The transmission of impulses in the ectodermal slow conduction system of the sea anemone Calliactis parasitica (Couch)

Recent electrophysiological investigations into the control of behaviour in sea anemones (McFarlane, 1969a, b) have revealed the presence of two conduction systems in addition to the nerve net. These ‘slow conduction’ systems, the SS1 and the SS2 are situated in the ectoderm and endoderm respectively, though unequivocal evidence that the epithelial cells themselves are directly involved in the transmission of excitation has not yet been obtained. Neuroid systems have been positively identified in a number of Hydrozoans (Mackie, 1965; Mackie & Mackie, 1967; Mackie & Passano, 1968) and in the epithelia of developing toad tadpoles (Roberts, 1969; Roberts & Stirling, 1971). More recently, Spencer (1974) has suggested that the colonial pulse system in the Hydrozoan Proboscidactyla flavicirrata may also be neuroid. Mackie (1970) gives a review of known neuroid systems.

The sea anemone slow systems have been shown to have an important role in many aspects of the control of actinian behaviour. For example, the SS1 is involved during pedal disc detachment in Calliactis parasitica (McFarlane, 1969b) and Adamsia palliata (Shelton & McFarlane, unpublished). In Tealia felina the SS1 responds to dissolved food extract applied to the column of the anemone (McFarlane & Lawn, 1972) and is active during preparatory feeding behaviour, causing expansion of the oral disc and lowering of the oral disc margin (McFarlane, 1970). This effect is brought about in part by the action of the SS1 on the oral disc radial muscles (ectodermal) leading to inhibition of spontaneous contractions and an increase in length of the muscles (McFarlane & Lawn, 1972). The SS2 can also have an inhibitory effect, in this case on certain endodermal muscles, and interacts with the nerve net in the control of circular and parietal muscle contraction cycles (McFarlane, 1974a).

A system showing many of the properties of the sea anemone slow systems has recently been identified in the Octocorallian Pennatula phosphorea (Shelton, 1975), and it would not be surprising to find that the majority of the Anthozoa possess similar systems. Despite the probable widespread occurrence of such control pathways in the Anthozoa and their fundamental importance to the regulation of behaviour in these animals, we know very little about the nature of ‘slow conduction’ or its morphological basis, although the suggestion has been made that it could be neuroid (McFarlane, 1969b).

This paper describes a series of experiments, carried out on the sea anemone Calliactis parasitica (Couch), designed to give some insight into the conduction of SS1 pulses. A computer model based on the experimental evidence is described and analysed.

Specimens of Calliactis parasitica were obtained from the Marine Laboratory, Plymouth and anemones with an expanded oral disc diameter of 3–5 cm were used. The animals were maintained in running sea water aquaria at 10–15 °C. Extracellular, polythene suction electrodes (after Josephson, 1965) with tip diameters of 100–300 μm were used to make recordings of electrical activity. Similar electrodes were employed to administer electrical stimuli. Recorded activity was amplified using differential pre-amplifiers and displayed on a Tektronix Type 564 storage oscilloscope.

Recordings were made from a variety of preparations-whole or half-animals and a range of strips of column of different widths. The strips were cut to include part of the pedal disc and part of the oral disc and pinned out flat (ectodermal surface upper-most) on a piece of cork submerged in the recording bath. Preparations were left over-night before use to recover from the operation. Although the largest SS1 pulses may be recorded from the tentacles (McFarlane, 1969b) all recordings were made from the sphincter region of the column in order to ensure that the population of conducting cells was as uniform as possible. Stimulation of the SS1 was made through a shallow flap cut in the column ectoderm. This ensured that only the SS1 was excited by the stimuli (McFarlane, 1970) but repeat experiments stimulating the intact column showed that stimulation via a flap did not affect the rate or the manner in which the SS1 fatigued.

The effect of repetitive stimulation of the SS1 in Calliactis (McFarlane, 1973 a) and of Tealia felina (McFarlane & Lawn, 1972) is an increase in the delay between the delivery of the stimulus and the arrival of the pulse at the recording electrode. This is coupled with a reduction in the amplitude of the evoked SS1 pulse. A similar effect is observed following repetitive stimulation of the SS2 in Calliactis (McFarlane, 1973a) The experiments were carried out by stimulating the intact column and recording from a tentacle and the effect was most pronounced at relatively high stimulus frequencies (once every three seconds). These results have been confirmed using flap stimulation and recording from the sphincter, though it should be noted that the SS1 conducts more rapidly across the oral disc than up the column (McFarlane, 1969b). This may be related to the greater diameter of the oral disc ectoderm cells.

Following repetitive electrical stimulation of an ectodermal flap, a marked increase in SS1 pulse conduction delay occurred (Fig. 1 and Fig. 2,a). The build up of fatigue in the system was marked by the failure of occasional SS1 pulses to reach the recording electrode. Finally the whole system became refractory and would not conduct further SS1 pulses until it had been allowed to recover. Full recovery took a surprisingly long time (Fig. 2 b) (see also McFarlane, 1969a). At 12 °C, the effects of a single SS1 pulse were detectable up to 40 sec afterwards. These effects were manifested by means of their delaying action on subsequent SS1 pulses. The recovery occurred most rapidly, however, in the first 5 sec following the transmission of an SS1 pulse. Recovery from total failure following a long train of stimuli could take several minutes. At lower temperatures, the delaying effect was even more long-lasting, and as a consequence refractoriness and failure occurred after fewer SS1 pulses had been conducted at any given frequency of stimulation. The conduction delay of an SS1 pulse was often doubled before the final failure of the system.

Concomitant with the reduction of conduction velocity of successive SS1 pulses, there was also a reduction in the amplitude of recorded activity (Fig. 3,a). Note the variability in the amplitude when the system was fatigued. Fig. 3,b shows the relationship between the conduction delay and the size of the pulses recorded. With increasing fatigue, the duration of evoked SS1 pulses became very variable (Fig. 4).

The use of two recording electrodes provided further data on the nature of fatigue in the SS1. Fig. 5 a shows an interesting and atypical case. In this experiment, there was a sudden jump in the conduction delay on channel 2 to a new and much higher value; after eleven pulses had conducted at this lower rate the conduction velocity returned to a value comparable to that of channel 1. This was quickly followed by total failure of the system.

The next series of experiments was designed to investigate the effect of the width of the conducting tissue on its ability to conduct long trains of SS1 pulses. Every effort was made to ensure that the experimental conditions were as uniform as possible for every experiment. The temperature was maintained at 14 °C; the same electrodes were used in each case; all stimuli were applied at a frequency of 1 per 3·2 seconds; and the distance between stimulating and recording electrodes was kept as nearly uniform as possible for every preparation. Fig. 5,b shows clearly that in a narrow strip, conduction failure occurred much more quickly than in a wider strip cut from the same animal. In this case a 33% reduction in width caused conduction failure after 27% fewer SS1 pulses had been conducted. Note that the rate of increase in conduction delay was not greater in the narrow strip. The two curves are nearly superimposed upon one another until conduction failed in the narrow strip. Another point of interest is the upturn of the curve for delay in the final stages of fatigue. Compare Fig. 5,b (curves for narrow strips) with Fig. 2 a (curves for half-animal preparation).

Electron microscopical evidence (McFarlane, 1969a; Shelton, unpublished) suggests that ectodermal cells in Calliactis parasitica are columnar structures up to several hundred μm long with a diameter of about 4–6 μm. Certain of the epithelial cells bear cilia and may have a sensory function. Desmosomal junctions (2–3 μm in length) join the cells together in a roughly hexagonal array to form a sheet of epithelium one cell thick. The external surface of the epithelium is covered with a border of microvilli. Half-desmosomes connect the epithelium with the collagenous matrix of the mesogloea. Membrane specializations have been observed between adjacent cells and these may be associated with the transmission of electrical excitation between ectoderm cells. Typically, these membrane specializations consist of a region (about 350 nm long) where the membranes are a fixed distance apart (about 25 nm). On each side of this, the intercellular space is occluded with electron-dense material. This may act as a seal to minimize current leakage.

In an unfatigued preparation, SS1 pulses conduct up the column at about 6 cm sec⁻¹. During an experiment in which a half-animal was stimulated on a flap at the base of the column and a recording made 2·5 cm away over the sphincter muscle, there was a delay between stimulus and response of 440 msec.

By combining the electrophysiological and electron microscopical evidence given above, it is possible to calculate the amount of delay as an SS1 pulse crosses each cell boundary, assuming that :

• (a) the ectodermal cells truly represent the conducting units of the SS1;

• (b) in the unfatigued SS1, conduction of excitation between the stimulating electrode and the recording electrode takes place along the shortest route;

• (c) the delay in conducting across an ectodermal cell and the delay in generating the response beneath the stimulating electrode is negligible. This calculation was made and a value of 0·088–0·1 msec was obtained for the junctional delay. This compares with an average value for the delay of conduction across crayfish giant motor synapses (believed to be electrotonically coupled) of 0·12 msec (Furshpan & Potter, 1959).

This paper has provided data which enable a more detailed analysis of the process of slow conduction to be made. The small size of the ectodermal cells and the large amount of mucus secreted have so far prevented successful intracellular recording. Though even this would not be conclusive, it would be the best direct evidence for ectoderm cells as the basis for the SS1.

If the SS1 is a neuroid system, the value obtained for the theoretical delay as SS1 pulses conduct between ectoderm cells in an unfatigued system is consistent with there being electrical coupling between the cells. This is supported by the limited electron microscopical evidence showing possible low-resistance junctions between the cells. The alternative to a purely neuroid SS1 is a system involving nervous elements in a net distinct from the well-documented nerve net described by Pantin (1935a, b, c, d) and more recently by McFarlane (1969a, 1973a,b, c; 1974a, b). In favour of the nervous nature of the SS1, it should be noted that there are distinct similarities between the properties of the nerve net and the slow system in certain octocorals (Shelton, 1975), but this may merely reflect similarities in the lay-out and mode of action of the two systems. Horridge (1956), using histological methods, showed that there were two nerve nets in ephyra larvae of Aurelia, and Passano (1965, 1973) has been able to demonstrate two different kinds of electrical activity from Cassiopea. As yet, however, there is no published evidence for the presence of nerve net cells in the column ectoderm of Calliactis, though this species has been intensively studied. The ‘classical’ nerve net (Pantin, 1935) is endodermal in the column. The weight of evidence so far accumulated tends to favour the idea that SS1 pulses are conducted between ectoderm cells.

As a further test of the suggestion that the SS1 is a neuroid system, it was decided to construct a computer model which displayed the experimentally observed properties in an attempt at gaining a greater insight into the conduction mechanism. The assumption was made that the system was made up of a flat sheet of discrete units each connected to six nearest neighbours in a hexagonal array. Increases in the conduction delay of pulses between two fixed points in the system might then be explained in one or both of two ways. Firstly, conduction might always be by the shortest route but with progressively increasing junctional delays as the system fatigued; secondly, it could be that in the fatigued system, the conduction path becomes longer and more tortuous and thus a pulse must cross more junctions (Shelton, unpublished thesis). The experiments with strips of column suggest that the latter explanation may be an important factor. While the system is unfatigued and conduction occurs via pathways which do not deviate markedly from straight lines, one would expect conduction delays in both wide and narrow strips to be closely similar. As fatigue built up, however, one would reach a point when the only pathways available are so winding that, with narrow strips, excitation is lost over the edge of the strip (Fig. 6).

It is postulated that there is a population of conducting units with a range of thresholds. In the unfatigued state, all the conducting units have a low threshold, but with repetitive stimulation the population of units with a higher threshold increases so that in any given area there will be a lower number of spiking cells. An unbroken chain of spiking units between any two points would thus be thinner and more tortuous than in the unfatigued state. The low density of spiking cells would further account for the observed reduction in the amplitude of the pulses recorded. Note that with a recording electrode tip of 200 μm diameter, there will be about 1600 ectoderm cells beneath the recording electrode. The SS1 pulses thus represent activity in a large number of cells and a reduction in the number of excitable cells would lead to a reduction in the amplitude of the pulse recorded.

The computer model sought to test the hypothesis outlined above. For the purposes of computation, the problem was reduced to its simplest form. An array of units (not necessarily single cells) each connected to its six nearest neighbours was set up. This corresponds to the observed packing of the ectoderm cells. A random population was given a high threshold (inexcitable and therefore non-spiking) and the computer instructed to find the fastest pathways for the transmission of excitation through the maze of excitable and inexcitable units. The percentage of randomly ‘fatigued’ units could be changed to test whether a high population of such units led to a significantly larger number of steps being taken between any two points in the array. To test such a model by hand would have taken an extremely long time. The results of this operation on the model proved most interesting. There was only a very narrow range (between 50 and 60 % excitable units) separating total rapid through-conduction from almost total failure to conduct with the size array which was used (80 units by 80 units). It would seem, therefore, that such a simple model is inadequate to explain the observed experimental data. Nevertheless, the model did show many of the right properties (Fig. 7). With a high proportion of inexcitable units, very long and tortuous pathways were formed and it was frequently observed that the number of steps taken to reach units close together in the array could be widely different. In other words, when the system was fatigued two things would happen:

• (a) there would be a relatively small number of spiking units in any given area;

• (b) the duration of spiking activity in that area would be variable but longer than in the non-fatigued condition. In terms of recordable activity, this would lead to SS1 pulses of small amplitude and long and variable duration. This corresponds with the observed response under such conditions. In the case of a high proportion of excitable cells, the model responded with a large number of spiking units in any given area, with all the spiking taking place over a much shorter time period, i.e. SS1 pulses of larger amplitude and shorter, less variable duration. This again is confirmed by experimental data from the living animal. The experiments using strips of column of different widths suggest that, in the non-fatigued condition, pathways of excitation should be direct, and in the fatigued condition pathways of excitation should be long and winding. The model behaves like this. As it stands, however, it suffers from the fault that it will not give a smooth progression in the lengths of conduction paths in response to progressive increases in the number of excitable cells. It will only work in this manner over an extremely narrow range.

So far, only the spiking units have been considered, but it has been previously suggested in this paper that the ectodermal cells are all electrically coupled together via low-resistance junctions. If this is the case, electrotonic current flow will occur even between non-spiking cells. The amount of electrotonic spread required to depolarize a cell surrounded by high threshold cells is unknown; we have no information on the amount of current leakage which occurs between cells. Work on autonomic smooth muscle (Bennett, 1972, 1973) suggests that in order to evoke propagated spiking activity, an area of tissue (consisting of a number of cells) must be depolarized. The results reported here have shown that as the SS1 fatigues, the properties of the system change from those of a sheet of conducting tissue to those of a reticular system. The model considered ‘units’ rather than cells and it may be that groups of cells make up each unit. Fig. 7 showed that under fatigued conditions some units with low thresholds were not spiking because there was no direct connexion to another spiking unit. Electrotonic current flow could induce spiking even though a low threshold unit was not directly coupled to another spiking unit. This would give the model the remaining property which it lacks, namely a smooth progression in the length of conducting pathways over a wider range of ‘percentage of excitable units’.

Differences in threshold may be related to the time taken to repolarize following a spike. It has been shown that the effects of the transmission of an SS1 pulse are still detectable many seconds after the passage of the pulse. In myoepithelial cells of the Siphonophore Nanomia bijuga, Spencer (1971) quotes a repolarization time of up to three seconds.

The model has been described in terms of a neuroid epithelium; it could have properties not so far experimentally demonstrated for the SS1 For instance, under certain conditions it could show facilitation of the rate of conduction. The slow system in Pennatula phosphorea does exhibit this property (Shelton, 1975). It is interesting to note that when the model is ‘fatigued’ it behaves as a local system. This could have important consequences in the animal and may help to explain certain aspects of pedal disc detachment which is controlled by the SS1 (McFarlane, 1969c). The model was designed to describe conduction in the SS1; the SS2 may well work in a similar way.

A type of fatigue similar to that reported here for the SS1 has recently been shown in the colonial pulse system (possibly neuroid) of the Hydrozoan Proboscidactyla flavicirrata (Spencer, 1974). There are also parallels in the fatigue shown by chick cardiac muscle (Lieberman et al. 1973).

McFarlane & Lawn (1972) have suggested that if the SS1 is a neuroid system, the ionic changes associated with the transmission of a pulse could affect ectodermal muscles lying in close proximity to spiking ectodermal cells. Generally speaking, the changes in ion concentration associated with a spiking nerve are exceedingly small and localized, but it may be that, with a ‘blanket’ of depolarization such as that envisaged, changes in the molar concentrations of ions could be significant. There could also be an interaction, similar to the one they have suggested, between the SS 2 and endodermal muscles. In the oral disc of Mimetridium cryptum, Batham (1965) has shown that the nerve net is ectodermal. The Hertwigs (1879, 1880) reported similar findings for Sagartia (Calliactis) parasitica. We know that the SS2 can interact with the nerve net (McFarlane, 1974a, b) and it may be that the SS1 and the nerve net are similarly linked perhaps in the region of the oral disc where the two conducting systems are in very close juxtaposition. This possibility should be investigated.

It is a pleasure to thank Dr I. D. McFarlane for his advice and help. Mr P. R. Balch very kindly wrote the computer programme. Dr J. L. S. Cobb made many useful suggestions and criticized the manuscript. This work was undertaken during the tenure of an S.R.C. Research Studentship.