frequencies as low as 1 Hz. largely unknown, but two distinct models have arisen out of unresolved differences in interpretation among the classic MATERIALS AND METHODS release. ultrastructural studies of vesicular neurotransmitter Hippocampal CA1-CA3 regions were dissected from 3- to Heuser, Reese, and colleagues (1-3) proposed a model in 5-day-old Sprague-Dawley rats, dissociated, and plated onto which all recycling vesicles undergo complete fusion with the coverslips coated with Matrigel (Collaborative Research) in- plasmalemma after releasing their contents, followed by a side microwells formed by sealing a 6-mm-diameter glass selective but rather slow (tens of seconds) endocytic retrieval cylinder (Bellco Glass) onto the coverslip using silicone sealant of vesicle membrane at sites distinct from the active zone for (Dow-Corning). Most details are as described (15). Culture release. Ceccarelli and colleagues (4), on the other hand, media consisted of minimal essential media (GIBCO), 0.6% proposed that vesicles are physiologically retrieved intact at the glucose, 0.1 g/l of bovine transferrin (Calbiochem), 0.25 g/l of active zone, after a very brief (<1 s) and reversible opening insulin (Sigma), 0.3 g/l of glutamine, 5-10% fetal calf serum event. Models of this type have come to be known colloquially (HyClone), 2% B-27 (GIBCO), and 8 ,uM cytosine ,B-D- as "kiss-and-run." It has been suggested that the slower, arabinofuranoside. Cultures were maintained at 37°C in a 95% complete-fusion model is only manifest under heavier non- air/5% CO2 humidified incubator, and culture media were physiological stimulation (5, 6). These suggestions have been replaced every 3 days. supported by recent measurements from tonic retinal synapses Cells were used 3-5 weeks after plating, and the coverslip (7, 8), which indicate that endocytosis is slowed under pro- was mounted in a low volume (75 ,ul) laminar superfusion microscope chamber and superfused at a rate of 1.5 ml/min. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" inL Abbreviation: [Ca +]i, intracellular Ca +. accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed.

5567 Downloaded by guest on September 26, 2021 5568 Neurobiology: Ryan et aL Proc. Natl. Acad. Sci. USA 93 (1996) The chamber, with agar bridges and Ag-AgCl electrodes fixed on opposite sides, was mounted on the stage ofthe microscope. Unless otherwise noted, cells were continuously superfused in a saline solution consisting of 119 mM NaCl, 2.5 mM KCI, 2 mM CaCl2, 2 mM MgCl2, 25 mM Hepes (buffered to pH 7.4), 30 mM glucose, and 10 ,M 6-cyano-7-nitroquinoxaline-2,3- dione (Research Biochemicals, Natick, MA). Action potentials were stimulated by passing 1-ms current pulses yielding fields of -10 V/cm through the chamber. Cells were labeled with 15 ,uM FM 1-43, 2 ,uM FM 1-84, or 1.5 mM FM 2-10 as indicated; stimulated electrically to fire a defined train of action potentials; and then rinsed for 5-10 min in saline before destaining measurements. Changes in [Ca2+], were measured using the fluorescent Ca2+ indicator fluo-3 1.25, AM. All measurements were performed at room temperature U, 3 1.00 (-24°C). Scanning fluorescence and differential interference 0 20H contrast images were acquired at a rate of 1 image every 2.2 s cU) 0.75 using a modified Bio-Rad MRC 500 laser scanning unit CD coupled to a Zeiss IM-35 inverted microscope and a Nikon 0 x40 1.3 NA Fluor objective and stored digitally. Fluorescence 'L 0.25 was excited using the 488-nm line of an argon laser, and _ 0.500 D emissions were detected through a 515-nm-long pass filter. 0'P0.00 Digital time-lapse sequences were analyzed as described ULa (15) with the following modifications. Estimates of vesicular IL. -0.25 release of fluorescence from a given bouton (AF; see Fig. 2A) 0 20 40 60 80 100 were obtained by calculating the magnitude of the difference time (sec) in the fluorescence intensity averaged over the first 10 time FIG. 1. (A) A network of hippocampal axons and dendrites typical points (before electrical stimulation) with that averaged over of those used in these experiments, imaged by scanning Nomarski the asymptotic phase of release (the last five time points) of a differential interference contrast microscopy optics. (B) The localiza- given run. Errors given are computed SEM. tion of many individual presynaptic boutons is revealed after the cells have been bathed in saline containing 15 mM FM 1-43, stimulated for 20 s at 20 Hz, and rinsed for 5 min. Staining patterns such as these have RESULTS previously been shown to require the presence of extracellular Ca2+ Fig. 1 illustrates the imaging procedures used in the present and to colocalize with markers of presynaptic proteins (14). (C) The same field of view as in B after 100 s of stimulation at 20 Hz. The loss study. When action potentials are fired in the presence of FM of fluorescence from individual boutons, typically >90% of the 1-43, membrane endocytosed during vesicle recycling becomes original intensity, is interpreted as the result of release of the dye to labeled and provides a means of visualizing active synapses the extracellular milieu during the exocytosis of labeled synaptic after rinsing away the excess dye (Fig. 1B). This result allows vesicles. (Bar = 10 jum.) (D) The kinetics of the loss of fluorescence for later measurement of exocytosis, because vesicles then vs. time measured from 20 individual fluorescent puncta like those release the trapped dye and lose their fluorescence upon shown in B during a train of action potentials at 20 Hz. The initial slope further electrical stimulation (Fig. 1C). The magnitude of the computed from the first 5 s after the beginning of the electrical fluorescence decrease at individual puncta measures the stimulation (dashed line, slope 4.4%/s) gives an estimate ofthe relative amount of labeling that occurred during the loading phase. The impact of a train of action potentials at 20 Hz on the use of the entire, relative turnover of a releasable pool of vesicles as a function of time. This has previously the vesicle pool for given stimulus been determined by comparing the amount of dye incorporated as a frequency is determined from the kinetics of the fluorescence function of stimulus length with the rate of destaining during electrical decrease at terminals during action potential firing. The stimulation (14). kinetics of the fluorescence decrease for terminals stimulated at 20 Hz is illustrated in Fig. 1D. Analysis of these kinetics after the beginning of endocytosis were -2-3 times above allows us to estimate the extent of pool turnover for a defined background in the case of the 100 AP load and represent a train of action potentials at this frequency (15). We estimate that signal from potentially as little as 2.5% of the vesicle pool. the first 20 action potentials (1 s) release 5% ofthe total recycling Fig. 2A shows a series of the dye release curves used to pool, whereas the first 100 action potentials release -25%. measure a set of endocytosis time points after a 20 AP train. To determine the time course of endocytosis after release of The measurements were made from the same bouton, using either 5% or 25% of the recycling vesicle pool, we presented repeated loadings at different At after the same stimulus train. dye at various times (At) after the beginning of defined action Dye uptake time course curves compiled from many such potential trains at 20 Hz (Fig. 2A, Upper). The uptake of dye measurements are illustrated in 2B. One curve shows into recycling synaptic vesicles was quantified for each At as the Fig. dye fluorescence released (AF) during a later episode of maximal uptake after a 20 AP train, whereas the other depicts uptake stimulation [delivered after a 5-min delay to allow for the after a 100 AP train, both at 20 Hz. The two curves must reflect repriming ofvesicles endocytosed during the test dye exposure the time course of vesicle membrane becoming inaccessible to (14, 15)]. This procedure corrected the labeled vesicle mea- extracellular dye, and should thus represent accurately the surement for a small component of residual nonvesicular time course of endocytosis. The data for each time point is membrane staining (see Fig. 1C). It is important to note that normalized to the uptake occurring when dye is present during, the measurements of dye uptake at various times (At) after the as well as after, stimulation. For example, the first data point stimuli are normalized to the At = 0 condition, corresponding on the 20 AP load curve shows that only -10% of the total to a stimulus delivered in the presence of the dye. Measure- amount of endocytosis takes place during the time of stimu- ments for this time point should thus determine the total lation and the following 5 s (Fig. 2B, dashed line). This data, amount of endocytosis that follows a given action potential therefore, set a very restrictive upper limit on the magnitude train. The uptake of dye at later times should be diminished in of any very fast component (subsecond) of endocytosis. Con- proportion to the rate of endocytosis. The AF signals 40-60 s sequently, these data strongly suggest that most or all of the Downloaded by guest on September 26, 2021 Neurobiology: Ryan et al. Proc. Natl. Acad. Sci. USA 93 (1996) 5569 B 1.0

0.8 * -A 20 action Polential load A I \ G~~0 0100 action potential load 20 AP 100 AP S 0.6 a -- i-i- 20Hz 0.4 oc I 0 U. <-At- NFM 1-43 0.2

0.0 20 AP 0 15 30 45 60 75 At- s At=5 s Atulos at (am) 10Hz -I-- 3 0.5 20A 100SAP (I 20 Hz) (5s20Hz)1- - -- - C I 100 X 0.4 0.3 50 AF 0.2 25 q 0.1

0 40 80 120 I 0.0 tIm <"ec) Ua. FIG. 2. The kinetics of endocytosis measured at individual synaptic boutons. (A, Upper) Experimental protocol is shown. Field stimulation was used to trigger action potential firing at 20 Hz for either 1 s (20 AP) or 5 s (100 AP). All experiments were carried out in the presence of 6-cyano-7-nitroquinoxaline-2,3-dione to prevent synaptically mediated burst discharges, which otherwise could produce excess action potential firing. The time course of endocytosis was measured by applying the dye after a variable interval (delay time At) with respect to the beginning of the action potential train; endocytosis occurring during the interval At was determined from the reduction of dye uptake relative to that with At = 0. Dye uptake was quantified by measuring fluorescence release signals during a subsequent maximal stimulus. (Lower) Representative release signals for three different values of At at a single synaptic bouton (20 AP load). (B) Endocytosis time course, determined from ensemble averages of FM 1-43 uptake as a function of At, normalized to uptake at At = 0. Release signals for a given At were measured at least twice at an individual bouton and compared with at least two measurements with At = 0. The dashed line represents the kinetics of endocytosis after release of 5% of the vesicle pool (20 AP stimulus), whereas the solid line depicts that after 25% release (100 AP stimulus). Each data point was averaged from measurements of 20-100 individual boutons in seven experiments. Because the data are normalized to the At = O s condition, we have only drawn a line through the points for subsequent times, starting at the earliest measured delay of At = 5 s. (C) The fractional completion of endocytosis at At = 5 s for 20 AP and 100 AP stimuli was compared at the same individual boutons. Note that endocytosis with the 100 AP stimulus reaches greater fractional completion at the end of 5 s, even though the last 80 of the 100 action potentials are fired later in the 5-s interval than the entire 20 AP stimulus.

endocytosis measurable by our method has the slow time beginning of the action potential train (At = 0 s). The relative course, which has been associated with the complete-fusion impact of differing extracellular Ca2+ concentrations was model of vesicle membrane reuptake (half-times on the order determined by comparing this load with that occurring for dye of 20 s at 24°C), and that endocytosis proceeds at roughly application at At = 3 s in one of three different solutions: similar rates for the brief and prolonged stimulus conditions. normal saline (2 mM Ca2+), Ca2+-free saline (50 ,uM EGTA), At 33°C, we found that at 10 s after the stimulus, endocytosis or high-Ca2+ saline (10 mM Ca2+). The data depicted in Fig. was 49% (±2% SE, n = 105) complete, indicating a Qlo (rate factor for of -2 for this process. Sjl.0 10°C increase) Io A modest acceleration of the endocytic process with the mM CaCJ mM ca ua stronger stimulus is apparent at the earliest time points 40.8 depicted in Fig. 2B. Endocytosis reached 36% (±1.5% SE, n = 104) completion for the 100 AP stimulus but only 15% I0.6 (±3% SE, n = 77) completion for the 20 AP. This effect is highly reproducible and was even more marked when extra- 0.4 neous variability was minimized by pairing measurements of the earliest time point (At = 5 s) for the 20 and 100 AP stimuli U. on the same individual boutons. Such direct comparisons of paired measurements for the At = 5 s time point of the 20 and 100 AP stimuli are summarized in Fig. 2C. FIG. 3. Dependence of endocytosis upon Ca2+. Measurements of In previous studies (14), we found that after large stimuli the amount of endocytosis occurring in either Ca2+-free (50 ,uM (KCl-induced membrane potential depolarizations), the en- EGTA), 2 mM CaC12, or 10 mM CaCl2 solutions during the 60-s period docytic phase of vesicle recycling was independent of extra- after a 20 AP train (At = 3 s) are presented, normalized with respect cellular Ca2+. To determine the role extracellular Ca21 might to the total amount of endocytosis elicited by 20 AP (At = 0 s) in play in endocytosis under more physiologic stimuli, we have normal saline (2 mM CaCl). These data indicate that endocytosis measured the amount of endocytosis following brief trains of during this period is insensitive to external Ca2+ over a 107-fold range action potentials, using a variety of extracellular Ca2+ con- in concentration. Endocytosis with Na+ replaced with Li+ is shown in the last bar; measurements indicate the amount of endocytosis occur- centrations. The protocol used was similar to that depicted in ring in the 60 s after exocytosis (At = 3 s), for Na+-free LiCl saline, Fig. 2A. Exocytosis was stimulated by a 20 AP train at 10 Hz normalized to the total amount of endocytosis elicited by 20 AP (At in normal saline (2 mM Ca2+). The total amount of endocy- = 0 s) in the same solution. (Bars = SEM; from left to right, n = 131, tosis (full load) was measured by applying FM 1-43 at the 112, 131, and 112.) Downloaded by guest on September 26, 2021 5570 Neurobiology: Ryan et al. Proc. Natl. Acad. Sci. USA 93 (1996) 3 are presented as the amount of endocytosis that remains 0 after At = 3 s in each solution, normalized to the 2 mM Ca2+ 0 0 saline condition measured from At = Os. These results indicate FM 2-10 that endocytosis appears insensitive to changes in extracellular 0 0 o Ca2+ over a wide range of concentrations (_107-fold) during o c; FM 1-43 the period in which 70-80% of endocytosis is taking place. We o 00 C verified the effectiveness of the solution change in rapidly FM 1-84 (1-2 s) changing the extracellular Ca2+ concentration by a 6 measuring changes in [Ca2+]i using the fluorescent Ca2+ indicator E0 o fluo-3 AM during trains of action potential (data not shown). z We also performed measurements of endocytic uptake for FM 2-10 FM 1-43 FM 1-84 these light stimuli under conditions in which [Ca2+]i remains t=.7s -r=2.7s 'r=5.9s elevated after a train of action potentials. Replacement of A NaCl with LiCl effectively blocks the Na+/Ca2+ exchanger, which normally functions to reduce [Ca2+]i after the action- 0 4 8 12 16 20 24 timalrotW- potential-driven elevations in [Ca2+]1 (16). This procedure 0 Ll ||VjUVL; leads to a prolonged elevation in [Ca2+]i after a train of action 0 (0 potentials compared with that in normal saline (16). We O) F)L 1 Hz measured the fraction of endocytosis taking place in the 60 s 00 after exocytosis (At = 3 s; Fig. 2A) for stimulation and C) endocytosis in either normal saline or Na+-free LiCl saline 0 0 (120 mM LiCl/2 mM CaCl2). The results are depicted in Fig. LO 3 and indicate that endocytosis is not affected by conditions 2-10 (CH2CH,) *.@ that inhibit Na+/Ca2+ exchange. 1-84 ((CH2)4CH3) ooo .20 We to determine whether the dissociation rate of the to sought -90 dye from the membrane would influence the efficiency of 0o destaining of dye-labeled vesicles during recycling when stim- ulated at frequencies as low as 1 Hz. Analogues of FM 1-43 0 i with both shorter (FM 2-10) (17) and longer (FM 1-84) B hydrocarbon chains are available, and their structures are 0 200 400 600 800 1000 depicted in Fig. 4A (Inset). We surmised that these two time (sec) analogues might dissociate from membranes faster and slower, respectively, than FM 1-43. This expectation was confirmed by FIG. 4. (A) (Inset) Chemical structure of different variants of the direct measurements of dye release from exposed surface amphipathic endocytic tracers FM 2-10, FM 1-43, and FM 1-84, which membrane in a rapid flow experiment, as illustrated in Fig. 4A. differ only by the length of the nonpolar hydrocarbon chain by which Ifvesicle membrane recapture occurs via a fast kiss-and-run they insert onto the membrane: FM 2-10, two carbons; FM 1-43, four efficient carbons; and FM 1-84, five carbons. (A) Measurements of the disso- process, two different mechanisms might prevent ciation times for different amphipathic styryl membrane probes. The destaining of the vesicle. If the fusion machinery consists of a dissociation times for different probes were determined by measuring physical barrier, such as a proteinaceous channel (18), that the decay of fluorescence after a brief dye-application pulse (30 ms). would prevent membrane components from mixing by lateral Dyes were applied through the tip of a micropipette to the surface of diffusion and if the recapturing times scale is similar to or much hippocampal neurons mounted in a fast perfusion chamber (flow faster than the dissociation time of the slowest dye, one would velocity = 60 mm/s, chamber volume = 75 ,ul). The fluorescence decay expect dye dissociation rates to influence strongly the degree curves were well fit in each case by a single exponential. The decay of vesicle destaining. Fig. 4B shows a comparison of the times indicated are representative of many experiments (n = 5) and kinetics of release by 1-Hz stimulation of FM 2-10 and FM verify the more rapid dissociation of the more water-soluble amphi- same boutons. The very close pathic probes. (B) A comparison of the destaining rates of individual 1-84, measured at the synaptic synaptic boutons labeled with either FM 2-10 or FM 1-84, stimulated agreement between these two curves, despite the large differ- at 1 Hz. Boutons were loaded with one of the probes by bath ence in dye dissociation times (700 ms and 6 s), suggests either application during 60 s of 10 Hz action potential firing and rinsed for (i) that most vesicles remain open for times substantially 5-10 min. The loss of fluorescence stimulated by a 1-Hz train of action greater than 6 s or (ii) that dye escapes via lateral diffusion into potentials was monitored by time-lapse imaging at 10-s intervals. After the plasma membrane and so need not first dissociate from 15 min of stimulation and 5 min of rest, the boutons were stimulated membrane into aqueous solution. at 10 Hz for 100 s to measure the total remaining fluorescence. The initial slope indicates that each action potential releases <0.3% of the recycling pool. The boutons in question were then similarly labeled and DISCUSSION destained using the other probe. The data shown represent ensemble averages of 20 boutons. The individual fluorescence traces obtained In this report, we have extended previously described methods during 1-Hz stimulation were normalized to the sum of the fluores- (15) to obtain measurements of synaptic vesicle recycling after cence loss occurring in the 1- and 10-Hz runs. These normalized very brief trains of action potentials (Fig. 2). This has allowed individual traces were then averaged together. us to determine the kinetics of endocytosis using stimulation levels an order of magnitude lower than those previously reported, in the range of stimulus trains (20 action potentials increased vesicular load as well as the greater residual eleva- in <1 s) typically measured in vivo (19). These measurements tion in [Ca24]1, which must follow the heavier stimulus. Inhi- have revealed that even after as little as 5% of the recycling bition of the Na4/Ca2+ exchanger, however, does not appear vesicle pool (- 10 vesicles per bouton) is caused to fuse with the to shift the time scale of endocytosis (Fig. 3B), even though it membrane, endocytosis is relatively slow, delayed in relation to leads to still greater increases in residual [Ca2+]1 (16). exocytosis with a t½/2 of 20 s. When exocytosis is increased The possible regulatory influences of Ca2+ ions on synaptic 5-fold with an additional 80 action potentials, the bulk of vesicle endocytosis are of considerable interest. Unfortunately, endocytosis proceeds with a similar slow (Q9/2 = -20 s) time the strong Ca24 dependence of exocytosis, and the necessity course, although a modest acceleration is apparent during the for exocytosis before endocytosis, makes any Ca2+ effects on first 5 s. Possible explanations of this acceleration include the endocytosis difficult to separate experimentally from effects Downloaded by guest on September 26, 2021 Neurobiology: Ryan et al. Proc. Natl. Acad. Sci. USA 93 (1996) 5571 on exocytosis. Our results clearly demonstrate, however, that Here we have demonstrated that all vesicles accessible to FM extracellular Ca2+ is not required for endocytosis. Indeed, 1-43 during recycling are recaptured by a slow endocytic there is no evidence for any influence of extracellular Ca>2 on process, which takes many seconds even at low stimulation endocytosis beyond that to be expected from the known role levels. This slow time scale of endocytosis is more consistent of Ca2 in triggering exocytosis. A similar conclusion was with the ultrastructural observations that suggested the com- reached in studies at the neuromuscular synapse of the Dro- plete-fusion model than it is with those that suggested the sophila shibire mutants (20). On the other hand, a different kiss-and-run model. At present, it is unknown if vesicles that conclusion was suggested by earlier results from frog neuro- might undergo very rapid endocytosis could escape labeling by muscular junction, where endocytosis was shown to require FM 1-43. During the staining procedure, the dye is present extracellular Ca>2 in a-latrotoxin-stimulated terminals (21). both in solution and bound to the plasma membrane. Although More recent studies, however, have suggested that this may the dissociation of these probes occurs on the seconds time have reflected a Ca2+ gating effect on a-latrotoxin channels scale, the association time scale is faster than can be resolved (22). Our results also do not support the notion that [Ca2+], with these methods (much less than 1 s; data not shown), plays a strong modulatory role during endocytosis. The timing suggesting that even rapidly retrieved vesicles are likely to of the endocytic process appears to be relatively insensitive to become labeled. Nonetheless, there remains the possibility either inhibition of the Na+/Ca2+ exchanger, continuous that another population of vesicles, wholly separate and inac- electrical activity (Fig. 2), or stimulation in higher Ca2+ cessible to FM 1-43 staining, may recycle via the faster concentration (10 mM). The fact that these varied conditions endocytosis process. It will be important to determine whether all have significant impact on [Ca2+]i and on rates of exocy- or not any such parallel kiss-and-run vesicle population actu- tosis, without significant change in endocytic rates, suggests ally exists, and, if it exists, to determine the relative contribu- that endocytosis in these nerve terminals is not subject to tions of fast and slow endocytosis vesicle populations to significant modulation by [Ca]2+]. These results differ from physiological synaptic transmission. those obtained from ribbon synapses of the retina (7, 8). In bipolar cell synapses, brief stimuli led to rapid recovery of The authors wish to thank Dr. Noam Ziv for helpful discussion membrane (1 s), whereas more prolonged stimuli resulted in a during the course of this work. This work was supported by grants from slower endocytic rate (30 s). This slowing of endocytosis was the National Institutes of Health (S.J S.), the Swiss National Science shown to be the result of prolonged elevation in [Ca2+]i. This Foundation (H.R.), and the Stanford Medical Free Electron Laser discrepancy may indicate that endocytosis regulates differently in Program, funded by the Office of Naval Research. tonic synapses than it does in the phasic synapse we have studied. When synaptic vesicles cycle through the plasma membrane 1. Heuser, J. E. & Reese, T. S. (1973) J. Cell Biol. 57, 315-344. in the presence of FM 1-43, the dye becomes trapped within 2. Heuser, J. E., Reese, T. S., Dennis, M. J., Jan, Y. & Evans, L. the lumen of the vesicle. Because the dye is present both in (1979) J. Cell Bio. 81, 275-300. solution and bound to the plasma membrane, the labeling can 3. Miller, T. M. & Heuser, J. E. (1984) J. Cell Biol. 98, 685-698. occur via two different routes: bulk 4. Ceccarelli, B., Hurlbut, W. P. & Mauro, A. (1973) J. Cell Biol. 57, potentially by diffusion 499-524. into the lumen from solution or by lateral diffusion from the 5. Fesce, R., Grohovaz, F., Valtorta, F. & Meldolesi, J. (1994) plasma membrane. Both mechanisms should be quite fast (<1 Trends Cell Biol. 4, 1-4. ms). When exocytosis reopens the lumen to the extracellular 6. Meldolesi, J. & Ceccarelli, B. (1991) Philos. Trans. R. Soc. environment, destaining can occur by either or both of these London B 296, 55-65. two mechanisms. In this case, however, although lateral dif- 7. von Gersdorff, H. & Matthews, G. (1994) Nature (London) 367, fusion will still be fast, direct escape out of the lumen into 735-739. solution will be limited by the rate at which the dye dissociates 8. von Gersdorff, H. & Matthews, G. (1994) Nature (London) 370, from the vesicle membrane. In intact retrieval or kiss-and-run 652-655. exocytosis models, it is often postulated that a proteinaceous 9. Alvarez de Toledo, G., Femandez-Chacon, R. & Fernandez, J. pore structure allows neurotransmitter efflux while partition- (1993) Nature (London) 363, 554-558. ing vesicle membrane from the neurolemma. Such a protein- 10. Thomas, P., Lee, A. K., Wong, J. G. & Almers, W. J. (1994)J. Cell aceous structure well act as a to Biol. 124, 667-675. pore might barrier lateral 11. Harris, K. M. & Stevens, J. K. (1989) J. Neurosci. 9, 2982-2997. diffusion to a dye like FM1-43. The finding that dyes with 12. Betz, W. J. & Bewick, G. S. (1992) Science 255, 200-203. dissociation times as slow as 6 s (FM 1-84) are released with 13. Betz, W. J. & Bewick, G. S. (1993) J. Physiol. (London) 460, equivalent efficiency strongly suggests either that vesicle mem- 287-309. brane remains fully exposed for some time well in excess of 6 14. Ryan, T. A., Reuter, H., Wendland, B., Schweizer, F. E., Tsien, s before exocytosis or that any more rapidly opened fusion pore R. W. & Smith, S. J (1993) Neuron 11, 713-724. allows the near complete loss of dye by lateral diffusion within 15. Ryan, T. A. & Smith, S. J (1995) Neuron 14, 983-989. its shorter open time. 16. Reuter, H. & Porzig, H. (1995) Neuron 15, 1077-1084. Previous studies have shown that action-potential-driven 17. Ribchester, R. R., Mao, F. & Betz, W. J. (1994) Proc. R. Soc. fluorescence destaining of terminals loaded with FM 1-43 London B 255, 61-66. faithfully track the kinetics of neurotransmitter release (13) 18. Almers, W. & Tse, E. W. (1992) Neuron 4, 813-818. and the kinetics of dye loading (15). Furthermore, the frac- 19. Ylinen, A., Bragun, A., Nadasdy, Z., Jando, G., Szabo, I., Sik, A. tional & Buzsaki, G. (1995) J. Neurosci. 15, 30-46. release of dye per action potential is consistent with 20. Ramaswamy, M., Krishnan, K. S. & Kelly, R. B. (1994) Neuron estimates of release probabilities and vesicle numbers (15). 13, 363-375. These data suggest that the recycling vesicles stained by FM 21. Ceccarelli, B. & Hurlbut, W. P. (1980) J. Cell Biol. 87, 297-303. 1-43 could account for all regulated releases of synaptic 22. Hurlbut, W. P., Chieregatti, E., Valtorta, F. & Haimann, C. neurotransmitter. (1994) J. Membr. Biol. 138, 91-102. Downloaded by guest on September 26, 2021