Characterization of Lymnaea Neurones by determination of Action Potential Trajectories

The ability to identify mollusc neurones from which electrical recordings can be made has proved useful in biophysical (e.g. Aldrich, Getting & Thompson, 1979), neuropharmacological (e.g. Walker & Kerkut, 1977) and neuroethological studies (e.g. Kupferman, Carew & Kandel, 1974), but definite identification of a particular neurone may prove difficult, particularly if it is surrounded by others of a similar size or colour. Walker et al. (1970) have demonstrated that action potential shape can be useful in the identification of neurones of Helix pomatia and Benjamin & Winlow (1981) classified Lymnaea stagnalis neurones as having type 1 or type 2 action potentials according to the respective absence or presence of a pseudoplateau during repolarisation.

Here we demonstrate that phase plane displays provide a simple and compact method for identifying mollusc neurones by analysis of the properties of the action potential. In this technique the rate of change of action potential voltage is displayed against action potential voltage. The phase plane technique has been used to study action potentials of vertebrate skeletal muscle (Jenerick, 1961, 1963, 1964) and cardiac muscle (Sperelakis & Shumaker, 1968) as well as those of molluscan neurones (Roberge et al. 1977; Pinsker & Bell, 1981). We have shown elsewhere (Haydon, Winlow & Holden, 1982; Holden, Winlow & Haydon, 1982) that the technique may also be used to determine the effects of pharmacological agents on nerve cells.

Specimens of Lymnaea stagnalis weighing 1·4–2·4 g were obtained from animal suppliers or the Leeds-Liverpool canal, kept in tap water at room temperature (20 °C) and fed on lettuce. Brains were removed and prepared according to the methods of Winlow & Benjamin (1976) and superfused in aerated HEPES buffered saline (Benjamin & Winlow, 1981) at 20 °C. Individual cells were penetrated with glass microelectrodes filled with the supernatant from a saturated solution of K₂SO₄ and with resistances of 15 ·8±1·4 MΩ (mean±s.E.). A WPI M701 microprobe system incorporating an active bridge circuit was used both for recording signals and passing current via a single electrode. The current monitor of the amplifier was checked using an active current-voltage convertor on a virtual earth. Signals were conventionally displayed and recorded and also displayed as the rate of change of voltage with respect to time, dV/dt, against V.After electrode penetration we allowed a minimum of 5 min for each cell to settle before recordings were made. During constant current injection into cells, recordings were only made after the cells had reached a steady state of discharge, usually within 30 s to 2 min after the onset of the current.

A periodic, repetitive discharge V(t) generates a closed orbit in the phase-plane dV/dt against V, and so provides a compact description of the voltage trajectory. A disadvantage of examining V(t) at a fast sweep speed to describe action potential shape is that changes in the depolarizing phase of the action potential can be lost and can alter the phase at which the oscilloscope is triggered. However, the orbit in the phase-plane contains the entire trajectory.

dV/dt was obtained by electronic differentiation, using an active circuit with a cut-off frequency of 10 kHz. Differentiation enhances any noise components present in V(t). To reduce degradation of the signal moise ratio, K(t) was filtered by an active 4th order Butterworth filter with a corner frequency of 2·5 kHz. Any filter will introduce frequency-dependent phase changes: such phase lags will produce a counterclockwise rotation of the dV/dt — V(t) display. The cut-off frequency of the filter was sufficiently high for there to be no distortion in the phase-plane display (Fig. la-c).

Since the cascade of filter and differentiator contained an odd number of operational amplifiers, the resulting output was proportional to —dV/dt. Thus in the phase-plane displays the trajectory orbits in the counter-clockwise direction, and the rate of depolarization is in the downward direction. In an isopotential cell the ionic current is proportional to dV/dt, so for isopotential cells inward (depolarizing) current is downward and outward (hyperpolarizing) current is upward in our phaseplane.

Unless otherwise stated, all superimposed sweeps of action potentials, or their phase-planes, contain ten superimpositions. On all graphs, results are expressed as mean + s.E. of mean.

The precise locations and nomenclature of the neurones discussed in this paper are described elsewhere (Winlow & Benjamin, 1976; Benjamin & Winlow, 1981; Slade, Mills & Winlow, 1981).

Differences between type 1 (Fig. 1d) and type 2 (Fig. 1f) action potentials (Benjamin & Winlow, 1981) are obvious even in the absence of their phase-plane portraits (Fig. 1e,g). However within this basic classification considerable differences in action potential trajectory are visible (Fig. 2). For example the superimposed spontaneous action potentials of the giant cell L.Pe.D. 4 (Fig. 2f) and a R.Pe.C. cluster cell (Fig. 2g) appear very similar in their superimposed traces. However their phase plane portraits immediately indicate that the maximum rates of depolarization and repolarization of L.Pe.D. 4 are considerably greater than those of the R.Pe.C. cluster cell. Furthermore, type 1 cells such as L.Pe.D. 3 and L.Pe.D. 7 (Fig. 2a, b) usually have a higher than type 2 cells (Fig. 2c-i). Thus different cells have characteristic phase-plane trajectories and these are consistent with only minor variations, from preparation to preparation (Fig. 3).

Although the phase-plane portrait for any given cell is quite characteristic it may be altered according to whether the cell is endogenously active or driven by synaptic input. Bursts of spikes in the giant neurone R.Pe.D. 1 are capable of initiating a powerful synaptic input (input 3 of Benjamin & Winlow, 1981) which in turn reexcites R.Pe.D. i itself (Fig. 4 and Winlow, Haydon & Benjamin, 1981). Action potentials induced by input 3 show a significant decrease in (Fig. 4b) as compared with endogenous action potentials (Fig. 4a) or action potentials immediately following input 3 (Fig. 4c). Furthermore there is a pronounced prepotential during input 3 induced spikes and this is recognizable as a hump preceding ⁠. This prepotential is probably an axon spike (see P. G. Haydon & W. Winlow, in preparation) which electrotonically invades the soma and triggers the soma spike.

Hyperpolarizing currents will increase both and and cause narrowing of action potentials whilst depolarizing currents have the reverse effects. In the example shown in Fig. 5 the spontaneous discharge of an R.Pe.A cluster cell was first recorded (Fig. 5a) and the cell was then silenced by a 0·8 nA hyperpolarizing step lasting for 70 s. The hyperpolarizing current was then reduced to — 0·4 nA and action potentials resumed (Fig. 5b). On return to o nA injected current (Fig. 5c) there appeared to be a reduction in and as compared with the resting discharge in Fig. 5,a. Since the discharge rate in Fig. 5 a was 4 spikes s⁻¹ we then hyperpolarized the cell by 0·2 nA to give a discharge rate of 3·8 spikes s⁻¹(Fig. 5,d), at which the phase plane portrait closely resembled that in Fig. 5a. Similar processes occurred to a greater or lesser extent in all the cell types investigated and this suggests that modifications of and are determined by spike rate, which is itself determined by membrane potential and the previous history of the neurone.

Many authors have shown that molluscan action potentials broaden as spike frequency increases (e.g. Strumwasser, 1967; Aldrich et al. 1979; van Swigchem, 1979; Gillette, Gillette & Davis, 1980). We therefore determined the alteration of and (for a variety of neurones) that could be induced by gradually increasing the amount of current injected into the cell via the recording electrode. We adopted the approach of first hyperpolarizing the investigated cell through several steps to silence unless the cell was naturally silent. We then gradually decreased the hyperpolarizing current in a stepwise manner and recorded steady state action potential trajectories at each step, eventually passing through zero nA and then continuing the procedure with stepwise depolarizing pulses. Examples of action potential trajectories at steady rates of discharge are illustrated for the type 1 action potential of a visceral K cell (Fig. 6) and for the type 2 action potential of R.Pe.D. 1 (Fig. 7). For each of the cell types investigated there is a characteristic relationship between injected current, and (Fig. 8) and (Fig. 9) and these relationships can be used to discriminate between cell types.

For type 1 action potentials, and are proportional to maximal net inward and outward current respectively. However for type 2 cells it is necessary to subdivide into and (Fig. 7,a). Thus, is proportional to the net outward current during the first phase of repolarization prior to the pseudoplateau whilst is proportional to the net outward current during the second phase of repolarization following the pseudoplateau. For cells with type 1 action potentials the curve relating both and to injected current is characteristically steep (Figs. 8a and 9a). However the curve for could be confused with that for the type 2 cells of the L.R.Pe.A clusters (Fig. 8b) were it not that the differences between type 1 and type 2 action potentials are so marked (Fig. 2).

In any case comparison of for the visceral HIJK cells (Fig. 9a) with and for the L./R.Pe.A clusters indicates that for the type 1 cells there is a characteristically greater rate of outward current movement than for type 2 cells. Lying within the L./R.Pe.A clusters there are three pairs of giant cells, L./R.Pe.D. 4, L./R.Pe.D. 8 and L./R.Pe.V. 2. These giant cells have a similar appearance and colour to the L./R.Pe.A. cluster cells surrounding them and have similar action potentials to one another (Fig. 2d,f), both in terms of action potential shape and trajectory. Their responses to injected current are shown in Figures 8c and 9c and are very similar to the responses of the L./R.Pe.A clusters (Fig. 8 b and 9b). Their curve for is shifted both up and to the right relative to the L./R.Pe.A. clusters and the range over which current injection is possible is much greater for these giant cells than for smaller cells of the surrounding L./R.Pe.A. clusters. This latter point indicates that cell size is a very important determinant of the response of a cell to injected current, but the fact that the curve for is shifted upwards implies that there is also a greater inward current density in these larger cells. Differences in and for these cell types are much less marked than for ⁠.

Further evidence of the importance of inward and outward current density in determining spike trajectories is provided by the giant cells R.Pe.D. 1, VV1/2 and VD1 (Figs. 8d-f and 9d-f)These cells are of similar sizes to one another (Winlow & Benjamin, 1976; Benjamin & Winlow, 1981), but their curves for ⁠, and in response to injected current differ markedly. Of the three cells VDi is particularly interesting since it is never silent unless hyperpolarized and quite strong hyperpolarization is often necessary to silence it. Perhaps this is due to its strong electrical connection to another giant neurone, RPD 2(Benjamin & Winlow, 1981; Soffe & Benjamin, 1980).

Phase plane portraits of action potentials for any given neurone are quite characteristic for that cell provided that the cell’s recent past history and the amount of current injected into it are known. The most marked differences between neurones (particularly those with type 2 action potentials) are shown by plotting against injected current, as summarized for the cells in this study in Figure 10. When neurones cannot be easily identified by morphological criteria, then phase plane techniques, and particularly plots of against injected current, can provide a valuable means of physiological identification.

The net inward current is the difference between the inward and outward ionic currents, and so would be proportional to the inward ionic currents only if the outward ionic currents were negligible during the upswing of the action potential. This has shown to be the case in numerical solutions of the Beeler & Reuter (1977) equations for the mammalian ventricular action potential (Hondgehem, 1978), but it is not the case for the Hodgkin-Huxley (1952) equations for the squid axonal action potential (Strichartz & Cohen, 1978). A full voltage clamp analysis for the neurones examined in this paper is not available, but from the voltage clamp analyses of other molluscan neurones (see Adams, Smith & Thompson, 1980 for a review) it is likely that the outward K⁺-currents are small but not negligible, and so is not proportional to the maximal inward ionic currents. Further, the inward ionic currents are carried by two independent systems, a faster Na⁺-selective and a slower Ca²⁺-selective system. Thus is not proportional to the maximal inward Na⁺ current.

An ion-selective current is the product of the maximal conductance, functions of voltage-dependent activation and inactivation variables, and the electrochemical gradient: thus even if was proportional to the maximum inward Na+-current it would not provide a measure of the maximal Na+-conductance, or Na⁺-channel density.

Thus is not a measure of the Na+-channel density, and and are not measures of the channel densities of the main repolarization conductances. However, and reflect membrane rather than geometric properties: there is no correlation between and cell size (Table 1). The simplest explanation for the range of and seen in different neurones is that the different neurones have characteristically different densities of their ion-selective channels, with larger and being associated with higher channel densities. The effects of channel density have been reviewed by Holden & Yoda (1981). The activity dependent changes in and are due to changes in the voltage dependent activation and inactivation variables. Since and for a given neurone type are tightly distributed about their mean values at any level of activity (see Figs. 8 and 9), the somatic ionic channel densities of a neurone are regulated about values that are different for different neurones.

There is only a single peak for during type 1 action potentials suggesting either that only a single conductance acts during repolarization or that if several outward currents exist they act in synchrony. For type 2 action potentials the situation is markedly different and the two peaks designated and (Fig 7a) exist in their phase planes. We presume that the outward current leading to repolarization is driven by the potassium selective current system, I_K(Adams, et al. 1980) and that the peaks and are caused by interruption of repolarization by the inward calcium current I_Ca(Meech, 1978) which may in turn, activate the calcium dependent potassium current, I_ca(Meech, 1978; Thompson, 1977). At least two potassium currents may act during repolarization and both may participate in determining the value of and ⁠. Since both and decline in parallel in response to increasing levels of injected current (Fig.9b-f) either value is characteristic of a given neurone.

Action potential plasticity, such as we have observed here, implies that when neurones are using action potentials as signals, rather than the local non-propagative potentials of spikeless neurones (see Pearson, 1976; Burrows & Siegler, 1976), all action potentials are not necessarily equivalent. Indeed, ‘neuronal integration must be considered to be a deeply state-dependent property’ (Llinas, Yarom & Sugimori, 1981) of the nervous systems of both vertebrates and invertebrates and the discovery of action potential plasticity based on modification of ionic conductances adds another level of complexity to our knowledge of the cellular mechanisms underlying animal behaviour.

Broadening of action potentials as impulse frequency increases is a well known phenomenon in molluscs (e.g. Strumwasser, 1967; Kandel, 1978; van Swigchem, 1979; Aldrich et al. 1979; Benjamin & Winlow, 1981). Similar broadening of action potentials can also be produced either by extracellular or intracellular application of TEA (e.g. Koketsu, Cerf & Nishi, 1959; Bryant & Weinreich, 1975; Schwartzkroin & Prince, 1980; Holden, Winlow & Haydon, 1982). One property of action potentials whose duration is artificially increased with TEA is that the amount of transmitter released from their axon terminals is greatly increased (Bryant & Weinreich, 1975; Berry & Pentreath, 1976; Kandel, 1976; Winlow, Haydon & Benjamin, 1981). This implies that axon terminal spikes are broadened as well as soma spikes, which remains to be directly demonstrated. Thus the plasticity in action potential trajectory shown above may have important behavioural consequences. For example strong excitatory synaptic inputs increase spike frequency and cause spike broadening, whilst strong inhibitory inputs may prevent neurones from firing for a period and cause narrowing of subsequent action potentials (W. Winlow & A. V. Holden, submitted for publication). Thus, following strong excitation, spikes are broadened and transmitter release could be enhanced leading to augmentation of a behavioural sequence such a4 occurs in the feeding system of Pleurobranchea(Gillette et al. 1980). Broadened spikes could also deplete transmitter stores, as occurs with high frequency discharges in cat sympathetic ganglia (Pysh & Wiley, 1974) and this could lead to habituation of a response. Following periods of strong inhibition post-inhibitory rebound excitation (PIR) often occurs and has been shown to be the basis of oscillatory discharges in Aplysia (Kandel, Frazier & Wachtel, 1969). During PIR, spikes rapidly broaden (W. Winlow, unpublished observations) and could thus enhance transmitter release as spike frequency starts to decline. Thus the plasticity of action potential trajectories that we have demonstrated here may influence the output of neuronal networks.

We thank Miss Heather Davies for her expert technical assistance.