Activation Of A Peptidergic Synapse Locally Modulates Postsynaptic Calcium Influx

Neuropeptides are known to have a wide array of modulatory actions on neuronal targets. One of their effects is to regulate synaptic transmission through the modulation of ion channels (Walker et al. 1988; Man-Son-Hing et al. 1989). In particular, neuropeptides have been shown to modulate calcium currents (Brezina et al. 1987; Walker et al. 1988; Man-Son-Hing et al. 1989; Bley and Tsien, 1990) in many neurones. Since neuropeptides can act distally as neurohormones or serve paracrine functions, in addition to their direct neurotransmitter roles, it is often difficult to determine the source of peptides underlying this type of modulatory action. In the present study, we examined the connection between a putative peptidergic neurone (neurone VD4) and its target (neurone Pl) (N. Syed, personal communication).

Neurone VD4 is a putative homologue of the Lymnaea stagnalis visceral white interneurone (VWI) (Benjamin, 1984), which has been shown to be a crucial element of the respiratory central pattern generator (Syed et al. 1990). By isolating VD4 and its target neurone Pl in culture, we were able to study the events underlying presynaptic modulation by a peptidergic source.

The peptide FMRFamide was initially identified in the clam Macrocallista nimbosa (Price and Greenberg, 1977). It has since been found in several molluscan species (Price, 1986), together with other FMRFamide-related peptides (FaRPs). FaRPs have been shown to have potent modulatory actions on central neurones of these molluscs (Cottrell et al. 1984; Colombaioni et al. 1985; Belardetti et al. 1987; Brezina et al. 1987; Kramer et al. 1988). In the nervous system of the pond snail Helisoma trivolvis, FMRFamide and two other FaRPs have been identified (Bulloch et al. 1988) and shown to exert modulatory effects both peripherally and centrally (Coates and Bulloch, 1985; Murphy et al. 1985; Bulloch et al. 1988).

The sources of the FMRFamide responsible for the modulation of these H. trivolvis neurones have not yet been established, although there are numerous FMRFamide-like immunoreactive neurones within the central nervous system (CNS), including VD4 (Murphy et al. 1985). The goals of this study were to identify the FaRP complement of neurone VD4; to determine the actions of these FaRPs on the calcium currents of the target cell Pl, and to examine the regulation of calcium influx in Pl when synaptically connected neurone VD4 was stimulated to release its neuropeptides.

All experiments were performed on laboratory-raised, adult specimens of the albino (red) pond snail Helisoma trivolvis. Animals were maintained in aquaria on a diet of lettuce and Purina Trout Chow.

Neurones Pl and VD4 were isolated from the pedal and visceral ganglia, respectively, using the techniques outlined by Haydon et al. (1985). In brief, the central ganglia were removed from the animal and treated with 0.2% trypsin for 25 min. Neurones were plated in either adhesive or non-adhesive culture conditions. Cells plated on 35 mm Falcon (no. 3001) Petri dishes coated with poly-L-lysine adhered to the substratum, a prerequisite for neurite outgrowth. For voltage-clamp experiments in which spherical neurones were required, cells were plated onto 35 mm Falcon (no. 1008) Petri dishes containing 2 ml of defined medium [DM; 50% Leibowitz-15 (Gibco) with H. trivolvis salts; Wong et al. 1981] to which 20μl of fresh snail haemolymph was added. Haemolymph prevents neuronal adhesion to the culture dish and suppresses neurite extension (Haydon, 1988).

Indirect immunocytochemical methods were applied to cultured neurones VD4 and Pl. Neurones were left overnight at 4°C in Zamboni’s fixative, rinsed in 0.187moll⁻¹ phosphate-buffered saline (PBS), and exposed for 10min to 0.3% Triton-X and 0.02% sodium azide in PBS containing 3% goat serum, prior to application of the primary antibody. A rabbit polyclonal antiserum raised against FMRFamide (1:200 dilution) was provided by J. Bishop (Chronwall et al. 1984; O’Donohue et al. 1984) and was applied to the neurones for 24 h at room temperature (RT, 22-25°C). Following further washes in PBS, the preparations were incubated with fluorescein-conjugated goat anti-rabbit IgG (Sigma, 1:100 dilution) for 2 h at RT. To test the specificity of the primary antibody, preadsorption controls were conducted using FMRFamide and GDPFLRFamide at concentrations of 5×10⁻⁴moll⁻¹.

Isolated VD4 neurones (10 per sample) were dissolved in acetone (0.5 ml); the acetone was then filtered through a Nylon syringe filter (0.45 μm pore size), dried (Speed-Vac) and the residue resuspended in aqueous high performance liquid chromatography (HPLC) solvent (1 ml of 0.1% trifluoroacetic acid, TFA). Half of the extract was loaded onto the HPLC directly and the other half was oxidized with hydrogen peroxide before HPLC, as previously described (Price et al. 1990). A flow rate of 0.5 ml min⁻¹ through an RP-300 column (2.1mm×220mm) with a gradient from 16% acetonitrile (ACN) in 0.1% TFA to 36% ACN in TFA over 20 min was used. Fractions (0.5 min) were collected from both runs, dried (Speed-Vac) and resuspended in 0.1 ml of radioimmunoassay (RIA) buffer; 25 μl samples of each fraction were used for RIA with antiserum Q2 (Bewick et al. 1990; Price et al. 1990). The RIA was standardized with the peptide NDPFLRFamide. Peaks were identified by comparing their elution times with those of synthetic peptides and by their behaviour after oxidation.

Conventional electrophysiological techniques were used to record synaptic interactions between neurones Pl and VD4 within the nervous system and in cell culture. To record from neurones in ganglia, microelectrode penetration was facilitated by a 30 s trypsin treatment of the visceral and pedal ganglionic sheaths. The intracellular recording electrodes were filled with 1.5 mol l⁻¹ KC1 (resistance, 20–30 MΩ) and the preparation was bathed in normal H. trivolvis saline containing; 51.3mmoll⁻¹ NaCl, 1.7 mmol l⁻¹ KC1, 4.1mmoll⁻¹ CaCl₂, 1.5 mmol l⁻¹ MgCl₂ and 5 mmol l⁻¹ Hepes at pH 7.3. Neuronal activity was monitored using Getting preamplifiers (model 5A) and stored on videotape using a Vetter 420F recorder.

Neurones were isolated from the nervous system and were cultured in 1% haemolymph in DM for 2–3 days. Immediately prior to the experiment, cells were individually transferred to a recording chamber (a Falcon no. 1008 dish previously coated with poly-L-lysine) containing DM. The spherical neurone adhered to the base of the dish and remained immobilized for the duration of the voltage-clamp experiment. Patch pipettes (with d.c. resistances of 1–2MΩ) were filled with a solution containing 35 mmol l⁻¹ CsCl, 5 mmol l⁻¹ MgCl₂, 5 mmol l⁻¹ EGTA, 5 mmol l⁻¹ ATP, 1 mmol l⁻¹ GTP and 5 mmol l⁻¹ Hepes (pH adjusted to 7.3 with CsOH). The external solution consisted of 4.1 mmol l⁻¹ CaCl₂, 1.5 mmol l⁻¹ MgCl₂, 1.7 mmol l⁻¹ KC1, 30 mmol l⁻¹ tetraethylammonium bromide, 10 mmol l⁻¹ 4-aminopyridine, 30 mmol l⁻¹ sucrose and 10 mmol l⁻¹ Hepes (pH adjusted to 7.3 with tetraethylammonium hydroxide). Signals were filtered with a corner frequency of 1kHz. Cells were held at a potential of -60mV and depolarizing command potentials were delivered at 30 s intervals. Leakage and capacitative transients were digitally subtracted using appropriately scaled hyperpolarizing pulses during data acquisition using pClamp software (Axon Instruments, CA).

Membrane-impermeant Fura-2 pentapotassium salt (Molecular Probes Inc., CA) was pressure-injected (using a Picospritzer; General Valve) into neurone Pl following penetration of the cell with a micropipette containing 10 mmol l⁻¹ dye in 10mmoll⁻¹ Hepes. The dye was allowed to diffuse through the neurites before data acquisition. We estimated that Fura-2 was at a concentration of no more than 100 μmol l⁻¹ after microinjection. Neurone VD4 was penetrated with a microelectrode containing 1.5 mol l⁻¹ KC1.

On completion of calcium imaging, the KCl-filled microelectrode was removed from neurone VD4 and the cell was re-impaled with a Lucifer-Yellow-filled microelectrode. Lucifer Yellow was ionophoresed into neurone VD4, and the fluorescence of the dye was excited using a 485 nm bandpass filter, a 510 nm dichroic mirror and a 515 nm longpass filter. Lucifer Yellow images of presynaptic neurone VD4 were compared to the Fura-2 images of postsynaptic neurone Pl and to the phase-contrast images obtained prior to experimentation to determine sites of contact between the neurites of the two cells. In some preparations successful Lucifer Yellow injections were not achieved because of the fragility of neurone VD4. However, since we found little evidence of neurite fasciculation between Pl and VD4 it was possible to determine discrete regions on Pl neurites that contacted VD4, and other regions that were without apparent contact.

The effects of bath-applied FMRFamide or GDPFLRFamide on calcium transients of neurone Pl were determined in some cell pairs. Calcium transients were acquired either using images as described above or with a photomultiplier tube using UMANS type 5.0 software (C. M. Regan, Urbana, IL).

Stimulation of VD4 produced a slow hyperpolarization of neurone Pl in situ (Fig. 1A). This synaptic connnection was reconstructed in cell culture. Neurones VD4 and Pl were isolated from the CNS and plated in culture conditions that permitted neurite outgrowth. After 3–8 days in culture, both neurones extended neurites that allowed contact to be established between the cells. Stimulation of neurone VD4 reliably produced a slow inhibitory postsynaptic potential in neurone Pl, a response similar to that observed in situ (Fig. 1B). This synaptic (action is chemically mediated since it is (i) reduced in magnitude by experimental hyperpolarization of neurone Pl, (ii) accompanied by an increase in membrane conductance in Pl, and (iii) reversibly abolished by high-Mg²⁺/zero-Ca²⁺ saline (not illustrated).

Neurite-bearing neurones VD4 and Pl, grown for 3–8 days in culture, were fixed, permeabilized, incubated with an FMRFamide antiserum and labelled with a fluorescein-conjugated second antibody (see Materials and methods). The neurites and cell body of neurone VD4 stained positively (N=7), while neurone Pl was unlabelled, displaying only somatic autofluorescence (Fig. 2). Preincubation of the FMRFamide antiserum with either FMRFamide (N=4) or GDPFLRFamide (N=3) at concentrations of 5×10⁻⁴moll⁻¹ successfully preabsorbed the antiserum, preventing positive immunoreactivity of neurone VD4. Thus, neurone VD4, but not neurone Pl, exhibits FMRFamide-like immunoreactivity.

An extract of five cells supplied sufficient material for either HPLC or RIA. Two major peaks are apparent with the unoxidized extract. The largest is at the elution position of FMRFamide and the next largest is at the position of FLRFamide. A small peak is seen near the position expected for GDPFLRFamide (Fig. 3). After oxidation, the FMRFamide peak disappears (oxidation shifts the position of the peptide but, more importantly, makes it almost unreactive with the antiserum used), but the putative FLRFamide and GDPFLRFamide peaks remain, as expected. Since the RIA was standardized with NDPFLRFamide, which is about 10 times more immunoreactive than FMRFamide, the major peak in Fig. 3A actually represents about lOpmol of peptide or 2prnol per cell.

We examined the actions of FMRFamide, FLRFamide and GDPFLRFamide on the calcium currents of neurone Pl. Two macroscopic calcium currents were identified in Pl; a transient, low-voltage-activated current (LVA) was detected on depolarization to -20 mV and a sustained, high-voltage-activated current (HVA) was detected at command potentials of -10 mV or greater (see Haydon and ManSon-Hing, 1988). The tetrapeptides FMRFamide and FLRFamide were equipotent in reducing the HVA calcium current of Pl (Fig. 4A,C,D). However, the heptapeptide GDPFLRFamide, exerted only a small modulatory effect on the HVA calcium current and only at high concentrations (Fig. 4C,D). The threshold concentrations of both tetrapeptides was 3×10⁻⁸moll⁻¹, whereas for the heptapeptide it was 3x ICT⁶ mol l⁻¹. At a concentration of 10⁻⁵moll⁻¹, the highest dose tested, the tetrapeptides FMRFamide and FLRFamide reduced the current by 45.4±12.6% (mean±s.D., N=6) and 51.0±9.4% (N=5), respectively. The heptapeptide GDPFLRFamide only reduced the current by 7.7±6.4% (N=9) (Fig. 4C). The actions of these FaRPs were fully reversible upon washout. The LVA calcium current was unaffected by the FaRPs.

To determine whether the FaRPs might act through G-protein-coupled receptors, GTPγS (100 μmoll⁻¹) replaced the GTP in the patch pipette to promote irreversible receptor-activation of G proteins. In large neurones, such as Pl, the concentration of dialyzed GTPγS is small, resulting in little G protein activation; thus, the ligand (FMRFamide or FLRFamide) was applied to promote exchange of GDP for GTPyS. FMRFamide (N=10) and FLRFamide (N=4) caused irreversible reductions in the HVA current of Pl in the presence of GTPγS (Fig. 4B). After peptide washout, subsequent application of peptide was without effect on the calcium current magnitude. These observations are consistent with the hypothesis that FaRPs act through receptor-coupled G proteins to modulate the HVA calcium current.

Since synaptic interactions frequently occur on neuronal processes, we investigated whether the modulatory actions of the peptides present in VD4, could be detected in neurites of Pl. Neurone Pl was plated alone, or in the presence of VD4, on an adhesive culture substratum that supports neurite extension. After 3–8 days of culture, the action-potential-evoked calcium transients of neuritebearing cells were examined. To monitor internal calcium levels, Fura-2 was pressure-injected into the cell body of neurone Pl from a recording microelectrode. A constant train of action potentials (characteristically consisting of 10 action potentials at 2 Hz) was evoked in Pl at intervals of 1–2 min to promote calcium loading of neurites. The fluorescent emission of Fura-2 within a selected region of the neurites from Pl was detected using a photomultiplier tube or, in some experiments, a field of neurites was imaged using a SIT camera. Calcium levels were estimated using ratiometric techniques.

Bath application of FMRFamide at 5×10⁻⁷moll⁻¹ (a concentration found to affect HVA calcium current without preventing spike initiation) caused a 78.3±7.7% (mean±S.E.M, N=4) reduction in the calcium transient of Pl cell neurites evoked in response to a constant train of action potentials. Similar results were obtained with FLRFamide (N=2). Since the threshold concentration for GDPFLRFamide on the HVA calcium current was 10⁻⁵moll⁻¹, we used this higher dose to test for the effects on calcium transients. Fig. 5 is a recording demonstrating the effects of bath-applied FMRFamide and GDPFLRFamide on the action-potential-evoked calcium transient in a region of the neurites of a Pl cell. In contrast to the minor effect of GDPFLRFamide on the macroscopic calcium current of Pl (7.7±6.4%, N=9; 10⁻⁵ mol l⁻¹, see Fig. 4C), the heptapeptide (N=10 from five cells; 10⁻⁵moll⁻¹) reversibly reduced the calcium transient by 60.4±8.1%. In addition to modulating calcium transients, application of all three FaRPs caused a simultaneous hyperpolarization of neurone Pl.

To determine whether VD4 stimulation also reduced the calcium transient in neurites of Pl, pairs of VD4 and Pl neurones were plated into adhesive culture conditions and allowed to extend neurites to form cell-cell contacts. Neurone Pl was pressure-injected with Fura-2 and the calcium transients in its neurites were imaged using quantitative ratiometric imaging techniques. Fig. 6 shows a pair of synaptically connected neurones injected with fluorescent dyes. Two regions of interest in which ratiometric images were obtained are highlighted. The calcium levels of region 1 (an area of contact) are shown in Figs 7A and 8. Image pairs (340nm, 380 nm excitation) were recorded at 4-s intervals and, after the third image pair had been acquired, neurone Pl was stimulated at a frequency of 4 Hz for 10s to evoke 40 action potentials. Action potentials reliably increased the calcium level of Pl cell neurites. After recording three calcium transients (at 3-min intervals), peptidergic neurone VD4 was depolarized for 10s to evoke a train of action potentials. In all preparations (N=4 cell pairs) the subsequent calcium transient in Pl was reduced in magnitude (Figs 7A, 8; N=8 sites of contact). In the example shown in Fig. 7, stimulation of Pl elevated the calcium level of region 1 (control conditions) from 87 to 322 nmol l⁻¹ (Fig. 7A). Following VD4 stimulation, action potentials in Pl only raised the calcium level from 70 to 214 nmol l⁻¹. Thus, the calcium transient was reduced from 235 to 144 nmol l⁻¹ following stimulation of peptidergic neurone VD4. Fig. 8 shows the magnitude of the calcium transient over time and the modulatory action of VD4. Multiple stimuli applied to VD4 at 3-min intervals caused a cumulative reduction in the calcium transient of Pl, which reversed after cessation of VD4 stimulation.

Data were recorded from regions of Pl cell neurites that had direct physical contact with VD4 processes and from regions in which there was no discernible contact. VD4 stimulation reduced the calcium transient in Pl cell neurites by 30.6±2.79% (mean±s.E.M.) at regions of contact with VD4 (N=8 regions from four cells). In contrast, VD4 stimulation had no significant effect on calcium transients (3.1 ±5.1% reduction) in regions of non-contact (N=7 regions from four cells). This observation is illustrated in region 2 of the preparation in Fig. 6, where simultaneous dye-fills of Pl and VD4 were achieved. Stimulation of VD4 did not appreciably modulate the calcium transient in the non-contact region of Pl (7% reduction; Fig. 7B), whereas the transient was reduced by 37% in the region of contact (Fig. 7A).

The exogenous application of synthetic peptides has been used, in the past, to study peptidergic modulation of identified neurones. Through this approach the importance of peptides in the modulation of several ionic currents (Walker et al. 1988; Taussig et al. 1989; Bley and Tsien, 1990) and in the regulation of the secretory machinery (Man-Son-Hing et al. 1989; Dale and Kandel, 1990) has been demonstrated. However, global application of neuropeptides may not simulate the true nature of synaptic peptide release. In the present study we have established that neurone VD4 contains multiple peptides and that this neurone is the presynaptic element of an inhibitory connection with neurone Pl in the central nervous system of H. trivolvis. By successfully isolating the synapse in culture, we have been able to use this system to investigate the nature of the modulatory actions of a peptidergic neurone on its target.

Although VD4 exhibited FMRFamide-like immunoreactivity, the precise FaRP complement was unknown. Previous work on the CNS of H. trivolvis has identified FMRFamide, FLRFamide and GDPFLRFamide as the main congeners of FMRFamide (Bulloch et al. 1988). In Lymnaea stagnalis the mRNA encoding the tetrapeptides (FLRFamide and FMRFamide) is found in neurones distinct from those expressing heptapeptide-encoding mRNA (J. F. Burke, personal communication). The high levels of FMRFamide and FLRFamide found in VD4 of H. trivolvis suggest that this neurone is primarily a ‘tetrapeptide’ neurone. However, the existence of a small peak at the GDPFLRFamide elution position indicates that, in H. trivolvis, individual neurones may express both tetrapeptides and extended FaRPs.

The reconstruction of the VD4/P1 synapse in culture provides an accessible system for investigating the consequences of peptide release from neurone VD4 onto its target. Stimulation of VD4 has multiple modulatory actions on neurone Pl. First, a hyperpolarization occurs in Pl as a result of activity in VD4. The effect of this action of VD4 stimulation is to reduce the excitability of Pl. The ionic basis of the hyperpolarization in Pl has not been fully characterized, but preliminary evidence suggests that it is mediated by an increase in a potassium conductance (J. E. Richmond, unpublished observations). Such increases in potassium conductance due to exogenous FMRFamide application have been reported in other molluscan neurones (Belardetti et al. 1987; Taussig et al. 1989). Second, the activity-dependent calcium transient in Pl is reduced following stimulation of VD4. This decrease in calcium entry may be attributable, in part, to a reduction in the voltage-activated calcium current of Pl by the release of FMRFamide and FLRFamide, as demonstrated by the exogenous application of FaRPs to voltageclamped spherical Pl neurones. However, it is likely that the modulation of the calcium transient is also due to additional mechanisms, since the heptapeptide GDPFLRFamide, which had little effect on the calcium current, reliably reduced the calcium transient. We propose that the FaRPs endogenous to neurone VD4 have a compound action on calcium influx (1) through a reduction of the HVA calcium current and (2) through a hyperpolarization (possibly due to an increased potassium conductance), causing a further reduction in calcium influx during Pl cell action potentials.

We wish to thank Drs Mark Zoran, Fred Bahls and Helen Man-Son-Hing, and Trent Basarsky, for constructive criticism of the manuscript, and Drs A. Don Murphy, N. Syed and J. F. Burke for personal communications. This work was supported by NIH grants NS26650 and NS24233 (P.G.H.) and NIH grant HL28440 (D.A.P.) and the Iowa State University Biotechnology Council. P.G.H. is an Alfred P. Sloan Fellow.