Development of Ca²⁺ hotspots between Lymnaea neurons during synaptogenesis

Abstract

Calcium (Ca²⁺) channel clustering at specific presynaptic sites is a hallmark of mature synapses. However, the spatial distribution patterns of Ca²⁺ channels at newly formed synapses have not yet been demonstrated. Similarly, it is unclear whether Ca²⁺ ‘hotspots’ often observed at the presynaptic sites are indeed target cell contact specific and represent a specialized mechanism by which Ca²⁺ channels are targeted to select synaptic sites. Utilizing both soma–soma paired (synapsed) and single neurons from the mollusk Lymnaea, we have tested the hypothesis that differential gradients of voltage-dependent Ca²⁺ signals develop in presynaptic neuron at its contact point with the postsynaptic neuron; and that these Ca²⁺ hotspots are target cell contact specific. Fura-2 imaging, or two-photon laser scanning microscopy of Calcium Green, was coupled with electrophysiological techniques to demonstrate that voltage-induced Ca²⁺ gradients (hotspots) develop in the presynaptic cell at its contact point with the postsynaptic neuron, but not in unpaired single cells. The incidence of Ca²⁺ hotspots coincided with the appearance of synaptic transmission between the paired cells, and these gradients were target cell contact specific. In contrast, the voltage-induced Ca²⁺ signal in unpaired neurons was uniformly distributed throughout the somata; a similar pattern of Ca²⁺ gradient was observed in the presynaptic neuron when it was soma–soma paired with a non-synaptic partner cell. Moreover, voltage clamp recording techniques, in conjunction with a fast, optical differential perfusion system, were used to demonstrate that the total whole-cell Ca²⁺ (or Ba²⁺) current density in single and paired cells was not significantly different. However, the amplitude of Ba²⁺ current was significantly higher in the presynaptic cell at its contact side with the postsynaptic neurons, compared with non-contacted regions. In summary, this study demonstrates that voltage-induced Ca²⁺ hotspots develop in the presynaptic cell, concomitant with the appearance of synaptic transmission between the soma–soma paired cells. The appearance of Ca²⁺ gradients in presynaptic neurons is target cell contact specific and is probably due to a spatial redistribution of existing channels during synaptogenesis.

Calcium (Ca²⁺) influx through voltage-gated Ca²⁺ channels is required for transmitter release (Katz & Miledi, 1967; Charlton et al. 1982; Simon & Llinas, 1985; Llinas et al. 1992; Heidelberger et al. 1994; Mackenzie et al. 1996; see also Augustine et al. 1987; Regehr & Tank, 1994; Borst & Sakmann, 1996; Reuter, 1996; Catterall, 1998), its modulation (see Littleton & Bellen, 1995; Bennett, 1997), and neuronal plasticity (see Finkbeiner & Greenberg, 1996) at both vertebrate and invertebrate synapses. In most instances, the voltage-induced Ca²⁺ influx appears to be localized to specific subcellular domains, termed ‘transient calcium microdomains’ or ‘hotspots’(Chad & Eckert, 1984; Connor, 1986; Lipscombe et al. 1988; Smith et al. 1993; Cooper et al. 1996; Llano et al. 1997). A variety of immunocytochemical (Westenbroek et al. 1995; see Reuter, 1996), freeze fracture (Dreyer et al. 1973; Heuser et al. 1974; Pumplin, 1983, Roberts et al. 1990) and toxin labelling (Robitaille et al. 1990; Cohen et al. 1991) techniques have been used to demonstrate that the Ca²⁺ hotspots may result from Ca²⁺ channel clustering at presynaptic endings. For instance, histochemical studies using specific Ca²⁺ channel toxins, such as ω-conotoxin GVIA, have provided direct evidence that N-type Ca²⁺ channels ‘cluster’ at specific presynaptic sites at the neuromuscular junction (NMJ; Robitaille et al. 1990; Cohen et al. 1991).

In addition to their roles at mature synapses, Ca²⁺ channel activities at specific synaptic sites are also remodelled during development. For example, in developing mouse sensory neurons, a Q-type Ca²⁺ channel is up-regulated, whereas the P-type Ca²⁺ channel is down-regulated, during synapse formation (Hilaire et al. 1996). Similarly, calcium channel ‘up-regulation’ occurs at synapses between cultured hippocampal neurons (Basarsky et al. 1994), rat cerebellar granule cells (D'Angelo et al. 1994) and leech neurons (Fernandez-de-Miguel et al. 1992). However, the spatial distribution patterns of Ca²⁺ channels at newly formed synapses have not yet been demonstrated. Moreover, it is unknown whether the Ca²⁺ hotspots are indeed target cell contact specific, and thus present a specialized mechanism by which Ca²⁺ channels are targeted to specialized synaptic sites. This lack of fundamental knowledge regarding developmental neurobiology is due, in most instances, to the anatomical complexity of the mammalian brain, where direct access to developing synapses is often not feasible.

To obtain direct and simultaneous access to individual pre- and postsynaptic somata and their synaptic sites, we have developed synapses between the cell bodies of identified Lymnaea neurons: right pedal dorsal 1 (RPeD1) and visceral dorsal 4 (VD4). The soma–soma synapses are morphologically and electrophysiologically similar to those seen in vivo (Feng et al. 1997, 2000; Hamakawa et al. 1999; Woodin et al. 1999). In this study, utilizing Ca²⁺ imaging and whole-cell patch clamp recordings, we demonstrate that voltage-induced Ca²⁺ hotspots/gradients develop in the presynaptic Lymnaea neurons VD4 and RPeD1 concomitant with the appearance of synaptic transmission with their respective postsynaptic partners, and that these gradients are target cell and contact site specific. Moreover, we show that the total Ca²⁺ current amplitude in single cells did not differ significantly from their paired counterparts. However, when paired with its synaptic partner, the intensity of Ca²⁺ current within the presynaptic cell was significantly higher at its contact site with the postsynaptic cell compared with non-contacted areas. These data demonstrate that Ca²⁺ hotspots develop at newly formed (inhibitory and excitatory) synapses in a target cell and contact site specific manner. We propose that Ca²⁺ hotspots seen at the synaptic site may be due to their redistribution from non-synaptic to synaptic sites during synapse formation.

METHODS

Animals and cell culture

All experiments were performed on neurons isolated from the intact nervous system of the fresh water pond snail, Lymnaea stagnalis. Animals were raised and maintained in the aquarium at the animal care facility of the University of Calgary and kept at room temperature (20–22 °C) on a 12 h–12 h light–dark cycle (See Feng et al. 1997 for details). Snails with a shell length of 10–15 mm (approximately 2–4 months old) were used.

Animals were dissected under sterile conditions as described previously (Syed et al. 1990). Briefly, the central ring ganglia were isolated and treated with trypsin (Type III, 3 mg ml⁻¹, Sigma) for 25 min, followed by trypsin inhibitor (Type III, 3 mg ml⁻¹, Sigma) for 10 min. The ganglia were then incubated in high osmolarity defined medium (DM, containing 40 mm glucose; Sigma G7021). DM consisted of serum-free 50 % (v/v) Liebowitz L-15 medium (Gibco -special order), with additional salts (mm: NaCl 40.0, KCl 1.7, CaCl₂ 4.1, MgCl₂ 1.5, Hepes 5.0), 10 mm glucose, 1.0 mm l-glutamine and 20 μg ml⁻¹ gentamicin. The pH was adjusted to 7.9 with 1 n NaOH. The ganglia were pinned down to the bottom of a dissection dish and the connective tissue sheath surrounding the neuronal somata was removed. Identified cells were isolated by applying gentle suction via a fire-polished pipette (2 mm, WPI, 1B200F) which was coated with Sigmacote (Sigma). Isolated cells were subsequently transferred to a poly-l-lysine (Sigma)-coated culture dish (Falcon 3001) containing either DM or brain conditioned medium (CM Ridgway et al. 1991). To obtain soma–soma synapses, identified neurons were isolated with their axon stumps attached and juxtaposed in cell culture. Within a few hours of contact, both cells resorbed their axon stumps completely, so that the somata were in direct contact with each other. All experiments were performed on cells that were maintained in culture for 12–24 h.

Fura-2 [Ca²⁺]_i imaging

The Ca²⁺-sensitive, membrane-impermeable pentapotassium salt of fura-2 (Molecular Probes Inc.) was used to determine intracellular Ca²⁺ levels. Fura-2 was dissolved in distilled water to a final concentration of 6 mm. The neurons were loaded with fura-2 using intracellular recording techniques. Specifically, sharp glass microelectrodes (resistance 40–80MΩ) were filled with fura-2 and connected to an amplifier headstage (Getting, model 5; Dagan, 8700) via silver chloride-treated wire. After impaling neurons, the neuronal resting membrane potential was measured and fura-2 was injected iontophoretically using 0.4 nA, 2 s hyperpolarizing current pulses applied at a rate of 0.3 Hz, for 3–10 min using a function generator (Master-8, AMPI). To ensure successful fura-2 loading, applied pulses and corresponding changes in cell membrane potentials were continuously monitored on a storage oscilloscope (Gould). The extent of fura-2 loading was monitored by visualizing the fluorescence intensities.

Intracellular Ca²⁺ levels were determined using a ratiometric fura-2 imaging technique, as described previously (Grynkiewicz et al. 1985; Tsien et al. 1985; Tsien & Poenie, 1986). Specifically, fura-2-loaded neurons were observed under an inverted microscope (Zeiss Axiovert 135; Zeiss, ON, USA) equipped with a Plan-Neofluar × 100/1.3 NA oil-immersion objective lens. Fura-2 fluorescence was excited at 340 and 380 nm wavelengths generated by UV light from a 100 W Hg/Xe-arc lamp, alternately passed through 340 and 380 nm excitation filters, which were controlled by a computer (Macintosh Quadra 950)-operated rotating filter wheel (Empex Imaging Inc., Mississauga, ON, USA). The neurons were exposed for 200 ms at each excitation wavelength. In order to minimize photobleaching of the fura-2 signal, saturation of the camera, and/or photodamage to neurons, a neutral density filter (0.3–3.0; Omega Optical) was placed in the excitation pathway to reduce the intensity of excitation light (Goldberg et al. 1992). Emitted fluorescence light, generated after the excitation light, was reflected via a 430 nm dichroic mirror, passed through a 510 nm emission filter, and ultimately detected by an intensified charged coupled device (ICCD) camera (Paultek Imaging, Nevada City, CA, USA). Fluorescence images were digitized by a QuickCapture frame grabber board (Data Translation, Mississauga, ON, USA) and stored on the computer. The images were acquired using Ratio 1.3 image capture software (kindly provided by Dr S. B. Kater, School of Medicine, University of Utah).

Free intracellular Ca²⁺ concentrations ([Ca²⁺]_i) were estimated as described by Grynkiewicz et al. (1985). An in vitro calibration was performed using a series of calcium solutions (Calcium Calibration Buffer Kit with Magnesium II; Molecular Probes, Eugene, OR, USA) loaded into glass calibration tubes (20 μm width; In Vitro Dynamics, Rockaway, NJ, USA). Measurement parameters were Q = 1.95 (representing minimum fluorescence intensity divided by maximum fluorescence intensity, F_min/F_max); fluorescence ratio in the presence of saturating Ca²⁺, R_max = 3.75; fluorescence ratio in the absence of Ca²⁺, R_min = 0.7; K_d = 264.5 nm. These K_d values are consistent with that obtained in other fura-2 Ca²⁺ imaging systems (Connor, 1986; Goldberg et al. 1992; Rehder & Kater, 1992).

To estimate [Ca²⁺]_i, the images acquired at 340 and 380 nm were ratioed on a pixel-by-pixel basis using Ratio 1.3 software. Mean values for [Ca²⁺]_i of any given rectangular region of interest were measured after a correction for a background signal recorded in the absence of the fluorescent dye.

Two-photon laser scanning microscopy

Two-photon excitation was performed using a Coherent Ti:Sapp Mira IR femtosecond laser pumped with a 5 W Verdi solid state laser. A laser scanning microscope (LSM 510, Zeiss attached to an Axioskop2 FS) was used for imaging neurons, which were simultaneously recorded using the whole-cell voltage clamp technique. The acquisition of images by the LSM 510 was synchronized with the recordings using digital pulses generated by Clampex. Neurons were patched with electrodes containing 500 μm Calcium Green. Stable recordings were maintained initially for 25 min to allow neuronal loading with the dye prior to imaging experiments. The Mira laser was tuned to 840 nm for visualization of Calcium Green-injected cells. A series of images in the Z-plane were obtained to reconstruct the whole cell, and to ensure that images were obtained from regions of the neurons that were clearly directly apposed to the adjacent paired cell. Fluorescence images were obtained coincidently with transmitted images (from the IR beam) to monitor the morphology of the recorded cell. After an initial morphological survey of the cell, responses to depolarizing command potentials were observed in full frame mode to estimate regions that showed calcium increases indicated by large fluorescence changes. Then line scans were obtained to monitor calcium transients on a fast time scale from different regions of the cells. For example, a line would stretch from the active zone adjacent to the paired neuron to the membrane distal to the paired surface. Fluorescence changes at the membrane from different regions of the cells were expressed as values relative to the base line fluorescence intensity. There were no striking differences or gradients in the resting fluorescence intensities across the cell. The dye loading appeared to be equal across the cell cytoplasm. The two-photon laser scanning image only obtains fluorescence values from a very narrow volume at the focal plane, making contribution from dye excitation above or below the focal plane unlikely.

Electrophysiology

Whole-cell patch clamp recordings

A whole-cell, voltage clamp (membrane ruptured) recording technique was used to study Ca²⁺ current activities (Hamill et al. 1981). Briefly, patch electrodes were prepared from glass haematocrit tubes (Western Scientific, Richmond, BC, USA), which were pulled in two steps on a vertical pipette puller (Kopf model 750, Tujunga, CA, USA). Pipettes were filled with filtered (0.22μm) pipette solution consisting of CsCl (29 mm), CaCl₂ (2.3 mm), EGTA (11 mm), Hepes (10 mm), ATP-Mg (2 mm) and GTP-Tris (0.1 mm), adjusted to pH 7.4 with CsOH. Pipettes with resistance in the range 0.8–2.0 MΩ were used in this study. The bath reference electrode, consisting of an agar bridge and a chlorided silver wire (World Precision Instruments), was filled with the bath solution consisting of tetraethylammonium chloride (TEA-Cl, 47.5 mm), MgCl₂ (1 mm), CaCl₂ (4 mm), Hepes (10 mm) and 4-aminopyridine (4-AP, 2 mm), adjusted to pH 7.9 with TEA-OH. The pipettes were mounted onto the input headstage of a patch clamp amplifier (Axopatch 1D, Axon Instruments, Foster City, CA, USA). After obtaining a gigaohm-seal, cell capacitance was neutralized, and the series resistance was compensated. The Ca²⁺ current measured in this study was filtered at 1 kHz using a 4-pole Bessel filter and digitized at a sampling frequency of 5 kHz. The voltage command generation and data acquisition were carried out using a 486 IBM compatible computer equipped with a Digidata 1200 interface (Axon Instruments) in conjunction with pCLAMP-6 software (Axon Instruments).

Whole-cell Ca²⁺ current data were analysed using Clampfit (pCLAMP 6, Axon Instruments) software and plotted using Sigmaplot 4.0 (Jandel Scientific).

Differential perfusion system

Cells were constantly perfused with Ca²⁺-free extracellular solution consisting of Tris-HCl 50 mm, MgCl₂ 10 mm, 4-AP 4 mm, pH 7.9. Under these experimental conditions no current was recorded. An additional microperfusion system utilized a fire-polished glass pipette (4 μm diameter) to deliver a narrow stream of Ba²⁺ solution (BaCl₂ 10 mm, NMG (N-methylglucamine)–HCl 50 nm, Hepes-NMG 5 mm, 4-AP 3 mm, pH 7.9). Because NMG had different optical density than the normal saline solution, the Ba²⁺ solution stream was clearly distinct and easily discernible. Under phase contrast optics, it was thus relatively simple to direct the stream of Ba²⁺ solution onto a select area of interest on neuronal somata. This method allowed us to measure Ba²⁺ current selectively from only those areas of the cell that were exposed to the stream of Ba²⁺ solution.

All experiments involving whole-cell current analysis from single and paired cells were performed blind, i.e. the patch clamper was unaware of the precise identity of the paired cells.

Statistics

All parametric data are presented as means ± s.d. Statistical analysis was carried out using Sigmastat (Jandel Scientific). Differences between mean values from each experimental group were tested using Student's t test for two groups, or one-way analysis of variance (ANOVA) for multiple comparisons. Differences were considered significant if P < 0.05.

RESULTS

Ca²⁺ gradients develop at the contact site between soma–soma paired neurons VD4 and RPeD1

If synaptic sites between soma–soma paired Lymnaea neurons were indeed specialized (Feng et al. 1997), we hypothesized that voltage-induced Ca²⁺ gradients would develop in the presynaptic cell at its site of contact with the postsynaptic neuron. To determine whether a voltage-induced Ca²⁺ signal was spatially localized at synaptic sites between the soma–soma paired cells, the right pedal dorsal 1 (RPeD1) and visceral dorsal 4 (VD4) neurons were isolated and paired in CM, in which reciprocal inhibitory synapses reliably reform (Feng et al. 1997). Single or paired VD4 cells were studied12–24 h after cell isolation. Calcium images were acquired prior to (Ctrl), during (Stim, at the end of the 10th action potential) or after (recovery, Recov) intracellular stimulation. The rationale for using the 10 action potential protocol was based on our electrophysiological analysis, which revealed that the peak of postsynaptic response was achieved after 10 action potentials. To accomplish this, either unpaired or paired (with RPeD1) VD4 cells were injected intracellularly with the Ca²⁺ indicator dye, fura-2. The dye was allowed to equilibrate for 15–30 min and the injected cells were subsequently re-impaled with sharp intracellular electrodes and ratiometric fluorescence images were acquired under various experimental conditions. The stimulation frequency consisted of 10 action potentials and the images were acquired at the end of the 10th action potential. As shown in Fig. 1A, electrical stimulation of an unpaired VD4 elicited a diffuse Ca²⁺ signal, which was distributed evenly throughout the soma. The resting intracellular Ca²⁺ level in unpaired VD4 cells was 264 ± 172 nm (n = 13), which increased significantly (252 ± 47 %; n = 13) during a brief period of intracellular stimulation (10 action potentials; P < 0.05). When monitored 30 s after the stimulation, the Ca²⁺ signal had returned to its resting level (278 ± 125 nm; n = 13). These data show a voltage-induced rise in the intracellular Ca²⁺, which was distributed uniformly throughout the unpaired VD4 cells.

Figure 1. Ca²⁺ hotspots develop in VD4 at its contact site with RPeD1.

A, fura-2 images were acquired from an unpaired VD4 cell. Fura-2-injected cells were subsequently impaled with sharp intracellular electrodes and images were acquired prior to (Ctrl), at the end of the 10th action potential (Stim) and following recovery (Recov, 30 s interval after the acquisition of stimulus-induced images). A voltage-induced Ca²⁺ signal in an unpaired VD4 was evenly distributed throughout the cell. B, fura-2 was injected into a VD4, soma–soma paired with RPeD1 (as described above), and the images were acquired at rest (Ctrl), during electrical stimulation (Stim) and following recovery (Recov). The voltage-induced Ca²⁺ signal was localized at the contact region between the cells. The unpaired and paired cells are shown in the phase contrast pictures. Estimated [Ca²⁺]_i (colour scales) is in μm. Scale bar, 60 μm.

To determine the spatial patterns of Ca²⁺ signal in a paired VD4 neuron, fura-2 was injected intracellularly into VD4 and Ca²⁺ images were acquired as described above. The resting Ca²⁺ level of paired VD4 was found to be 130 ± 64 nm (n = 12). It is important to note that in most instances (80 %) of soma–soma pairing between VD4 and RPeD1, appropriate inhibitory synapses reform within 12–24 h of soma contact (Feng et al. 1997). The injected cells were re-impaled with sharp intracellular recording electrodes and Ca²⁺ images were acquired prior to, during, and following electrical stimulation with a frequency of 2 Hz. Compared with unpaired VD4, the voltage-induced Ca²⁺ signal in paired VD4 was predominantly localized at the contact sites with RPeD1 (Fig. 1B). Because this localized, voltage-dependent Ca²⁺ gradient appeared at cell contact/synaptic sites between the cells, for the sake of simplicity, we have used the term ‘Ca²⁺ hotspot’ to refer to the calcium gradient observed at cell-to-cell contacts/ synapses. It is also important to note that the number of action potentials (10) and the period of image acquisition (2 s) were kept constant for both unpaired and paired VD4 neurons.

To determine whether Ca²⁺ hotspots in a paired VD4 coincided with the appearance of synaptic transmission, the incidence of hotspots was compared indirectly with our previously published electrophysiological recordings obtained at 12–24 h (Fig. 1C, data modified from Feng et al. 1997). The incidence of Ca²⁺ hotspots (imaging data) and postsynaptic potential (defined as synapse) between VD4 and RPeD1 was similar (Fig. 1C, data from Feng et al. 1997), suggesting that the Ca²⁺ hotspots most probably reflect changes in the Ca²⁺ concentration that underlies synaptic transmission between the cells. Furthermore, because Ca²⁺ gradients developed at the soma–soma contact site, the above data lend further support to the idea that these synaptic/contact sites are, indeed, specialized.

Two-photon laser scanning microscopy (TPLSM) was next used to visualize, at a faster rate, the location and development of Ca²⁺ hotspots during membrane depolarization. Whole-cell voltage clamp recordings, and simultaneous measurements of Ca²⁺ dynamics, were obtained from the presynaptic VD4 neurons in pairs (n = 4). We monitored the development of the Ca²⁺ transient during voltage steps to 20 mV using line-scan mode, thereby allowing us to observe the onset and recovery of evoked Ca²⁺ fluxes. Figure 2A shows a representative example, in which a transmitted image was taken from RPeD1 paired with VD4. Calcium transients in response to 20 mV voltage steps were recorded from various regions. The greatest Ca²⁺ transient was observed in regions apposed to the postsynaptic neurons (Fig. 2B). The data showing the peak fluorescence intensity compared with baseline fluorescence are summarized in Fig. 2C. The voltage-induced increase in the peak fluorescence intensity is significantly lower at the non-contact site compared with the target cell contact site. After determining the region of greatest Ca²⁺ flux using the voltage step to 20 mV, we examined the magnitude of the Ca²⁺ signals at the hotspot region in response to a series of voltage steps. An I–V relationship with sample traces is shown in Fig. 2D. The calcium transients measured during the series of voltage steps showed graded amplitude with a peak at 20 mV (Fig. 2E). This indicates that the Ca²⁺ transients that we observed were principally due to calcium influx via voltage-activated channels.

Figure 2. Two-photon laser scanning microscopy showed that calcium influx was greatest at the membrane apposed to the RPeD1 paired VD4 neuron.

A, transmitted image of the neuron pair showing the patch pipette and the Calcium Green-filled VD4 neuron. Asterisks indicate the regions where line-scan imaging was taken. B, calcium transients were obtained using line-scan mode from the regions indicated by the asterisks in A. The cell was stepped to 20 mV to evoke calcium currents. The greatest change in dye fluorescence intensity was observed at the site of contact. ▪ indicates electrical stimulations. C, the average peak change in Calcium Green fluorescence (divided by total fluorescence; δF/F) during voltage steps to +20 mV recorded at the contact site (CS) versus that at the non-contact site (NCS). The peak δF/F was significantly greater at the contact site (P < 0.05). D, calcium currents (inset) and the I–V relationship obtained in RPeD1 paired VD4 cell (external Ca²⁺ 2 mm) is shown from which simultaneous calcium measurements (shown in E) were obtained. These calcium signals were obtained at the CS zone where large calcium signals were recorded. The calcium transients measured at the CS using the two-photon laser scanning microscope were graded with increasing depolarization and showed a voltage relationship similar to the whole-cell current. This indicates that the calcium transients were probably due to influx via voltage-gated channels.

We next sought to determine: (1) whether Ca²⁺ hotspots between soma–soma paired cells were common and fundamental to other presynaptic neurons, such as RPeD1, and (2) whether Ca²⁺ hotspots were target cell contact specific. To test the above two possibilities, RPeD1 neurons were simultaneously paired with both VD4 (target) and visceral dorsal 1 (VD1 – non-target) cells in a three-cell configuration. We have previously demonstrated that, in a three-cell mode, RPeD1 forms a specific synapse with VD4 but not with VD1 (Feng et al. 1997).

Ca²⁺ hotspots in a presynaptic neuron were target cell contact specific

Fura-2 was injected intracellularly into RPeD1 and Ca²⁺ gradients were monitored either at rest or during direct intracellular stimulation. Under these experimental conditions, Ca²⁺ hotspots were observed in RPeD1 at its contact site with VD4 but not at its contact site with VD1 (n = 5; Fig. 3A and B).

Figure 3. Voltage-induced Ca²⁺ hotspots observed in RPeD1 were target cell (VD4) contact specific.

To test for the specificity of Ca²⁺ hotspots, RPeD1 was soma–soma paired simultaneously with both VD4 and VD1 (in a three-cell configuration; see phase contrast image, inset). Fura-2 was injected into RPeD1 and images were acquired at rest (Ctrl), during electrical stimulation (Stim), and following recovery (Recov). Voltage-induced Ca²⁺ hotspots were observed in RPeD1 only at its contact site with VD4 but not at its contact site with VD1. A and B represent two different examples. Estimated [Ca²⁺]_i (colour scales) is in μm. Scale bar, 100 μm. C and D, summary data showing the specificity of target cell-induced, voltage-dependent Ca²⁺ gradient in RPeD1. In a three cell configuration, where RPeD1 was simultaneously paired with both target (VD4) and non-target (VD1) cells, voltage-induced Ca²⁺ hotspots were consistently observed in RPeD1 at its contact site with VD4 but not with VD1. C, ratio of [Ca²⁺]_i levels both at rest and during stimulation. D, absolute [Ca²⁺]_i. Data are presented as means ±s.d. (n = 5). *Significant difference (P < 0.05).

The intensity of the baseline (control) Ca²⁺ signal was significantly higher at the RPeD1–VD4 contact site (367 ± 141 nm, n = 5) compared with that at the RPeD1–VD1 contact site (116 ± 49 nm; P < 0.05), suggesting that an increase in the resting Ca²⁺ signal occurred at the target cell contact site. The evoked intracellular Ca²⁺ signal at each site was normalized against its respective control value (Stim/Ctrl). As shown in Fig. 3C, a two-fold increase in the evoked Ca²⁺ signal (230 ± 20 %, n = 5) was observed in RPeD1 at its contact site with VD4. This increase was significantly greater than that observed at the RPeD1–VD1 site (130 ± 10%, n = 5; P < 0.05, Fig. 3C). Absolute change in estimated [Ca²⁺]_i measured during stimulation is shown in Fig. 3D. The estimated Ca²⁺ values in RPeD1 at its contact site with the target cell (VD4) were significantly higher than the contact site with the non-target cell (VD1).

Together, the above data highlight two points. First, Ca²⁺ hotspots are a feature of inhibitory synapse between RPeD1 and VD4. Second, Ca²⁺ hotspots are target cell and contact site specific.

Target cell-induced Ca²⁺ hotspots also developed at the excitatory synapses

The data presented above demonstrate that Ca²⁺ gradients develop at the inhibitory synapses between RPeD1 and VD4 (mutually inhibitory synaptic partners). To determine whether Ca²⁺ hotspots were also a function of excitatory synapses, presynaptic neuron RPeD1 was soma–soma paired with its excitatory partner VD2/3 for 12–24 h, and fura-2 images were subsequently acquired as described above. Similar to the data described above for the inhibitory synapses (between RPeD1 and VD4), voltage-induced Ca²⁺ hotspots were observed in RPeD1 at its contact site with VD2/3 (100%, n = 6; as shown in Fig. 4). The resting Ca²⁺ level at the VD2/3 contact site was 328 ± 96 nm (n = 6). The stimulus-induced increase in Ca²⁺ signal intensity ([Ca²⁺]_i Stim/Ctrl) in RPeD1 was significantly greater at its contact site with VD2/3 (223 ± 31 %, n = 6) than that observed at the site opposite to that of the contact areas (141 ± 18 %, n = 6; P < 0.05). These results show that Ca²⁺ hotspots observed at inhibitory synapses between RPeD1 and VD4 are also a characteristic of the excitatory synapse between RPeD1 and VD2/3.

Figure 4. Voltage-induced Ca²⁺ hotspots developed in RPeD1 at its contact site with excitatory synaptic partner VD2.

Fura-2 was injected into RPeD1 soma–soma paired with its excitatory synaptic partner (VD2) and images were acquired at rest (Ctrl), during electrical stimulation (Stim), and following recovery (Recov). Ca²⁺ hotspots were observed in RPeD1 at its contact site with VD2. Estimated [Ca²⁺]_i (colour scale) is in μm. Scale bar, 100 μm.

Soma–soma synaptogenesis did not increase the whole-cell Ca²⁺ current density in synaptically paired RPeD1 or VD4 cells

To determine whether Ca²⁺ hotspots observed at the soma–soma contact site between RPeD1 and VD4 were due to an up-regulation of Ca²⁺ channel activity in the presynaptic cell, we compared the whole-cell Ca²⁺ currents in unpaired cells with those of synaptically paired RPeD1. We found that the Ca²⁺ current properties in paired RPeD1 cells were not significantly different from those of unpaired cells. Figure 5A shows the mean I–V curve of the single RPeD1 and representative tracings of Ca²⁺ current are shown in the insert. The half-activation voltage of the current in the single cells was found to be 2.33 ± 1.7 mV (n = 26) and the reversal potential was 55.2 ± 2.4 mV; whereas in the paired RPeD1 cells these values were 2.42 ± 1.4 mV (n = 24) and 55.9 ± 2.2 mV, respectively. Figure 5B shows that there was no significant change in the current density between paired (n = 24) and single RPeD1 cells (n = 26). Because VD4 also forms an inhibitory synapse with RPeD1, Ca²⁺ current densities in both unpaired and paired VD4 were also compared. As observed for RPeD1, no significant difference in Ca²⁺ current density was found between unpaired and paired VD4 cells (Fig. 5B). These data demonstrate that the density of whole-cell Ca²⁺ currents in either RPeD1 or VD4 cells does not change after synapse formation. These data do not, however, rule out the possibility that a localized increase in Ca²⁺ current may still occur at the contact compared with non-contact areas. To test this possibility, we took advantage of a differential perfusion system, which allowed us to selectivity superfuse either contacted or non-contacted sites in soma–soma paired cells.

Figure 5. The high voltage activated (HVA) Ca²⁺ current density in single RPeD1 was similar to that of synaptically paired cells.

Single or paired cells were maintained in culture for 12–24 h. Whole-cell Ca²⁺ currents were recorded in bath solution that contained 4 mm Ca²⁺ (Na⁺ -free), whereas the pipette solution contained Cs⁺ (K⁺ -free). A, the mean I–V relationships of the peak Ca²⁺ currents were obtained from the single RPeD1 cells. The current was recorded during 500 ms depolarization steps applied from a holding potential of −80 mV to +70 mV, in increments of 5 mV. Inset shows Ca²⁺ currents elicited in RPeD1 in response to various voltage steps (−25, −15, −10, −5 and 0 mV) lasting 500 ms from a holding potential of −80 mV. B, comparison of the peak Ca²⁺ channel density between single and paired RPeD1 and VD4 cells. Data are presented as means ±s.d.

Ba²⁺ permeable Ca²⁺ channels are differentially distributed at the presynaptic site in RPeD1 cells

As shown above, the density of the whole-cell Ca²⁺ current in RPeD1 did not change after synapse formation with VD4. These data do not, however, exclude the possibility that a selective aggregation of Ca²⁺ channels may occur at the presynaptic sites, with a concomitant down-regulation of these channels at the non-synaptic sites. If so, then an increase in Ca²⁺ current amplitude at contact sites is expected, whereas the intensity of Ca²⁺ current at the non-contacted sites (away from synapse) is likely to decrease. To test this hypothesis, a differential perfusion system was used in which the perfusate stream was visualized optically, and whole-cell recordings were made from the presynaptic partner of soma–soma paired neurons (Fig. 6A). Because most Ca²⁺ channels are also permeable to Ba²⁺ (see Hille, 1994), the Ba²⁺ conductance for the Ca²⁺ channel in RPeD1 was tested. We found that, consistent with earlier studies (Dreijer & Kits, 1995; Wildering et al. 1995), the peak current amplitude increased by 1.9 ± 0.5-fold (n = 8), when 4 mm external Ca²⁺ was replaced with 4 mm Ba²⁺. Therefore, to obtain a higher resolution of Ca²⁺ channel conductance, a 10 mm external Ba²⁺ solution was used in this study. Figure 6A shows that a stream of high optical density NMG solution containing 10 mm Ba²⁺ can be reliably visualized and selectively directed onto any given area of interest during the whole-cell recording. Figure 6B shows an example of evenly distributed Ca²⁺ channel activities throughout the entire soma of an unpaired RPeD1. The current recorded from either half of the cell in Ba²⁺ solution was virtually the same, and was approximately half the total whole-cell current (WC), which was recorded from the entire soma when it was superfused with the Ba²⁺ solution. In contrast, Fig. 6C shows that the portion of Ba²⁺ current amplitude in a paired RPeD1 somata at its contact side (CS) with VD4 was much greater than that at the non-contact side (NCS, away from synapse). Consistent with the data presented above, the sum of the total current recorded from both contacted and non-contacted sides was approximately equal to the WC recorded from the entire soma superfused with the Ba²⁺ solution. These data are summarized in Fig. 6D. A 3-fold increase in Ba²⁺ current activity in RPeD1 at its contact point with VD4 suggests that Ba²⁺ permeable Ca²⁺ channel proteins may locate at the contact site following synapse formation between soma–soma paired cells. To rule out the possibility that the enhanced Ca²⁺ current activity at the contact site between the paired cells (compared with single cells), did not result from perfusion artifacts, or from the physical barrier resulting from soma–soma pairing, RPeD1 was also paired with its non-target cell VD1. We found that the Ca²⁺ current activity recorded from RPeD1 at its contact site with VD1 was similar to that of a single RPeD1 neuron(Fig. 6D).

Figure 6. Differential distribution of Ca²⁺ channel in target-cell paired RPeD1 cells.

Isolated cells were paired in a soma–soma configuration. The cells were initially perfused with Ca²⁺-/Ba²⁺-free external solution as described in Methods. A 10 mm Ba²⁺ solution, containing high optical density NMG, was applied via an additional microperfusion pipette and the Ba²⁺ stream was visualized under phase contrast optics (A). A, phase contrast picture shows the high density perfusion stream directed at the non-cell contact side of a paired RPeD1 (with VD4). The pipette in the upper left corner contained the Ba²⁺ solution perfusion pipette, and the pipette in the lower right corner is the patch pipette. B, the peak Ba²⁺ current recorded from the whole cell, or two halves of an unpaired RPeD1 cell is presented. Perfusion of the high optical density Ba²⁺ solution on either half of the cell yielded an equal current activity, which was 50 % of the current recorded from the entire cell (WC). These values suggest that the channels were distributed evenly throughout the entire cell. C, the peak Ba²⁺ current recorded from a paired RPeD1 (with VD4). A RPeD1 soma was paired with VD4 and the whole-cell current was recorded from the RPeD1. The peak Ba²⁺ current recorded from the cell contact side (CS) was 3-fold larger than that of the non-contact side (NCS). The current recorded from the entire cell (WC) was the sum of the currents obtained from synaptic and non-synaptic sides. D, comparison of the ratio of the peak Ba²⁺ current from the cell contact side (CS) vs. that from the non-contact side (NCS) of RPeD1, paired with its target cell VD4 or non-target cell VD1. In the case of single cells, the ratio represents any half vs. the other half. Data were presented as means ±s.d. (n = 5). *Significant difference (P < 0.05). It is important to note that the Ca²⁺ current activity recorded from the contact site of a paired (with a non-target cell VD) RPeD1 was similar to that of a single RPeD1.

DISCUSSION

In this study we have demonstrated that Ca²⁺ gradients develop at specific presynaptic sites between soma–soma paired cells (both inhibitory and excitatory synapses), and that these hotspots coincide with the appearance of synaptic transmission between the cells. The voltage-induced Ca²⁺ gradients were target cell and contact site specific. Moreover, we have also demonstrated that while whole-cell Ca²⁺ currents do not change in synaptically paired cells, they are, nevertheless, differentially localized to the synaptic sites. Two-photon imaging in line-scan mode enabled us to determine the changes in [Ca²⁺]_i, concentration with excellent temporal resolution. These data thus provided us with the time course of changes in fluorescence intensity in different regions of the cell during a series of graded voltage steps. However, TPLSM necessitates the use of non-ratiometric dyes, such as Calcium Green, which do not provide absolute Ca²⁺ measurements. These data, nevertheless, support our claim that the Ca²⁺ hotspots result from Ca²⁺ influx. Our data do not, however, rule out the involvement of Ca²⁺-induced Ca²⁺ release from the intracellular stores in the formation of Ca²⁺ hotspots.

At the neuromuscular junction, the voltage-induced Ca²⁺ gradients result from the clustering of presynaptic Ca²⁺ channels. These clustered Ca²⁺ channels permit local influx of Ca²⁺ ions, which leads to transmitter release at specific synaptic sites (Regehr & Tank, 1984; Augustine et al. 1987). In addition to Ca²⁺ channels, other ion channels have also been shown to cluster at specific subcellular domains. For instance, Na⁺ channels were reported to cluster at presynaptic sites at the NMJ (Vabnick et al. 1996) as well as in Aplysia axons (Johnston et al. 1996; Salzer, 1997). Similarly, K⁺ channels have been shown to cluster at specific subcellular sites in rat central nervous system neurons (Alonso & Widmer, 1997). Moreover, Ca²⁺-gated K⁺ channels have been shown to co-cluster alongside Ca²⁺ channels where they regulate transmitter release at the frog NMJ (Robitaille et al. 1993; Yazejian et al. 1997, 2000). Voltage-dependent Ca²⁺ channels also have been shown to redistribute to specific synaptic sites during synapse formation between cultured hippocampal CA1 neurons (Jones et al. 1989). These studies suggest that ion channel clustering at specific and localized cellular regions is important and fundamental to various neuronal functions. Consistent with this notion are our data demonstrating that Ca²⁺ channel activity is, indeed, specialized at synapses between the somata of Lymnaea neurons. However, there exists a distinct possibility that the target cell contact may also induce protein phosphorylation-dephosphorylation changes, which, in turn, may alter the proportion of functional channels without an actual change in the Ca²⁺ channel distribution. Recent studies on Lymnaea neurons have used various conotoxins specifically to block voltage-gated Ca²⁺ channels in a subpopulation of peptidergic neurons (Fainzilber et al. 1996; Feng et al. 1997; Sasaki et al. 1999). In our hands, however, these toxins only partially block the total Ca²⁺ current in the cultured Lymnaea neurons used in this study. This differential sensitivity of Lymnaea neurons to toxins may be cell-type specific. Moreover, labelling these toxins with fluorescent probes has proved challenging (Feng et al. unpublished observation). Successful labelling of these toxins with fluorescent tags would prove beneficial in determining whether Ca²⁺ channels, or their various specific subtypes, redistribute to the synaptic sites between soma–soma paired cells.

The insertion of newly synthesized Ca²⁺ channels, or their redistribution at specific synaptic sites may not, however, be the only reason for Ca²⁺ gradients in the soma–soma paired cells. For instance, presynaptic, voltage-insensitive Ca²⁺ hotspots were detected immediately (within seconds to minutes) after contacts between pre- and postsynaptic cells in both vertebrates (Dai & Peng, 1993) and invertebrates (Funte & Haydon, 1993; Zoran et al. 1993). These Ca²⁺ gradients coincided with a target cell-induced activation of a cAMP-dependent protein kinase pathway in the presynaptic neuron (Funte & Haydon, 1993). In addition, Ca²⁺ channel activity has also been shown to be regulated by protein kinase C (Hall et al. 1995; Haydon & Drapeau, 1995; Stea et al. 1995) and cAMP-dependent protein kinase (Sculloreanu et al. 1993). Thus, a localized activation of protein kinase at the synaptic site may, in turn, selectively phosphorylate specific voltage-gated Ca²⁺ channels at the synapse, resulting in an enhanced Ca²⁺ influx (Funte & Haydon, 1993). For instance, in both frog (Dai & Peng, 1993) and the snail Helisoma (Funte & Haydon, 1993; Zoran et al. 1993), contacts with specific muscle target cells were considered necessary for an immediate increase in intracellular Ca²⁺ levels at the contact site.

Although the nature of cell–cell signalling mechanisms for these Ca²⁺ gradients has not yet been determined, either retrograde messengers or various membrane bound molecules may be involved (Haydon & Drapeau, 1995). For instance, cell adhesion molecules (CAM) such as NCAM and N-cadherin have been shown to activate both L- and N-type neuronal Ca²⁺ channels via G-protein-dependent pathways, leading to increased Ca²⁺ influx in neurons (Doherty et al. 1991). In addition to these membrane bound molecules, various target-derived retrograde molecules may also mediate Ca²⁺ channel modulation. For example, IGF-1 was shown to modulate N- and L- type Ca²⁺ channels in cerebellar granule cells via receptor tyrosine kinase-mediated pathways (Blair & Marshall, 1997). Because synapse formation between soma–soma paired Lymnaea neurons requires receptor tyrosine kinase activity (Hamakawa et al. 1999), it is plausible that a similar receptor tyrosine kinase-mediated signalling pathway may also be responsible for the target cell contact-induced Ca²⁺ hotspots observed in the present study. Taken together, the above studies suggest that a variety of cell–cell signalling mechanisms may generate either transient, or permanent, Ca²⁺ gradients in neurons by modulating their Ca²⁺ channel activities.

In this study, we did not observe significant differences in the whole-cell Ca²⁺ current density between single and paired RPeD1. Specifically, neither Ca²⁺ current activity nor I–V properties in single neurons differed from their paired counterparts. Our experiments with a differential perfusion system did, nevertheless, show either a target-cell contact induced up-regulation of Ca²⁺ channel activity at synaptic side, and/or a ‘down-regulation’ at the non-synaptic side. These data correlated well with our fura-2 and two-photon Ca²⁺ imaging results which showed that the paired cells are likely to up-regulate their Ca²⁺ channel activity at the contact sites, whereas these channels might be down-regulated elsewhere. Overall, this ‘redistribution’ may not, therefore, have altered the total Ca²⁺ current in RPeD1, since no change in the whole-cell current was observed. A similar redistribution and re-organization of Ca²⁺ current has been reported in other vertebrate (mouse: Hilaire et al. 1996) and invertebrate systems (leech: Drapeau et al. 1989, 1995; Catarsi & Drapeau, 1996). In leech, for instance, the postsynaptic Ca²⁺ current was found to be reduced during synapse formation as compared with the presynaptic cell (Cooper et al. 1992). Many other examples exist where Ca²⁺ channels are shown to be redistributed during development (Ritchie, 1982; Huguenard et al. 1988). This study is, however, the first to simultaneously monitor voltage-induced Ca²⁺ gradients at newly formed inhibitory and excitatory synapses, as well as the presynaptic somata.

Ca²⁺ influx through voltage-gated channels has been shown to regulate many aspects of neuronal development, ranging from ion channel maturation to neurite outgrowth and synapse formation (Spitzer, 1991). Since Ca²⁺ regulates a myriad of neuronal functions, its precise involvement in various developmental programs is often difficult to delineate from its conventional role(s) in neuronal excitability and transmitter release. However, the transient expression of various ion channels and neurotransmitter receptors during early development lends support to the idea that they may participate in various developmental programs (Spitzer, 1991). For example, in mouse embryonic sensory neurons, the T-, Q- and N-type current densities increase during early development, whereas P-type Ca²⁺ currents disappear altogether (Hilaire et al. 1996). Xenopus spinal neurons in vitro exhibit a 2-fold increase in Na⁺ channel density, whereas voltage-dependent K⁺ currents were found to increase 3-fold during development (O'Dowd et al. 1988). In chick embryonic neurons, Ca²⁺ current activity increases during synapse formation and decreases to lower steady-state levels thereafter (Jimenez et al. 1997). In addition to their spatial distribution, temporal patterns of Ca²⁺ channels may also play a pivotal role in the maturation of many developmental steps. Consistent with this idea are our earlier studies, which showed that blocking Ca²⁺ channel activity during soma–soma pairing between Lymnaea neurons VD4 and RPeD1 significantly reduces the incidence of synapse formation in cell culture (Feng et al. 2000). Although it is tempting to speculate that under such experimental conditions, Ca²⁺ hotspots would also fail to develop, further experiments are required to test this possibility.

In summary, the data presented in this study are consistent with our hypothesis that Ca²⁺ hotspots develop at contact sites between the somata of identified Lymnaea neurons, and that these hotspots are target cell and contact site specific and involve a redistribution of Ca²⁺ channels at the synaptic sites. The soma–soma synapse model provides us with an excellent opportunity to determine the significance of Ca²⁺ channel clustering and its activity in synapse formation, synaptic transmission and synaptic plasticity.