Aplysia californica (6–10 months old) purchased from the University of Florida were housed in tanks containing fresh aerated sea water (∼15°C) and fed with seaweed (Ulva lactuca) obtained from the Station Biologique at Roscoff, France.

Animals were anesthetized with an injection of a 60 ml isotonic MgCl₂ solution (in mM: 360 MgCl₂, 10 HEPES, adjusted to pH 7.5). Buccal ganglia were dissected under artificial sea water (ASW, in mM: 450 NaCl, 10 KCl, 30 MgCl₂, 20 MgSO₄, 10 CaCl₂, 10 HEPES, adjusted to pH 7.5). The isolated ganglion preparations were maintained at 15°C with a Peltier cooling device and bathed in standard ASW or modified saline during subsequent pharmacological experiments.

A “Low Ca + Co” solution was used to block all network chemical synaptic connections and was composed of: (in mM) 446 NaCl, 10 KCl, 30 MgCl₂, 20 MgSO₄, 3 CaCl₂, 10 CoCl₂, 10 HEPES, adjusted to pH 7.5. The NaCl concentration was adjusted to maintain the same osmolarity as ASW. Electrophysiological recordings under this saline started at least 20 min after perfusion onset, which was the time required to completely block chemical synapses and allow for recovery of neuronal resting membrane potentials to at least −50 mV (Bédécarrats et al., 2021). A calcium-free solution (“0 Ca + Co”) in which all calcium (CaCl₂) was replaced with equimolar (10 mM) cobalt (CoCl₂) was composed of: (in mM): 450 NaCl, 10 KCl, 30 MgCl₂, 20 MgSO₄, 10 CoCl₂, 10 HEPES, 0.5 EGTA, adjusted to pH 7.5). D-tubocurarine (Merck-Sigma-Aldrich) was used at a concentration of 1 mM in ASW. Tetrodotoxin (TTX, Tocris) was diluted in distilled water from a 0.1 mM stock solution and added to Low Ca + Co saline at a concentration of 1 μM. Tetraethylammonium (TEA, Merck-Sigma-Aldrich) was diluted in Low Ca + Co saline at a concentration of 50 mM. Flufenamic acid (FFA, Merck-Sigma-Aldrich) was first diluted in 100 mM DMSO and then in Low Ca + Co saline at a final concentration of 100 μM (DMSO, 0.1%). Cyclopianozic acid (CPA, Merck-Sigma-Aldrich) was diluted to 50 mM in DMSO and used at a final concentration of 20 μM in Low Ca + Co (DMSO, 0.04%).

Spontaneously expressed cycles of radula motor output (BMPs) were monitored with wire pin extracellular electrodes placed in direct contact with the cut stumps of the protraction I2 (I2 n.), retraction 2,1 (n. 2,1) and radular closure motor (Rn.) nerves. Intracellular electrodes were made from pulled glass capillaries (15–30 MΩ) filled with KCH₃CO₂ (2 M). Intracellular signals were amplified with an Axoclamp-2B amplifier (Molecular Devices, Palo Alto, CA), digitized with a CED (Cambridge Electronic Design) interface (sampling rate 5 kHz), then acquired and analyzed with Spike 2 software (Cambridge Electronic Design, UK). B63 interneurons and B31/B32 motoneurons were identified as previously described (Susswein and Byrne, 1988; Nargeot et al., 2009). Relatively brief intracellular current pulses (duration 5 or 8 s) were injected to trigger plateau-like activity in B63 under ASW. These pulse durations correspond approximately to the rising phase durations of plateau-like depolarizations triggered spontaneously by the ongoing intracellular calcium oscillation in B63. It is noteworthy that shorter pulses did not, or were less likely to, elicit plateaus. On the other hand, longer injected pulse durations did not enhance the expression or durations of plateaus because of the onset of the subsequent retraction phase of the BMP, which inhibits B63 activity. After chemical synapse blockade under Low Ca + Co saline, longer pulses of 30 or 60 s, corresponding to the spontaneous depolarization durations of B63 in such a condition, were used.

The amplitudes of B63’s spontaneous and current pulse-evoked plateau-like potentials were measured from the resting (or pre-pulse) membrane potential until the initial level of step-like, maintained depolarization occurring either spontaneously or after termination of a triggering current pulse. The simultaneous voltage variation in a contralateral B31/B32 neuron, which is postsynaptic to a given B63, was measured using the same criteria. The durations of spontaneous or evoked plateau-like potentials in each recorded B63 and its contralateral B31/B32 partner were measured from the time point of B63’s spontaneous step-like depolarization, or from the end of a triggering current pulse in B63, until the neuron’s spontaneous repolarization to resting membrane potential.

The underlying subthreshold oscillations in membrane potential of B63 neurons were analyzed between a bandwidth of 0.00390 to 0.125 Hz by Fast Fourier Transform (FFT) analysis using R software (R Core Team, 2021). Intracellular voltage recordings were first smoothed with a Spike 2 “Smooth” filter with a time constant of 500 ms to suppress action potentials and down-sampled at 1 Hz to decrease the time of computer processing. The peak magnitudes of spectral densities were determined in 400 s excerpts of B63 intracellular recordings. Sinusoidal waveforms corresponding to plateau-triggering oscillations detected on the FFT periodograms were computed and reconstructed from Wavelet decomposition using the R-CRAN package “WaveletComp” (Roesch and Schmidbauer, 2018).

Two-tailed Mann–Whitney U tests were used for unpaired comparisons of two independent groups (U statistic). Comparisons of the proportions of neurons from different preparations expressing evoked plateau-like potentials were made using the two-tailed Fisher’s exact test in unpaired-sample procedures. A Kruskal–Wallis test was used to assess the significance of the overall group differences between >2 independent groups (H statistic). Post hoc pairwise comparisons were performed with a Conover’s test with single-step adjustment (q statistic). All statistical tests were conducted with the R-CRAN “Base” and “PMCMRplus” package (Pohlert, 2019) and any comparison differences were considered significant for a probability level of p < 0.05. Bars in figures represent Mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, n.s., non-significant.

Under normal saline (ASW) conditions, the premotor pacemaker B63 neuron spontaneously expresses endogenous oscillations in membrane potential (Figures 1A–C) that occasionally trigger a large amplitude, plateau-like depolarization and intense burst discharge. This in turn is associated with a plateauing depolarization in B31 neurons and buccal motor pattern (BMP) genesis (Figure 1A). In the condition of chemical synaptic blockade, B63’s membrane potential oscillation and its plateau-like activity persisted, whereas plateauing activity ceased in the B31 neurons (Figures 1D–F). These data therefore suggested that the plateau-like potentials occurring in B63 may arise from membrane property that is intrinsic to pacemaker neuron itself.

In a second set of experiments using isolated buccal ganglion preparations, we asked whether the plateau-like depolarizations in B63 can also be expressed under normal ASW in the absence of corresponding B31/B32 neuron activation. The impetus for these experiments was that activity in the B63 neurons was previously thought to drive BMP generation through a combination of monosynaptic fast cholinergic EPSP production in, and electrical coupling with, the contralateral B31/B32 motor neurons (Figures 2A–C; Hurwitz et al., 1997, 2003). In this scheme, when sufficiently depolarized, the B31/B32 cells express a plateau-like state and a prolonged burst of action potentials via self-excitatory synapses, which in turn sustains B63’s depolarization through their electrical coupling (Figures 2D1, E1, F1; Saada et al., 2009).

With sufficient strength, brief pulses (5 or 8 s) of depolarizing current injected into a recorded B63 neuron could trigger long-lasting plateau-like depolarizations of similar amplitude in both the injected B63 itself and a simultaneously recorded B31 neuron (Figures 2D2, E2, F2). Although single current pulses did not always trigger a plateau-like activity in recorded B63/B31 pairs, presumably due to ongoing or elicited synaptic inhibition within the buccal network, the ability for successful pulses to trigger conjoint B63/B31 plateau-like depolarizations was observed in all tested preparations (5/5). Since the B31 neuron’s depolarization is thought to be evoked by fast cholinergic excitation from B63, the nicotinic antagonist D-tubocurarine (1 mM) was applied to test whether the two neuronal responses could be dissociated. The presence of D-tubocurarine in the bath saline effectively blocked the chemical component of EPSPs in B31/B32 elicited by impulses in B63 (Figure 2C). Consequently, the former’s synaptic response to B63 activation was now strongly reduced (Figures 2D3, E3, F3) and was insufficient to reach threshold for impulse firing, as indicated by the total absence of spikes in B31/32’s axons monitored in the I2 motor nerve (Figure 2D3). In contrast, in all preparations (5/5) the initial injected current pulse still elicited a plateau-like depolarization and prolonged burst firing in B63 itself, both of which outlasted the triggering pulse and with a plateau amplitude that remained unchanged from control (compare Figures 2D2, D3).

This dissociation of the membrane behavior of the B63 and B31/32 neurons was further confirmed by exposure to a Low Ca + Co bath solution to block chemical synapses throughout the buccal network (Figure 2B). In this condition, brief current injection into a B63 again elicited a sustained, large amplitude voltage shift but produced only a weak subthreshold depolarization in a simultaneously recorded B31 neuron (Figures 2D4, E4). Presumably, this residual response resulted from the latter’s electrical coupling with B63 (see Figures 2A, B). It is noteworthy, furthermore, that in Low Ca + Co saline, the mean duration of evoked B63 plateaus was significantly increased compared to the duration observed in standard ASW conditions (Figure 2F4; cf. Figures 2F1, F2). This increase was an expected consequence of the loss of chemical inhibitory synaptic influences throughout the buccal network, such as those associated with the retraction phase of normal BMP genesis.

Altogether, these results are consistent with the idea that although chemical excitation from B63 acting in combination with B31/B32’s autapses may contribute to plateau-like behavior in these two cell types, this synaptic mechanism is not exclusively responsible for B63’s plateauing capability, but rather, it results at least in part from an endogenous membrane property.

Plateau potential generation is the expression of a bistable, all-or-none electrical behavior that allows cells to maintain a lasting depolarized membrane potential before switching back to resting potential, either spontaneously or in response to hyperpolarizing synaptic inputs or negative current injection (Russell and Hartline, 1982; Hartline and Graubard, 1992). The depolarized plateau state of this bistability is typically maintained by persistent inward currents through voltage-dependent cationic channels, while plateau termination is due to voltage-induced closure of these channels and/or to outward currents though calcium-gated potassium channels (Golowasch and Marder, 1992; Russo and Hounsgaard, 1996; Tabak et al., 2011).

Under chemical synaptic blockade with Low Ca + Co exposure, the B63 neurons continued to express such bistable membrane behavior, with a prolonged plateau state that could be triggered by depolarizing current pulse injection and terminated prematurely by hyperpolarizing current injection (Figure 3A). To investigate the contribution of Na⁺ and/or Ca²⁺ ions to the onset and maintenance of B63’s active depolarized state, experiments (n = 5) were conducted in which the sodium channel blocker TTX (1 μM) was added to the Low Ca + Co bathing solution. In all preparations under this condition, injected depolarizing current pulses failed to trigger B63 plateau potentials (Figures 3B–F). In contrast, in a separate group of preparations (n = 5) that were bathed in 0 Ca + Co saline, B63 was still able to express sustained plateau potentials in response to positive current pulse injection (5/5 preparations), and then terminated spontaneously or in response to brief hyperpolarizing current pulses (Figures 3C, D3). On this basis, therefore, extracellular Na⁺ influx through TTX-sensitive membrane channels, but not extracellular Ca²⁺ influx, appears to be necessary for the expression of B63’s endogenous plateau property. Nevertheless, the amplitude of B63 plateaus was significantly reduced under 0 Ca + Co compared to Low Ca + Co, thereby suggesting that, although Ca²⁺ is not necessary for the expression of a mainly Na-dependent mechanism, calcium influx nonetheless contributes partly to B63’s plateau depolarization (Figure 3E).

In normal saline (ASW) conditions, the spontaneous termination of individual plateaus and associated impulse bursts in B63, as in other buccal neurons generating the protractor phase of each BMP, is triggered by an inhibitory synaptic drive from neurons active during the retraction phase of each radula bite cycle (Hurwitz and Susswein, 1996; Sasaki et al., 2013). Under Low Ca + Co or 0 Ca + Co solutions, which suppress these inhibitory influences, B63 expressed greatly extended plateau potentials that eventually terminated spontaneously (Figures 3D1, D3, E, F). This step-change cessation could also be prematurely induced by the intracellular injection of a hyperpolarizing current (see Figures 3A, C), thereby suggesting the contribution of a voltage-dependent mechanism.

To test for a probable contribution of potassium channels in B63’s plateau termination, preparations were superfused with or without the potassium channel blocker TEA (50 mM) in Low Ca + Co saline. Under this condition, plateaus were still able to be evoked in 5/5 preparations. In the presence of TEA (n = 5), as compared to in the blocker’s absence (n = 5), the durations of B63’s plateau potentials were considerably and significantly prolonged (Figure 3D2, compare with Figures 3D1, E). Plateau durations were also considerably modified by changes in the concentration of calcium ions, with plateaus lasting several tens of seconds in the presence of Ca²⁺ (3 mM in the Low Ca + Co solution, n = 5), but were maintained for hundreds of seconds in the cation’s absence (0 Ca + Co solution, n = 5) (Figure 3D3, compare with Figures 3D1, F). These findings thus lead to the conclusion that a voltage-dependent potassium current regulated by calcium contribute to B63’s plateau termination.

In normal ASW conditions, plateau potentials in the B63 neuron are triggered spontaneously by an ongoing low amplitude oscillation in the interneuron’s membrane potential (see Figure 1) that arises from organelle-derived fluxes in intracellular calcium (Bédécarrats et al., 2021). Both the plateaus (as described above) and their underlying voltage oscillation are blocked by TTX, indicating in each case, a contribution of sodium and/or non-selective cation membrane channels (see Figure 3B and Bédécarrats et al., 2021). To determine whether the same or different membrane channels are implicated in the expression of these endogenous properties, the specific effects of two different pharmacological agents was tested. Firstly, the membrane potential oscillation of B63 was suppressed by the presence of reticulum endoplasmic calcium pump inhibitor CPA (20 μM in a Low Ca + Co solution; Figures 4A1, D, E; see also Bédécarrats et al., 2021), and consequently, also prevented the spontaneous expression of plateau potentials (n = 8; Figure 4A1, compare with Figure 4C1). Nonetheless, in the majority of these preparations (6/8), B63 neurons were still able to produce plateau potentials in response to depolarizing current injection (Figure 4A2, compare with Figures 4C2, F, G). This proportion of preparations in which B63 still expressed current pulse-evoked plateaus was not significantly different from that found in ASW (5/5, p = 0.487), in Low Ca + Co (5/5, p = 0.487) or in Low Ca + Co + DMSO (8/8, p = 0.467) thereby indicating that CPA did not in fact suppress this membrane property. In contrast, the non-selective cation channel blocker FFA (0.1 mM in a Low Ca + Co solution, n = 8) did not block B63’s spontaneous membrane oscillation (Figures 4B1, D, E), which led to action potential bursts, but solely with short durations, on nearly every cycle. Moreover, in almost all of these tested preparations (7/8), experimental depolarization by current pulse injection, also failed to trigger any sustained plateau-like potentials (Figures 4B2, D). This proportion of preparations was now significantly different from that in ASW (0/5, p = 0.003), in Low Ca + Co (0/5, p = 0.003) or in Low Ca + Co + DMSO (0/8, p = 0.0). In addition, quantitative comparisons between the amplitude and cycle period of B63’s spontaneous oscillations and of the amplitude and duration of evoked B63 plateaus further indicated that these membrane properties are differently affected by FFA and CPA (Figures 4D–G). Therefore, although the B63 neuron’s oscillatory and plateauing capability both depend on TTX-sensitive membrane channels, the above pharmacological results indicated the involvement of two distinct types of sodium channels.