Abstract
Membrane-bound organelles such as mitochondria and the endoplasmic reticulum play an important role in neuronal Ca2+ homeostasis. Synaptic vesicles (SVs), the organelles responsible for exocytosis of neurotransmitters, occupy more of the volume of presynaptic nerve terminals than any other organelle and, under some conditions, can accumulate Ca2+. They are also closely associated with voltage-gated Ca2+ channels (VGCCs) that trigger transmitter release by admitting Ca2+ into the nerve terminal in response to action potentials (APs). We tested the hypothesis that SVs can modulate Ca2+ signals in the presynaptic terminal. This has been a difficult question to address because neither pharmacological nor genetic approaches to block Ca2+ permeation of the SV membrane have been available. To investigate the possible role of SVs in Ca2+ regulation, we used imaging techniques to compare Ca2+ dynamics in motor nerve terminals before and after depletion of SVs. We used the temperature-sensitive Drosophila dynamin mutant shibire, in which SVs can be eliminated by stimulation. There was no difference in the amplitude or time course of Ca2+ responses during high-frequency trains of APs, or single APs, in individual presynaptic boutons before and after depletion of SVs. SVs have a limited role, if any, in the rapid sequestration of Ca2+ within the neuronal cytosol or the synaptic microdomain. We also conclude that SVs are not important for regulation of synaptic VGCCs.
- calcium regulation
- calcium channels
- synaptic vesicles
- Drosophila
- temperature-sensitive mutant
- shibire
Introduction
In presynaptic nerve terminals, synaptic vesicles (SVs) occupy a larger volume than any other intracellular organelle. Apart from their principal role in sequestering and releasing transmitter substances, data from in vitro studies indicate that SVs may also participate in Ca2+ sequestration (Israel et al., 1980; Michaelson et al., 1980; Rephaeli and Parsons, 1982; Gonçalves et al., 1998). In addition, voltage-gated Ca2+ channels (VGCCs), which respond to depolarization by admitting Ca2+ close to sites of exocytosis in the synaptic active zone (Catterall, 2000), may be regulated through interactions with SVs. If SVs are significantly involved in such activities, they should modify the amplitude and kinetics of changes in the presynaptic free Ca2+ concentration ([Ca2+]i) after action potentials (APs). To test this hypothesis, we compared Ca2+ transients in motor nerve terminals depleted of SVs with Ca2+ transients in nondepleted terminals.
To deplete nerve terminals of SVs, we used the Drosophila temperature-sensitive dynamin mutant shibire (Chen et al., 1991; van der Bliek and Meyerowitz, 1991), in which SVs do not recycle at temperatures above 29°C (the “nonpermissive” temperature) (Koenig and Ikeda, 1983; Koenig and Ikeda, 1989). Previous work on this mutant concluded that nerve-evoked Ca2+ signals disappear entirely from shibire nerve terminals depleted of SVs at nonpermissive temperatures (Umbach et al., 1998). This was taken to indicate that SVs must be present for VGCCs to operate properly. The close association between VGCCs and SVs observed in detailed electron microscope tomographic reconstructions of the amphibian neuromuscular junction (NMJ) (Harlow et al., 2001) lends support to this view. Alternatively, this association may ensure rapid transmitter release independently of any modulation of the VGCCs by SVs.
The evidence that SVs sequester Ca2+ in situ is unclear. Although there have been numerous proposals of a role for SVs in presynaptic Ca2+ regulation, these are based on data from in vitro studies using cytochemical, electrophysiological, and biochemical procedures. For instance, in thin sections of neuronal tissue, electron spectroscopic imaging shows a higher concentration of calcium in the lumen of SVs than in the cytosol (Torpedo nerve electroplaque synapse, Parducz and Dunant, 1993; rat brain, Mizuhira and Hasegawa, 1997; frog NMJ, Pezzati and Grohovaz, 1999). Other studies show that there is an increase in the proportion of Ca2+-containing SVs after nerve stimulation (rat superior cervical ganglion, Parducz et al., 1987; Torpedo, Parducz and Dunant, 1993, Parducz et al., 1994). Ion channels in the SV membranes may allow Ca2+ entry (for review, see Rahamimoff et al., 1990; Woodbury, 1995). A Ca2+-H+ exchange has been demonstrated in SV membranes in which elevated Ca2+ (∼500 μm) induced H+ release from SVs isolated from sheep brain cortex (Gonçalves et al., 1998). In addition, a high-affinity ATP-dependent Ca2+ pump takes up Ca2+ in isolated SVs (Torpedo electric organ synapses, Israel et al., 1980; Michaelson et al., 1980; Rephaeli and Parsons, 1982). Thus, although Ca2+ can be found concentrated in SVs, and SVs accumulate Ca2+ under some circumstances, it remains to be demonstrated that SVs can influence [Ca2+]i regulation in an intact nerve terminal. If SVs sequester Ca2+ in Drosophila nerve terminals under conditions of elevated [Ca2+]i, their absence should alter the amplitude and time course of Ca2+ signals during stimulation.
The experiments reported here were designed to test the need for SVs in Ca2+ entry and regulation during stimulation. By completely depleting SVs in shibire nerve terminals, we were able to test two previous hypotheses: (1) SVs are required for VGCC function; and (2) SVs sequester Ca2+ during and after nerve impulses, thereby modifying the amplitude and/or time course of the changes in [Ca2+]i. We found that elimination of intraterminal SVs by stimulating shibire motor nerves at nonpermissive temperatures had little impact on the intracellular Ca2+ signal.
Materials and Methods
Fly stocks. All experiments were performed on wandering third instar larvae of Drosophila melanogaster. The Drosophila mutant shibire [temperature-sensitive allele 1 (shits1), referred to henceforth as shi] was originally generated by ethyl methanesulfonate mutagenesis of Oregon Red (OR) flies (Grigliatti et al., 1973). OR flies were used here as the wild-type control. Canton Special (CS) and shi strains were gifts from Dr. K. E. Zinsmaier (University of Arizona, Tucson, AZ), and the OR strain was a gift from Dr. M. L. Suster (McGill University, Montreal, Quebec, Canada). All fly stocks were kept in noncrowded conditions at 22 ± 1°C on cornmeal agar with dry yeast.
Larval movement assay. A custom-built apparatus was used to monitor the movement of shi, OR, and CS larvae during temperature cycles between 22 and 34°C (see Fig. 1). Five third instar larvae were placed on a flat agarose gel substrate in a circular chamber (60 mm in diameter, 3 mm deep) formed on the upper surface of a fan-cooled Peltier device (Tellurex, Traverse City, MI). A transparent heater (Minco Products, Minneapolis, MN) closed the top of the chamber, facilitating rapid increases in temperature. To alter the temperature of the chamber from 22 to 34°C and back to 22°C, we controlled the current in Peltier device and heater by a temperature controller (model 5C7-362; Oven Industries, Mechanicsburg, PA) and feedback thermistor. Changes in the controller set point were made by a computer program via a serial port on the controller that also recorded the real temperature in a disk file. This system was designed to produce rapid and reproducible changes in temperature. Video images of larvae were obtained with a standard monochrome CCD camera (MK-1312E; Marscam, Taipei, Taiwan). A record of the larval positions was captured at 5 sec intervals on a personal computer through a frame-grabber board (DT3155′ Data Translation, Marlboro, MA) supported by AIW 4.0 software (Axon Instruments, Union City, CA). A pixel-by-pixel subtraction of consecutive images using ImageJ software (http://rsb.info.nih.gov/ij/) provided a measure of relative movement calculated for 5 sec intervals over the entire 1 hr period. Four trials were averaged for each strain to produce records of larval movement over the temperature cycle (see Fig. 1).
The larval NMJ preparation and temperature control. Larvae were dissected in chilled Schneider's insect medium (Schneider's) (Sigma, St. Louis, MO) to make a filleted preparation pinned to the Sylgard (Dow Corning, Midland, MI) base of a 0.5 ml perfusion bath. All nerves issuing from the ventral ganglion were severed. Before Ca2+ imaging, Schneider's was replaced with physiological solution [Hemolymph-Like number 6 (HL6)] (Macleod et al., 2002) that continuously superfused the preparation, except in fura-dextran (fura) imaging experiments in which HL6 was renewed at least every 30 min. [Ca2+]o and [Mg2+]o in HL6 were 0.5 and 15 mm, respectively, or, when specifically indicated, 2 and 4 mm, respectively. The temperature of the preparation was controlled by the flow of HL6 superfusate (3 ml/min) that was heated over a Peltier device (Melcor, Trenton, NJ), connected to an adjustable 6 V DC power supply. To obtain adequate temperature control, the thermally conductive aluminum nose cone of the water-immersion objective (40×, 0.55 numerical aperture, 160 mm; Nikon, Tokyo, Japan) was replaced with a custom-made Delrin nose cone that is much less thermally conductive than the standard aluminum nose cone. Two thermocouples, 2 mm apart on either side of a larva, were used to monitor the bath temperature.
Loading calcium indicators into motor neuron terminals. Motor neurons (MNs) of the longitudinal abdominal muscles were forward filled with 10 kDa dextran conjugates of Oregon Green 488 BAPTA-1 (OGB-1) [Ca2+ dissociation constant (Kd), 454 nm] or fura (Kd, 594 nm, measured in 100 mm KCl and 30 mm MOPS) (Molecular Probes, Eugene, OR) dissolved in water, using the method described by Macleod et al. (2002). Dextran-conjugated Ca2+ indicators were used because they are not taken up by intracellular organelles. OGB-1 10 kDa dextran was used because it is visible at resting [Ca2+]I and has a Kd well matched to the [Ca2+]i range we wanted to observe and a good dynamic range. It also appears to have little impact on the time course of the Ca2+ signal decay after short trains of pulses at 20 Hz, or single APs, in Drosophila motor neuron terminals. This was ascertained by comparing Ca2+ signals in the presence of OGB-1 and the low-affinity indicator Fluo-4 10 kDa dextran (Macleod et al., 2002). To load indicators, a hemisegment nerve was cut close to the ventral ganglion and then drawn by suction into the end of a heat-polished glass pipette (∼12 μm diameter) to form a fit just tight enough to exclude further entry of Scheider's from the preparation bath. A fine plastic tube was introduced into the suction pipette to apply the indicator, at a final concentration of 0.5-2 mm in Schneider's, to the cut end of the nerve; this was done within 5 min after cutting the nerve. After 20-40 min, the fine plastic tube was used to flush the indicator from the tip of the pipette with fresh Schneider's. The suction pipette, now filled with Schneider's, was left in place as the anode for stimulating the nerve.
Calcium imaging with Oregon Green 488 BAPTA-1. Imaging of OGB-1-loaded MN terminals was performed on a Nikon microscope (Optiphot-2) fitted with a Bio-Rad-600 confocal scan head (Bio-Rad, Mississauga, Ontario, Canada). Only type-1b boutons on muscles 6 and 7 were examined. To reduce phototoxicity, the argon ion laser was operated at low power, and the output was attenuated to 0.5% by neutral density filters. A light-emitting diode inserted into the optical path via a side inspection port of the scan head was lit for 2 msec to mark the first pulse in stimulation trains. Scanned images were saved in eight-bit PIC format and then converted to TIFF format for analysis using ImageJ software. Both frame scans (for whole-bouton measurements) and line scans (for higher temporal resolution of Ca2+ changes within a single bouton) were collected as described by Macleod et al. (2002). The principle for analyzing frame scan data were the same as for analyzing line scan data. Average pixel values were measured within selected regions of each image, or each line, and also in an adjacent region not containing any OGB-1 fluorescence to provide a background value. This background value was subtracted numerically from the average pixel value of the fluorescence-containing region to generate the fluorescence value (F). Images of fluorescence are shown without the background subtracted. ΔF/FR is calculated by dividing the change in F in response to stimulation (ΔF) by the value of F at rest (FR) before nerve stimulation (see Figs. 3D, 4B, 5, 6, 7C, 8). ΔF/FS is calculated by dividing the change in F in response to the cessation of stimulation (ΔF) by the value of F during nerve stimulation (FS) (see Fig. 3B). All estimates of the time course of decay were made using Sigma Plot (SPSS, Chicago, IL) by fitting a single-exponential curve to line scan data.
Calcium imaging with fura-dextran. Imaging of fura-dextran-loaded nerve terminals was performed on a Nikon (Optiphot-2) microscope with an Olympus Optical (Tokyo, Japan) water-immersion objective (40×, 0.7 numerical aperture). The microscope was fitted with an intensified CCD camera (model IC-100; PTI, Princeton NJ) and a filter wheel (model 5240; Pacific Scientific, Metaltek Instruments, Raleigh, NC) controlled by an Axon Instruments (Foster City, CA) Digidata 2000 frame grabber supported by AIW 2.2 software. Images were acquired through a 530 ± 35 nm bandpass filter as a mercury arc lamp was used to illuminate the preparation alternately through 340 ± 5 and 380 ± 5 nm bandpass filters (Omega Optical, Brattlebro, VT). Equation 5 by Grynkiewicz et al. (1985) was used to calculate [Ca2+]i values. Values of Rmin and Rmax were obtained in situ through bath application of 1 mm ionomycin (Calbiochem, La Jolla, CA) in HL6 containing either no added Ca2+ and 10 mm EGTA (Rmin; Sigma) or 10 mm [Ca2+]o (Rmax) (both solutions were adjusted to pH 7.2 using NaOH). The Ca2+ dissociation constant has not been determined for fura-dextran in Drosophila MNs in situ. The value determined in situ for fura-2 in crayfish MNs (Kd, 865 nm) (Tank et al., 1995) has been used in this study.
The use of l-glutamic acid to reduce muscle contraction. High-frequency stimulation of MNs at physiological levels of [Ca2+]o results in significant Ca2+ influx, considerable neurotransmitter release, and corresponding muscle contraction, which interferes with Ca2+ imaging. To overcome the problem of muscle contraction at MN terminals not depleted of SVs, we added l-glutamic acid (LGA) to the HL6 superfusate. We found that glutamate receptors become sufficiently desensitized, at millimolar concentrations of LGA (≥5 mm), to allow 80 Hz stimulation of the nerve without muscle contraction (0.5 or 2 mm [Ca2+]o). Stimulation trains of 80 pulses at 80 Hz were chosen to test the Ca2+ handling properties of MN terminals during activity because this paradigm resembles MN activity during fictive locomotion (Cattaert and Birman, 2001; Barclay et al., 2002). We examined the amplitude of OGB-1 fluorescence transients evoked during 10 Hz stimulation in shi MN terminals in the presence of 7 mm LGA and found no difference (paired t test; p = 0.78) relative to the amplitude of transients in its absence (ΔF/FR of 28.2 ± 8.5% with and 29.3 ± 6.3% without; n = 5). The relationship between the frequency of stimulation and ΔF/FR in the presence of 7 mm LGA was linear in the 10-80 Hz range (data not shown), and 80 Hz stimulation did not result in dye saturation because 120 and 160 Hz stimulation always resulted in significantly greater changes in OGB-1 fluorescence, often with muscle contraction (data not shown).
Protocol to deplete MN terminals of SVs. Larval preparations intended for depletion treatment were heated by superfusing HL6 with increased [Ca2+]o (2 mm) and reduced [Mg2+]o (4 mm) to maximize transmitter release during stimulation (see Fig. 4A). After 3 min at 34 ± 1°C, the segmental nerve was stimulated at 30 Hz for 6 min (10,800 pulses). Periodic inspection of the OGB-1 fluorescence level within the boutons indicated that APs were being continually delivered to type-1b boutons throughout the impulse train. If the stimulus voltage was reduced below a certain threshold at any time during the train, the fluorescence level dropped immediately. Thus, depletion-inducing stimulation was effective in activating boutons throughout the entire period of stimulation.
Electron microscopic examination of terminal boutons, subsequent to stimulation, was used to verify complete depletion of SVs. Electrophysiological methods were not used to attempt verification, because the signs of SV depletion [the inability to evoke excitatory junction potentials (EJPs) and the inability to detect miniature EJPs (mEJPs)] can occur without SV depletion. The inability to evoke EJPs is not proof of SV depletion because one or both motor axons innervating muscle 6 or 7 can become inexcitable (especially at high temperatures) before SV depletion occurs. Although mEJPs may go undetected for a period, a combined electron microscopical and electrophysiological study has demonstrated previously that numerous SVs can occupy the active zones of shibire terminal boutons while mEJPs are very infrequent (Koenig and Ikeda, 1999).
Electron microscopy. Filleted larval preparations were fixed within minutes of depletion stimulation in a preheated mixture (34°C) of 3% glutaraldehyde, 1% formaldehyde, and 0.1 m sodium cacodylate buffer at pH 7.2 for 4 hr. After fixation, preparations were rinsed in the same buffer for 2 hr at 4°C and then trimmed to allow unambiguous identification of the stimulated segment. Preparations were postfixed in 1% osmium tetroxide for 1 hr, followed by tissue dehydration in a graded ethanol series (30-100%, 20 min) and propylene oxide. The preparation was infiltrated and embedded with Epon-Araldite for 2 d at 60°C. Sections of 75 nm were cut on a Reichert-Jung ultracut microtome, stained with uranyl acetate and lead citrate, examined in a Hitachi (Tokyo, Japan) H-7000 transmission electron microscope, and photographed with an AMT Advantage HR/HR-B CCD camera system (Advanced Microscopy Techniques, Danvers, MA).
Results
If SVs are substantially involved in the regulation of Ca2+ channels or the sequestration of Ca2+, stimulus-dependent Ca2+ signals would be altered in the absence of SVs. To test this prediction, we developed a protocol to deplete MN terminals of SVs in shi larvae without permanently compromising cellular physiology.
shi larvae are paralyzed at 34°C but recover rapidly when cooled
We first tested shi larvae for temperature-dependent paralysis and recovery. Delgado et al. (2000) established that larval MN terminals can be depleted of SVs by stimulating the nerve while holding the preparation at 34°C. We wanted to hold MN terminals at this nonpermissive temperature long enough to deplete them of SVs and then collect physiologically valid data from them. However, previous studies had indicated that shi flies recover very slowly, if at all, from extended exposures to nonpermissive temperatures (see Fig. 3) (Siddiqi and Benzer, 1976). To establish that shi larvae are capable of recovery, we monitored the activity of larvae while the temperature was cycled from 22 to 34°C and back to 22°C (Fig. 1). The positions of shi, OR, and CS larvae were monitored at 5 sec intervals, and a measure of movement was calculated (see Materials and Methods). At room temperature (22°C), all strains of larvae showed gradual reduction in activity over a 1 hr test period (Fig. 1A). When the temperature was elevated to 34°C, both OR and CS strains became more active, whereas the shi strain stopped moving within 3 min (Fig. 1B). When, after 20 min at 34°C, the temperature was reduced to 22°C, shi larvae began moving again and regained a level of activity comparable with OR and CS strains within 5 min.
The resumption of movement by shi larvae shown in Figure 1 indicates that normal physiology is restored after extended exposure to 34°C. From these behavioral data, we cannot assume that the MN terminals are fully depleted coincident with paralysis, because paralysis may result from depletion of SVs at central synapses (Kosaka and Ikeda, 1983). However, in shi adults, some MN terminals become depleted of SVs if the nerves are not severed from the CNS at nonpermissive temperatures (Kawasaki et al., 2000). Thus, it is likely that the MN terminals in freely moving shi larvae, exposed to 34°C, recover from at least a partial depletion of SVs.
Evidence of synaptic vesicle depletion
Up to 4000 pulses are required to deplete shi larval MN terminals in an isolated preparation at 34°C (Delgado et al., 2000), whereas stimulation below 30 Hz is insufficient to deplete the reserve pool of SVs (Kuromi and Kidokoro, 2000). Because our MN terminals were loaded with OGB-1 and because similar Ca2+ chelating compounds can compete for Ca2+ with the Ca2+ sensor for transmitter release and thus inhibit release of SVs (Adler et al., 1991), we stimulated at a higher frequency than in previous studies; high-frequency stimulation can overcome the effects of modest concentrations of chelators (Winslow et al., 1994).
Filleted shi larval preparations were heated with superfusing HL6, and, after 3 min at 34 ± 1°C, the segmental nerve was stimulated at 30 Hz for 6 min (10,800 pulses); release was maximized by the inclusion of 2 mm [Ca2+]o and 4 mm [Mg2+]o in the HL6. To determine that they were depleted of SVs, we examined stimulated boutons by electron microscopy and compared them with unstimulated boutons in the same preparation. Within minutes of the end of the stimulus train and without allowing the temperature to drop, we replaced HL6 with a preheated mixture (34°C) of 3% gluteraldehyde and 1% formaldehyde. The preparation was then processed for thin-section (75nm) transmission electron microscopy (see Materials and Methods). We note that it is unlikely that the application of fixative alone causes significant exocytosis, because exocytosis subsides within 500 msec of fixative application at the adult coxal muscle NMJ (Koenig et al., 1998) and within 1 sec at rodent hippocampal synapses in culture (Rosenmund and Stevens, 1997).
Serial sections of type-1b boutons in unstimulated segments adjacent to the stimulated segment were filled with SVs (Fig. 2A). In contrast, the boutons in the stimulated segment were completely devoid of SVs (Fig. 2B,C). All 18 sections from the stimulated boutons were examined, and no SVs could be identified. Sac-like membranous inclusions of variable size were present, and these internalized sacs were often confluent with the extracellular space (Fig. 2D). Mitochondria in the stimulated boutons appeared normal, indicating that gross changes in [Ca2+]i or osmotic pressure had not occurred. Although no SVs were found in MN terminal boutons stimulated at 34°C, larger vesicular structures arrested in the process of pinching off from the surface were observed. Some had electron-dense “necks” at their point of contact with the surface plasma membrane (PM). This feature has been observed in previous electron microscopy of dynamin mutants (Kosaka and Ikeda, 1983, 1989).
Ca2+ accumulation in presynaptic boutons during and after depletion of SVs
To confirm that MN terminals were continuously stimulated during depletion, we measured OGB-1 fluorescence as an indicator of [Ca2+]i. Elevated OGB-1 fluorescence during the stimulus train and its decline on cessation of stimulation is a good indicator of effective bouton excitation during our depletion trains. Images of OGB-1-loaded boutons were acquired with a laser-scanning confocal microscope every 680 msec as the depletion stimulation train ended (Fig. 3A,B). OGB-1 fluorescence dropped immediately on cessation of stimulation between one image (Fig. 3A, third image) and the next (fourth image). The average fluorescence for five preparations (Fig. 3B, open circles) declined by ∼50% on cessation of stimulation. This drop indicates that [Ca2+]i had been elevated during the depletion train. Elevations in [Ca2+]i could also be evoked by short-stimulation trains after depletion stimulation (Fig. 3C,D). The average change in fluorescence increased linearly with stimulation frequency in the 10-40 Hz range (Fig. 3D, inset) (n = 3). This result is in good agreement with the single compartment model of Tank et al. (1995) in which Ca2+ reaches a plateau when entry and removal-sequestration are in balance. The boutons sectioned in Figure 2, B and C, are the same as those from which the records shown in Figure 3C were taken, 2 min before fixation.
Ca2+ accumulation is reduced in shi boutons at 34°C before SV depletion
OGB-1 fluorescence responses to trains of APs were compared between shi and wild-type MN terminals. At 22°C, there was no difference in the response to stimulation between strains (p ≫ 0.05) (Fig. 4). In shi, the ΔF/FR response was reduced at 34°C relative to 22°C. In CS, there was no decrease in the ΔF/FR response at 36°C relative to 22°C (data not plotted). However, in OR, from which shi was generated (Grigliatti et al., 1973), there was a reduction in ΔF/FR response between 22 and 34°C. There was a similar reduction in the ΔF/FR response in shi between 22 and 34°C; however, there was no difference between depleted and nondepleted shi MN terminals at 34°C (unpaired t test; p = 0.63; n = 6). Thus, increasing the temperature results in a decrease in the ΔF/FR response independent of depletion stimulation. The decrease in fluorescence signal is not a result of depletion; as shown in Figure 2A, boutons heated without stimulation did not become depleted. There was no difference between the first measurement of the ΔF/FR response in shi at 22°C and the measurement from the same MN terminals, given time to recover at 22°C, subsequent to depletion at 34°C (paired t test; p = 0.62; n = 5).
[Ca2+]i kinetics during short high-frequency stimulation trains are unchanged in the absence of SVs
To examine the effect of SV depletion on [Ca2+]i dynamics, we compared OGB-1 fluorescence with high temporal resolution in depleted and undepleted MN terminals. We collected these data by repeatedly scanning a single line through the center of type-1b boutons at a rate of 250 Hz (Macleod et al., 2002), while stimulating the nerve with a train of 80 pulses at 80 Hz (Fig. 5). In response to stimulation, there was an immediate elevation in OGB-1 fluorescence in nondepleted shi boutons at 34°C, followed by a plateau for several hundred milliseconds and a rapid decay on cessation (Fig. 5A). After shi MN terminals had been depleted using the protocol described previously, OGB-1 fluorescence also increased rapidly with a rapid decay on cessation (Fig. 5B). Depleted boutons are subjected to rigorous stimulation before gathering line-scan data, whereas nondepleted boutons are not. This treatment and the sustained Ca2+ load may compromise the ability of organelles (not SVs) to participate in normal Ca2+ handling during the short test trains that follow. We tested for the effect of the stimulation protocol alone, by subjecting MN terminals of OR larvae (Fig. 5C) to the depletion protocol at 34°C. OGB-1 fluorescence increased rapidly in response to 80 Hz stimulation as it did with nondepleted and depleted shi MN terminals and fell sharply with its cessation.
A plot of the average trace for each of the three treatments (Fig. 6A) indicates similarity in the Ca2+ dynamics between treatments. When the traces are normalized to the amplitude of the last 200 msec of stimulation (Fig. 6B), the similarity in the dynamics is more obvious. Three parameters of the OGB-1 ΔF/FR response were measured for each treatment shown in Figure 5: maximum ΔF/FR amplitude (Fig. 6C), rise time (Fig. 6D), and time constant of decay after the 80th pulse (Fig. 6E). When these parameters are compared, it is clear that the trace for shi-depleted boutons is little different from the trace for nondepleted shi boutons at 34°C, although both are similar to the traces for OR boutons subjected to the depletion protocol at 34°C.
Estimated [Ca2+]i during stimulation
We next asked whether volume-averaged [Ca2+]i reaches a level comparable with that observed for maximal Ca2+ uptake by SVs in vitro. The Km for ATP-dependent Ca2+ uptake by isolated cholinergic SVs has been estimated in the micromolar range (5 μm, Israel et al., 1980; 50 μm, Michaelson et al., 1980; 52 μm, Rephaeli and Parsons, 1982). Ca2+-H+ antiport activity in isolated SVs was optimally activated by 500 μm Ca2+ (Gonçalves et al., 1998). To obtain an estimate of [Ca2+]i during trains of high-frequency activity in the MN, such as those expected during locomotion, we used the ratiometric indicator fura (Fig. 7). The MN of fura-loaded terminals was stimulated at 80 Hz in the presence of either 0.5 or 2 mm [Ca2+]o, 15 mm [Mg2+]o, and 7 mm [LGA]. Fura fluorescence in OR and shi MN terminals indicated a [Ca2+]i resting level of ∼65 nm and an elevated level of [Ca2+]i during 80 Hz stimulation of 297 ± 30 nm in shi and 306 ± 31 nm in OR at 0.5 mm [Ca2+]o (Fig. 7E). [Ca2+]i rose to only 389 ± 34 nm in OR during 80 Hz stimulation in 2 mm [Ca2+]o (Fig. 7F). Thus, volume-averaged [Ca2+]i does not reach a level consistent with uptake observed in in vitro studies.
[Ca2+]i responses to single APs are unchanged in the absence of SVs
The large amplitude and long time course of [Ca2+]i changes in response to a high-frequency train of APs may overwhelm and mask more subtle differences in the influx and/or efflux of Ca2+ from the cytoplasm in the absence of SVs, which might indicate differences in Ca2+ influx and sequestration at the level of the microdomain. To investigate the possibility that VGCC opening probability and permeability is altered in the absence of SVs, or that less Ca2+ is intercepted and sequestered at the point of influx, we compared the amplitude and time course of volume-averaged [Ca2+]i in response to single APs (Fig. 8). If SVs are required for VGCCs to open in response to an AP, or to more generally relieve inhibition, we would expect a smaller Ca2+ response in the absence of SVs. Alternatively, if SVs play a role in sequestering Ca2+ on the periphery of the microdomains, their absence may result in a larger Ca2+ response.
The amplitude of the OGB-1 response to a single AP in depleted shi MN terminals at 34°C was unchanged (p = 0.34; unpaired t test) relative to the response in nondepleted MN terminals at 34°C (Fig. 8B). The time course of OGB-1 decay (extrusion-sequestration) after a single AP was also indistinguishable between treatments at 34°C (p = 0.89; unpaired t test) (Fig. 8C). The amplitude of the responses measured in shi at 34°C were not significantly lower that those at 22°C (Fig. 8B), but the time course of decay at 34°C was significantly faster than at 22°C for both depleted and nondepleted boutons (Fig. 8C). Faster rates of Ca2+ extrusion in the depleted and nondepleted MN terminals at 34°C, relative to 22°C, can account for the reduced amplitudes of the OGB-1 response to stimulation trains (Fig. 4) at the higher temperature (Tank et al., 1995).
Discussion
The present study used an endocytosis mutation of Drosophila to allow a comparison of AP-induced Ca2+ signals in nerve terminals containing SVs and those depleted of SVs. Calcium imaging data collected under the two contrasting conditions provided a test of the hypothesis that SVs directly regulate Ca2+ entry through VGCCs. The same data also allowed a test of the hypothesis that SVs contribute to neuronal Ca2+ homeostasis through rapid sequestration of Ca2+ during and shortly after APs.
Ca2+ influx during APs does not require SVs
In shi larval preparations maintained at 34°C, we observed an elevation in the level of fluorescence from the intracellular Ca2+ indicator in response to stimulation (30 Hz) of the segmental nerve. This fluorescence elevation was sustained for 6 min and fell immediately on cessation of stimulation (Fig. 3A,B). Subsequent examination of boutons subjected to this stimulation confirmed that all SVs were gone (Fig. 2). Because the fluorescence response was observed during the entire stimulus train and SVs were eliminated, we conclude that SV depletion did not affect Ca2+ entry and also that AP failure did not affect our SV depletion protocol. Subsequent short trains of stimuli at 34°C evoked fluorescence changes (Fig. 3C,D) indistinguishable in amplitude from those evoked before depletion at 34°C (Fig. 4). An analysis of fluorescence responses to single APs at 34°C, subsequent to depletion, showed no difference in the amplitude and time course of the Ca2+ signal relative to those before depletion at 34°C (Fig. 8). We conclude that regulation of VGCCs does not require SVs to bring presynaptic proteins into interaction with VGCC binding domains. These data do not preclude the possibility that either synaptotagmin and/or cysteine string proteins are regulators of VGCC activity and depolarization-invoked Ca2+ entry (Chen et al., 2002); however, if so, they are not required to be brought into organized interaction with the VGCC via an SV.
The data presented here are in contrast to another study that examined the Drosophila shi mutant (Umbach et al., 1998). Umbach and colleagues loaded the permeant AM form of the Ca2+ indicator Calcium Crimson into shi larval MN terminals. After incubating the preparation at 32°C for 10 min, the nerve was stimulated (10 Hz for 1-2 min) until muscle contraction ceased and was allowed to rest for an additional 5 min at 32°C. Subsequent stimulation at 10 Hz at 32°C caused no discernable increase in Calcium Crimson fluorescence. Although we note that 10 Hz stimulation for up to 2 min (1200 pulses) is unlikely to deplete the MN terminals of SVs (Delgado et al., 2000) and that muscle contraction ceases well in advance of SV depletion, we are unable to reconcile these data with our own. A possibility is that the loading procedure, requiring up to 3 hr incubation in a conventional saline solution (Jan and Jan, 1976) followed by preparation heating, may result in AP failure during repeated stimulation at 32°C in temperature-sensitive mutants such as shi.
A role for SVs in rapid presynaptic sequestration of Ca2+ could not be detected
Just as the amplitude of the Ca2+ response to stimulation in lizard MN terminals increases when sequestration of Ca2+ by mitochondria is blocked (David and Barrett, 2000) we predicted an increase in the Ca2+ response in Drosophila MN terminals depleted of SVs. However, in response to 80 Hz stimulation, MN terminals depleted of SVs showed no change in the rate of rise in [Ca2+]i, the amplitude of the plateau achieved after 1 sec, or the subsequent decay in [Ca2+]i relative to nondepleted MN terminal boutons (Fig. 6). In an attempt to detect subtle effects of Ca2+ sequestration by SVs, we examined [Ca2+]i responses to single APs in depleted boutons (Fig. 8). Numerous electron microscopy ultrastructural studies have shown that SVs densely crowd the active zone at the Drosophila NMJ, in which they present a significant surface area to intercept the incoming Ca2+ before it can diffuse throughout the cytosol (Atwood et al., 1993). In addition, some SVs are “docked” at the active zone beside PM-containing VGCCs; here, [Ca2+] in microdomains ([Ca2+]m), within 50 nm of the VGCC mouth, has been estimated to exceed several 100 μm [Llinas et al., 1992 (200-300 μm)] [but see Meinrenken et al., 2003 (10-30 μm)]. We predicted that, if SVs are no longer present to intercept a significant portion of Ca2+ influx through VGCCs, there would be a greater, or more rapid, increase in [Ca2+]i in response to a single AP. No such changes were observed, nor was there a change in time course of the decay of the Ca2+ signal.
In the analysis above, assumptions are made that the volume-averaged Ca2+ reported by the Ca2+ indicator reflects changes in sequestration by SVs at the level of the active zone and that the Ca2+ indicator is capable of revealing reductions in the time course of decay below 38 msec. Regarding the first assumption, numerical models indicate that all changes in microdomain Ca2+ will be proportionally reflected in the volume-averaged [Ca2+]i (Bennett et al., 2000). In the second major assumption, although OGB-1 has a high affinity for Ca2+, it is capable of reporting a rapid decay in cytosolic Ca2+. Decay time constants of ∼30 msec have been reported using OGB-1 at a concentration of 20 μm in dendrite spines of small volume (Sabatini et al., 2002). We have no estimate of cytosolic OGB-1 concentration in this study and cannot conclude that the absence of a difference in the time course of the Ca2+ signal, in response to single APs between depleted and undepleted boutons, constitutes a legitimate test of the sequestration role of vesicles at the microdomain. However, because the amplitude is unchanged between the depleted and undepleted preparations, we conclude that SVs are unlikely to have a role in Ca2+ sequestration at the level of the microdomain and are thus unlikely to affect transmitter release by this mechanism. Furthermore, it is unlikely that Ca2+ uptake by SVs alters the volume-averaged [Ca2+]i signal, which is important for signaling pathways such as CaMKII activation on SVs distal to the active zone.
Transmembrane ion gradients reverse across the SV membrane when SVs fuse with the plasma membrane
When SVs fuse with the PM, the SV membrane becomes continuous with the PM, with the lumenal side of the SV membrane now facing the extracellular solution. With prolonged stimulation of shibire boutons at 34°C, all of the SV membrane accumulates on the PM. The entire expanded membrane surface maintains contact with the extracellular solution until the temperature is reduced to permissive levels (Poskanzer et al., 2003). For this reason, the cytosol of a depleted bouton maintains contact with the same area of SV membrane as in a nondepleted bouton. In addition, the Ca2+-permeable macromolecules, originally located in the SV membranes, maintain their orientation relative to the cytosol. These Ca2+-permeable macromolecules, be they passive transporters or pumps, will not manifest the same activity in the PM as they did when located in discrete SVs. The H+ gradient is neutralized, and the Ca2+ gradient changes by as much as 4 log units. The trans-SV membrane proton and Ca2+ gradients, as well as the electrical potential, now favor movement of Ca2+ from the original lumenal side of the SV membrane (now facing the extracellular medium) into the cytosol. This would reverse the hypothetical Ca2+ transporting activity of SVs in their role as organelles capable of Ca2+ sequestration from the cytosol. Thus, depletion of SVs would be expected to eliminate any Ca2+ sequestration that SVs normally effect at nondepleted synapses. Our measurements indicate that SVs normally have a negligible role in Ca2+ sequestration.
Ca2+ uptake by SVs may be endocytotic
If SVs are not substantially involved in regulating [Ca2+]i in MN terminals during sustained stimulation, why is Ca2+ found concentrated in SVs in various nerve cells? One possibility is that Ca2+ in SVs is simply a consequence of endocytosis of an SV open to the extracellular milieu from which Ca2+ cannot be excluded. During endocytosis, synaptic vesicles acquire the pH of the external solution (Gandhi and Stevens, 2003), as well as compounds that are in the extracellular milieu [HRP, Jones et al., 1977, Meshul and Pappas, 1984, Parducz, 1986; fluorescein-conjugated dextran, Bonzelius and Zimmermann, 1990, Masur et al., 1990; the buffer Tris (during kiss-and-run events), Gandhi and Stevens, 2003], whereas Ca2+ is taken up through the exocytotic pore of mast cell granules (Raison et al., 1999). If Ca2+ is taken up into SVs during endocytosis, we would expect SVs in recently stimulated tissues to have higher concentrations of Ca2+ as is found in Torpedo (Parducz and Dunant, 1993; Parducz et al., 1994) and the rat (Parducz et al., 1987). We would also anticipate the number of SVs containing Ca2+ to increase over a time course similar to that of compensatory endocytosis rather than that of elevated [Ca2+]i evoked by stimulation, and this also is observed (Parducz and Dunant, 1993; Parducz et al., 1987, 1994). These data suggest that Ca2+ is concentrated in SVs as a result of endocytosis from the extracellular solution rather than by a process of Ca2+ uptake from the cytosol.
If an SV is endocytosed into the cytosol with a Ca2+ gradient of over 4 log units across its membrane, implications are evident for both the loading of transmitters into SVs and the facilitation of transmitter release. Ca2+ would exchange for protons via antiporters in the SV membrane, generating a proton motive force for “first stage” loading of transmitters, whereas the Ca2+ pump may run in reverse. Data supporting such an exchange of protons and Ca2+ was reported from cultured fibroblast cells, in which the time course of Ca2+ loss from endosomes matched the time course of acidification (Gerasimenko et al., 1998). Furthermore, acidification of endosomes was blocked by bafilomycin, a specific blocker of the vacuolar proton ATPase, which in turn blocked the leak of Ca2+ from the endosomes.
A slow leak of Ca2+ from a freshly endocytosed SV might contribute to the local [Ca2+]i and thus might contribute to short-term facilitation of transmitter release. The effect of the leak on volume-averaged [Ca2+]i may be difficult to detect because of its limited spatial extent (close to the PM, across which it is immediately extruded) and low flux rate. Indeed, if 25 vesicles, each of inner diameter 21 nm (Karunanithi et al., 2002) and containing 2 mm Ca2+, were endocytosed into a bouton 3 μm in diameter, where they instantaneously lost all of their Ca2+ to the cytosol, it could raise [Ca2+]i no more than 20 nm. Considering that acidification of the SVs (and presumably loss of Ca2+) has a time course of ∼430 msec (Gandhi and Stevens, 2003), the effect on volume-averaged [Ca2+]i would be undetectable.
SV Ca2+ transporters: physiological implications
The maximum volume-averaged [Ca2+]i attained in the MN terminals during high-frequency stimulation at 22°C was ∼0.4 μm. This value is in general agreement with [Ca2+]i reported in other MN terminals during stimulation (∼1 μm; crayfish, Tank et al., 1995, Msghina et al., 1999; frog, Suzuki et al., 2000; mouse, David and Barrett, 2000) but is well below estimates for microdomain [Ca2+] of 10-30 μm (Meinrenken et al., 2003). However, both values contrast sharply with the optimal Ca2+ concentration (500 μm) for inducing H+ release from SVs isolated from sheep brain (Gonçalves et al., 1998). If Ca2+ at a concentration comparable with extracellular levels (∼2 mm) is trapped inside newly endocytosed SVs, then the Ca2+ gradient across the SV membrane in vivo (10-3:10-7, intra-SV/cytosolic) contrasts sharply with the gradient manipulated in vitro (10-9:10-3) in which Ca2+ uptake was observed in isolated SVs (Gonçalves et al., 1998). Similarly, the proton gradient across the SV membrane in vivo (10-5.7:10-7; intra-SV, Miesenböck et al., 1998; cytosolic, Berridge et al., 2003) is unlikely, by exchange activity alone, to result in Ca2+ uptake, yet the levels manipulated in vitro are favorable for Ca2+ uptake (10-8:10-7). Thus, it is reasonable to expect that, in vivo, the SV membrane Ca2+-H+ exchanger would be important for the concentration of H+ in the vesicles, particularly in the presence of a high intra-lumenal [Ca2+], and not relevant to Ca2+ sequestration or intraterminal [Ca2+] homeostasis during or after APs. The Ca2+ affinity of the SV membrane Ca2+ pump (∼5-50 μm) falls well within the range of microdomain [Ca2+] estimates, but our measurements could detect no influence on volume-averaged [Ca2+].
Footnotes
This study was supported by grants from the Canadian Institutes for Health Research (G.T.M., M.P.C., H.L.A.) and from the Natural Sciences and Engineering Research Council of Canada (H.L.A.). We are grateful to Ken Dawson-Scully for assistance with depletion experiments, to Philip Francis for collection and analysis of behavioral data, to Chris Gergely for performing the fura-dextran calibration, to Amin Kay for analysis of Ca2+-imaging data, to Mingshan Xue for helpful discussions, and to Marianne Hegström-Wojtowicz for managing Drosophila stocks and for help with manuscript preparation.
Correspondence should be addressed to Dr. G. T. Macleod, Department of Physiology, 1 Kings College Circle, University of Toronto, Toronto, Ontario M5S 1A8, Canada. E-mail: greg.macleod{at}utoronto.ca.
Copyright © 2004 Society for Neuroscience 0270-6474/04/242496-10$15.00/0