INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As an experimental organism, Drosophila is at the forefront of genetic and molecular research, and in addition affords the opportunity for physiological experimentation. Despite much physiological work on synaptic transmission, progress has been impeded by lack of reliable measurements of the intracellular calcium concentration ([Ca²⁺]_i). Earlier attempts to measure presynaptic [Ca²⁺]_i employed membrane permeant Ca²⁺ indicators (Dawson-Scully et al. 2000; Karunanithi et al. 1997; Umbach et al. 1998). However, this technique is capricious, sometimes gives contradictory results, and loads all cell types, generating high background fluorescence. In addition, resolution of presynaptic Ca²⁺ signals for single action potentials (APs) has not been accomplished. To overcome current limitations, we employed a different technique: exposure of cut axons to dextran-conjugated fluorescent Ca²⁺ indicators (O'Donovan et al. 1993). We also designed a new physiological solution in which isolated preparations remain stable for many hours. We have measured resting [Ca²⁺]_i, and nerve-evoked Ca²⁺ signals and observed remarkably rapid Ca²⁺ kinetics that enable the neuron terminal to maintain Ca²⁺ homeostasis during high-frequency activity. These observations provide the basis for further comparisons with mutant phenotypes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were performed on Drosophila Canton S wandering third instar larvae, raised on cornmeal agar with dry yeast at 21 ± 1°C. Larvae were dissected (Jan and Jan 1976) in chilled Schneider's insect medium (Schneider's) (Sigma, St. Louis MO) and pinned to the Sylgard (Midland, MI) base of a 0.5-ml perfusion bath. Dissections were performed in Schneider's because it renders the preparation quiescent and stable. After dissection, except where indicated, Schneider's was replaced with the new physiological solution [Hemolymph-Like No.6 (HL6)], which was renewed every 30 min. HL6, shown in Table 1 (components from Sigma; pH adjusted to 7.20 with 1 M NaOH, approximately 1 ml/l), is based on recent hemolymph analyses (Pierce et al. 1999).

Electrophysiological measurements were obtained with sharp intracellular microelectrodes filled with 1.5 M KCl and 1.5 M K-acetate (50-70 M). Excitatory junction potentials (EJPs) and miniature EJPs (mEJPs) were recorded using an A-M Systems (Carlsborg, WA) Neuroprobe amplifier 1600, low-pass filtered at 5 kHz, and acquired to hard disk at 20 kHz by a PCI-6024E multifunction I/O card (National Instruments, Austin TX) and Strathclyde Electrophysiology Software (WCP for Windows V3.1.4, University of Strathclyde, Strathclyde, UK).

Motor neurons were forward-filled with fluorescent indicators [Oregon Green BAPTA-1 (OGB-1), 10 kDa; Calcium Green-1, 10 and 70 kDa; Calcium Crimson, 10 kDa; Fura-2, 10 kDa; and Fluo-4, 10 kDa; Molecular Probes, Eugene OR] dissolved in water. All Ca²⁺ imaging, except that on Fura-2, was performed using a BioRad-600 confocal scan-head (BioRad, Mississauga, Ontario, Canada) on a Nikon microscope (Optiphot-2) with a 40× Nikon water-immersion objective (0.55 NA). The Argon ion laser was operated at low power and further attenuated to 0.5% transmission using neutral density filters. The pinhole was opened to its maximum aperture. A green light-emitting-diode, placed in the optical path, was lit for 2 ms to mark the beginning of stimulation. Fluorescence (F) is consistently reported with background subtracted [arbitrary units (au); Figs. 3 and 4]. F/F is defined as the change in F during stimulation, relative to F prior to nerve stimulation. N denotes an animal while n denotes a measurement.

Fura-2 fluorescence images were acquired through a 530 ± 35-nm band-pass filter by an intensified CCD camera (model IC-100, PTI, Princeton NJ). A filter-wheel (model 5240, Pacific Scientific, Metaltek Instruments, Raleigh NC) was used to excite the preparation with light from a mercury arc lamp alternately through 350 ± 5 nm and 385 ± 5 nm band-pass filters (Omega Optical, Brattlebro, VT). The camera and filter-wheel were mounted on a Nikon (Optiphot-2) microscope (Olympus water-immersion objective: 40×, 0.7 NA) and controlled by an Axon Instruments (Foster City CA) Digidata 2000 frame-grabber supported by AIW 2.2 software. Equation 5 of Grynkiewicz et al. (1985) was used to calculate absolute [Ca²⁺]_i values. Values of R_min and R_max were obtained in situ through bath application of 20 µM ionomycin (Calbiochem, La Jolla CA) in HL6 containing either no added Ca²⁺ and 5 mM EGTA (Sigma; R_min) or 2 mM [Ca²⁺]_o (R_max). The value used for the dissociation constant (K_D; 371 nM) of Fura-2 was that determined in situ for an astrocyte cell line by Petr and Wurster (1997).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To maintain a physiologically stable preparation long enough to allow dissection, indicator transport, and measurement, we needed to develop a suitable physiological solution. Preliminary trials showed that previous solutions employed for this preparation (Table 1) do not maintain its physiological stability for an adequate period. In the standard solution introduced by Jan and Jan (1976), vacuolation of muscle cells and unstable resting membrane potentials (RMPs) are commonly observed (Stewart et al. 1994). The more "hemolymph-like" solution (HL3) introduced by Stewart et al. (1994) produces less muscle cell vacuolation and more stable RMPs than the standard solution. Even in HL3, some vacuolation is frequently observed (Fig. 1A), indicating that the solution is not fully compatible with the preparation. More significantly, in HL3 solution, nerve excitability is often lost after 2 h, while muscle cells often exhibit destructive spontaneous contractions.

View larger version (99K): [in this window] [in a new window]   Fig. 1. Physiological characterization of muscle cells and motor neurons in Hemolymph-Like No.6 solution (HL6). A: muscle cells 6 and 7 in Schneider's [top; arrowheads indicate nerve terminals (n)] and 3 min after an exchange of Schneider's for HL3 (bottom; arrows indicate several vacuoles). B: muscle cells 6 and 7 in Schneider's (top) and 3 min after an exchange of Schneider's for HL6 (bottom). Scale bar, 20 µm. C: excitatory junction potentials (EJPs) recorded in muscle cell 6 (61 mV), 0.5 h after exposure to HL6 (1.25 h after starting dissection) and 1 h after the axon was cut (top). EJPs recorded from the same preparation (47 mV) after 4.5 h in HL6 (bottom). D: amplitude frequency histograms of resting membrane potentials (RMPs) measured in HL6; 1.25 h after exposure to HL6 (top), and the RMPs of the same 41 cells after 5.5 hours in HL6 (bottom).

We tested the effects of hemolymph components (Pierce et al. 1999) on muscle cell vacuolation and RMP. Trehalose in combination with the principal hemolymph amino acids, at the concentrations reported by Pierce et al. (1999), eliminated vacuolation completely (Fig. 1B). Correspondingly, motor axon excitability was maintained with no change in threshold for a longer time. Axon excitability threshold (1.52 ± 0.68 V, mean ± SD, N = 61; 0.3 ms) remained constant when stimulation amplitude was maintained below double threshold. Although the improved solution has 31% less Cl than HL3, solutions in which only [Cl] was reduced produced inconsistent results for vacuolation and excitability. [K⁺] was increased from 5 to 25 mM, closer to hemolymph [K⁺] values. Thus the HL6 solution (Table 1) incorporated the amino acids, normal blood sugar trehalose, and [K⁺] of 25 mM. Glutamate, although reported in hemolymph (Pierce et al. 1999; but see Irving et al. 1979), was omitted because it interferes with synaptic transmission.

The time course of the EJP recorded in HL6 is indistinguishable from that in HL3, and EJPs summate when the nerve is stimulated at a sufficient frequency (Fig. 1C). EJP amplitude and time-course did not change appreciably for several hours. The average RMP in muscles 6 and 7 was initially 62.3 ± 5.7 mV and was still 53.1 ± 6.9 mV after 5.5 h in HL6 (41 cells, N = 6; Fig. 1D). The high RMP in this relatively high [K⁺] solution is predicted from the low [Na⁺]_o (54.7 mM), as the dependence of the muscle RMP follows the Goldman-Hodgkin-Katz equation with a ratio of 0.23 for Na⁺ to K⁺ permeability (Jan and Jan 1976). The frequency of mEJPs, generally a good indicator of the integrity of neuron terminals, did not change significantly over the same period: 1.17 ± 0.67 to 0.84 ± 1.14 Hz (21 cells, N = 6). mEJP amplitude, which reflects properties of both pre- and postsynaptic cells, also remained stable [0.77 ± 0.16 (SE) mV, initially; 0.85 ± 0.13 mV after 6.25 h; 6 cells, N = 3). Thus HL6 solution provides a much longer period of stability for dissected preparations than previous solutions.

To load motor neuron terminals with dextran-conjugated indicators, we cut a hemi-segment nerve close to the ventral ganglion with sharp microscissors, and drew it by gentle suction into a heat-polished glass pipette tip with an inner diameter of 12 µm (Fig. 2A). A fine plastic tube was then introduced (Fig. 2A, inset) to deliver an indicator solution (0.5-1 mM in Schneider's) to the nerve ending. The indicator solution requires placement on the cut nerve ending within approximately 5 min, before complete resealing of the axons occurs. After 20-40 min, the indicator solution was extracted from the tip of the pipette and replaced with Schneider's or HL6. Indicator appeared in motor neuron terminals within 1 h. The axoplasmic indicator concentration (fluorescence level) remained stable at sampled locations after 1-h equilibration. This technique consistently loaded bouton types Ib, Is, II, and III (Johansen et al. 1989) on body wall muscles 6, 7, 12, and 13 (Fig. 2D). We successfully loaded all dextran-conjugated indicators attempted: OGB-1, 10 kDa; Calcium Green-1, 10 and 70 kDa; Calcium Crimson, 10 kDa; Fura-2, 10 kDa; and Fluo-4, 10 kDa.

View larger version (133K): [in this window] [in a new window]   Fig. 2. Motor neuron terminals filled with dextran-conjugated indicators. A: severed end of a hemi-segment nerve fits snugly in the loading pipette. Scale bar, 50 µm. Inset: nerve (n) held in the tip of a pipette (p), showing the tip of the dye delivery tube (t). B: pipette in place over the mid-line of a filleted larval preparation. C: same field of view as in B, observed using epi-fluorescence optics. Fluorescent indicator is visible in the lumen of the pipette and in the nerve. Scale bar, 100 µm (B and C). D: segment nerve (3) at body-wall muscles 7, 6, 13, and 12, forward-filled with 1 mM Oregon Green BAPTA-1 (OGB-1) over 40 min, and viewed with confocal microscopy (17-µm depth, z-series projection). Indicator fluorescence is clearly visible in boutons of the motor neuron terminals. Scale bar, 20 µm. E: bouton types Ib and Is on muscles 7 and 6. F: bouton types Ib, Is, and II on muscle 13. Scale bar, 20 µm in both E and F (6-µm depth projections).

When the nerve was stimulated with trains of pulses, a change in fluorescence (F) occurred (Fig. 3A). Fluorescence (F) increased rapidly toward a plateau over the first 1 s of stimulation (Fig. 3B) with OGB-1 (K_D = 170 nM; Molecular Probes). The level of the plateau in Fig. 3C is approximately linearly related to the stimulation frequency from 5 to 20 Hz (Fig. 3D). Average F/F at 10-Hz stimulation is 60 ± 6% (mean ± SE, N = 6). The extracellular Ca²⁺ concentration ([Ca²⁺]_o) in HL6 was set at 0.5 mM, the highest level at which [Ca²⁺]_o could be maintained without muscle contraction at 20-Hz stimulation.

View larger version (26K): [in this window] [in a new window]   Fig. 3. Fluorescence changes in response to trains of stimulation pulses. A: 4 sequentially scanned images of type Ib boutons filled with OGB-1 on muscle 6. Each image is scanned left-to-right, 2 ms per line, and is 128 lines long (256 ms). Distance between consecutive images represents an acquisition delay (304 ms). White scan line in the second image marks the start of a 5-s train of electrical pulses (2 V, 0.3 ms) delivered to the nerve at 20 Hz. Pulses are represented by a series of arrow tics below the images. The look-up-table, used to represent fluorescence intensity (0-255) in each pixel, is shown on the right. Scale bar, 10 µm. B: time course of the average pixel fluorescence intensity minus background (F) of the bouton enclosed by a box in the 4th image in A, during a 20-Hz stimulus train [arbitrary units (au)]. Open symbols correspond to the 4 images presented in A. C: percentage change in bouton F (F/F) in response to 5-s stimulation trains of different frequencies (5, 10, 20, 30, 40 Hz: n = 3, 3, 3, 1, 1, respectively). Data for frequencies above 20 Hz could not be collected beyond the start of the trains due to muscle contraction. D: a plot of average maximum F/F reached during the trains of different frequencies in C.

Almost all loaded preparations showed a response to a single nerve stimulus. Fast sampling by confocal line scanning (Fig. 4) revealed fast intracellular Ca²⁺ dynamics. At 2-Hz stimulation (Fig. 4, A and B), the plot of F indicates that [Ca²⁺]_i does not accumulate from one pulse to the next, whereas at frequencies at 10 Hz and above (Fig. 4, C and D), [Ca²⁺]_i builds up. The consistent height of consecutive transients is a good indication that the calcium indicator (OGB-1) is not saturating in this [Ca²⁺]_i range. F decay of OGB-1 for 200 ms following a single isolated pulse was fitted by a single exponential [time constant () = 58.8 ± 8.4 (SE) ms, N = 6], as was F decay following the last pulse of a 1-s stimulus train at 20 Hz ( = 61.8 ± 6.6 ms, N = 3). Using the lower affinity indicator, Fluo-4 10 kDa (K_D = 3.1 µM; Kreitzer et al. 2000), F decay for 200 ms following a single pulse was 55.4 ± 8.6 ms (N = 4). The K_D of both indicators was established in vitro in a 100 mM KCl solution; however, the K_D of these indicators is likely to be several-fold higher in the motor neuron cytoplasm (Thomas et al. 2000). Measurements using Fluo-4 10 kDa were made in Schneider's which provided a high [Ca²⁺]_o (5.4 mM) while also preventing movement during high-frequency stimulation trains. To simulate locomotor activity, motor neuron terminals filled with Fluo-4 10 kDa were stimulated at 80 Hz for 2 s in the high [Ca²⁺]_o environment of Schneider's. F decay of Fluo-4 following the last pulse of the train was also rapid: 90.4 ± 6.5 ms (N = 4; single exponential fit over 750 ms). A curve fit to the sum of two exponentials did not provide a better description of F decay. Where high concentrations of calcium indicator are used in the loading solution (>2 mM), and to a lesser extent when loading times are excessive (>40 min),  measured using OGB-1 has been estimated at over 100 ms.

View larger version (59K): [in this window] [in a new window]   Fig. 4. Fluorescence changes in response to single pulse stimulation. Aa: scanned image of a type Ib bouton filled with OGB-1 on muscle 6 (shown in full in Ac; scale bar, 10 µm). Scale bar in Aa is 5 µm. Ab: a serial line scan through the center of the bouton, as marked in Aa, 4 ms per line, over a period of 2 s. Each arrow tic indicates the time at which a pulse was delivered to the nerve (2 V, 0.3 ms). Time scale bar, 200 ms. The look-up-table used for F intensity is shown on the right. B: plot of F for each line in Ab (au). Cb: a serial line scan through the bouton in Ca while stimulating at 10 Hz. D: plot of F for each line in Cb.

To verify that these techniques can be used with ratiometric imaging we used Fura-2 10 kDa dextran to estimate the resting [Ca²⁺]_i. In seven preparations, the average [Ca²⁺]_i was 23 ± 11 (SE) nM. Preliminary data indicate that resting [Ca²⁺]_i may be higher in the standard solution of Jan and Jan (1976) than in HL6, suggesting that the new solution better preserves presynaptic [Ca²⁺]_i homeostasis.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have provided a reliable assay for [Ca²⁺]_i in Drosophila larval motor neuron terminals; kinetics of signals for single stimuli and for trains; and a new physiological solution, HL6, which supports longer periods of vitality for the larval preparation, allowing the application of techniques that require longer execution times. The time-constant of Ca²⁺ clearance is rapid for both single pulses and high-frequency trains, indicating a very efficient extrusion mechanism.

The technique of forward-filling has the advantages of high sensitivity, good temporal resolution, exclusion from intracellular organelles, and stable indicator concentration. In addition, it provides the potential for loading terminals with other substances for probing synaptic function. We found that molecules as large as 70 kDa (Calcium Green-1 70 kDa dextran) can be loaded into axons. Thus other molecules such as chaperone proteins, competitive peptide fragments, and caged compounds may load successfully. This opens up a wide range of acute experiments in which intracellular proteins and reagents can be added or modified.

Kerr et al. (2000) used the GAL4-upstream activating sequence (UAS) system to drive expression of the Ca²⁺ indicator protein, cameleon, in the neurons of Caenorhabditis elegans. The same technique has been applied to the motor neuron terminals of larval Drosophila (Reiff and Schuster 2000). However, this technique cannot be used to detect responses to single pulse stimulation in single cells. Even the most recent version of cameleon indicators (Truong et al. 2001) has not demonstrated sufficient Ca²⁺ responsiveness to detect a signal from one AP in single cells.

The estimate of resting [Ca²⁺]_i reported here (23 nM) is similar to that reported in Drosophila giant neuron cell culture (22 nM; using a cell-free Fura-2 K_D estimate of 147 nM) (Berke and Wu 2002). However, estimated [Ca²⁺]_i values are very sensitive to the K_D used, and the Fura-2 K_D was not determined in motor neurons in the present study.

The plateau level of [Ca²⁺]_i accumulation increases in direct proportion to the frequency of pulses (see Helmchen et al. 1996), indicating that the Ca²⁺ clearance mechanism does not saturate at these levels of Ca²⁺ influx. The rapid fluorescence decay when stimulation ceases, indicates an efficient Ca²⁺ clearance mechanism. Even after long trains of stimulation, we did not observe a fluorescence decay time-constant on the order of minutes as reported by Umbach et al. (1998).

The technique described here has been used to examine whether there is a Ca²⁺ handling defect associated with the reduction in neurotransmitter release at the motor neuron terminals in the cacophony temperature-sensitive Drosophila mutant cac^ts2 (Kawasaki et al. 2000) at higher temperatures. There is a profound reduction in Ca²⁺ entry into boutons in response to single APs as part of the larval cac^ts2 phenotype at higher temperatures (Macleod et al. 2001). The technique has also been applied at the motor neuron terminals of csp mutant Drosophila larvae and preliminary data (K. Dawson-Scully and G. T. Macleod, personal observations) confirm earlier observations using the AM calcium indicator loading protocol of Dawson-Scully et al. (2000).

Detection of Ca²⁺ entry into individual boutons with single APs provides the means to test hypotheses for the regulation of [Ca²⁺]_i. The time-constant reported here, approximately 60 ms at 21°C, is fast relative to values for other neuron terminals at room temperature (Helmchen and Tank 1999). There are no adjacent sinks in the form of large inter-bouton connections to explain the rapid decay in terms of diffusion. We postulate that Drosophila motor neuron terminals possess a relatively large complement of Ca²⁺ extrusion molecules, consistent with their maintained high rates of impulse activity during locomotion when motor axons conduct AP volleys at 100 Hz, every couple of seconds (Barclay et al. 2002).