Ionic Composition of Haemolymph and Nervous Function in the Cockroach, Periplaneta Americana L

The ionic composition of the haemolymph of the American cockroach, Periplaneta americana, has been fully studied during the past 20 years (Tobias, 1948; Van Asperen & Van Esch, 1956; Treherne, 1961; Pichon, 1963; Pichon & Boistel, 1963a, b; Brady, 1967 a, b). It has been shown that whereas the sodium concentration is higher than that of potassium, thus resembling the blood of vertebrates, the potassium concentration may vary widely; the potassium level of the haemolymph increases, for example, when the insect is fed on a leaf diet (Tobias, 1948; Pichon & Boistel, 1963 b) or submitted to prolonged starvation and dehydratation (Pichon & Boistel, 1963 a) and may reach or even exceed 50 mM/1. Moreover, extremely large variations of the Na/K ratio have been recorded between different insects reared in apparently identical conditions (Pichon, 1963).

It is reasonable to suppose that such variations may appreciably modify the excitability of the neurones. The effects of inorganic ions on insect nerve fibres have been indeed thoroughly investigated using the microelectrode technique (see Yamasaki & Narahashi, 1959; Boistel, 1960; Narahashi & Yamasaki, 1960; Narahashi, 1966) and it has been concluded that the sodium theory was essentially valid for the giant axons of the cockroach. According to the curves given by Yamasaki & Narahashi (1959) the giant axons would be strongly depolarized in potassium concentrations such as those which have been measured in the haemolymph. Roeder (1948) found, however, that spontaneous activity and synaptic transmission in the isolated nerve cord of the cockroach were not immediately modified by variations in potassium concentrations in the medium between o and 50 mM/1. K⁺. More recently, we have shown that perfusion of the last abdominal ganglion of the cockroach with a saline containing 52 mM/1. K⁺ induced a transient increase in the ‘spontaneous’ activity recorded at the level of the abdominal nerve cord; this increase was followed by a blockage. The delay between the beginning of the perfusion and the maximum of activity is greatly reduced in de-sheathed preparations (Pichon & Boistel, 1965). This fact is in good agreement with observations by Twarog & Roeder (1956) that when an intact segment of the abdominal nerve cord is bathed in 180mM-KCl it takes 12-18 min. for conduction to be blocked, whereas this delay is only 10 sec. if a desheathed segment is perfused with this same saline. These authors concluded that the nerve sheath acts as a diffusion barrier against ions. Treherne (1961, 1962 a-c) has shown, however, that the nerve sheath is freely permeable to ions and that, rather than a static impermeability, a dynamic steady state can exist between haemolymph and the extracellular fluid. Removal of the nerve sheath would result in drastic changes in the composition of this extracellular fluid due to the disruption of a Donnan equilibrium with the haemolymph. These alterations would be responsible for the enhanced rates of depolarization observed in de-sheathed preparations. It has in fact been shown that elevation of the cation concentration to the extracellular levels resulted in a delayed conduction block in de-sheathed nerve cords of the cockroach irrigated with high potassium saline (Treherne, 1962c). In a similar way it has been shown that, whereas a saline containing 12·3 mw-K⁺ appreciably depolarized the giant axons of the cockroach in a de-sheathed nerve cord, a saline containing 17·1 mM-K⁺ did not alter significantly the resting potential provided the cation concentrations were elevated to the extracellular level (358·2 mM-Na+, 17·3 mM-Ca²⁺; Pichon & Boistel, 1967).

A new approach to this problem has been possible with the use of fine-tipped and mechanically resistant microelectrodes which allowed reliable intracellular recordings of membrane potentials to be made in sheathed nerve cords (Pichon & Boistel, 1966). Experiments carried out on isolated nerve cords bathed in different salines have given the following results:

The resting potentials were lower and the action potentials higher in sheathed preparations than in de-sheathed ones.

A saline containing 12·3 mM-K⁺ (instead of 3·1) produced only slight depolarization of the giant axons in sheathed preparations, leaving the action potential unchanged, whereas in de-sheathed preparations the resting potentials were decreased by some 10 mV. and the action potentials reduced by more than a half, thus demonstrating once more the disturbances involved by the de-sheathing procedure.

From the above results, it is logical to think that in the intact nerve cord, bathed in insect’s own haemolymph, the nerve fibres may have rather low resting potentials associated with high action potentials. It should be noted, however, that there may be a certain discrepancy between in vivo membrane potentials and those recorded in vitro. In the first place errors may arise from the nature of physiological salines used for electrophysiological investigations and which differ considerably from the haemolymph. Moreover, in Ringer solutions, all the cations are free, whereas part of them, for instance K⁺ or Ca²⁺ ions, may be bound to organic molecules in the haemolymph. In addition, it has been recently shown that an appreciable fraction of the haemolymph K⁺ might be related to the haemocytes and thus be inactive for excitable membranes (Brady, 1967 a, b). Another source of error may arise from the more or less extended lesion of the sheath and of the tracheae which necessarily occurs when the nerve cord is dissected away.

We shall describe in this paper the results of our investigations on intact nerve cords bathed in insect’s own haemolymph. They will concern the in situ values of the membrane potentials of the giant axons.

The same technique as that previously described (Pichon & Boistel, 1967) has been used throughout the present experiments, except for the reference electrode which consisted here of a hooked Ag-AgCl electrode held by a micromanipulator.

Except for long-term experiments, in which normal or ‘H’ salines (Pichon & Boistel, 1967) were used, no solution was employed in this work, the nerve cord being bathed in the insect’s blood.

Two different procedures were used:

In the first the insect was pinned ventral side upwards on a piece of cork. The animal was immobilized with plasticine and a small window was cut in the cuticle and also in the underlying epidermis just above the nerve cord between the 3rd and the 5th abdominal ganglia. A pair of hooked Ag-AgCl electrodes held on a micromanipulator were gently pushed beneath the connectives and lifted up in order to stretch the nerve cord slightly. One of these electrodes was connected to ground and was used as an indifferent electrode. The microelectrode was introduced into a giant axon between these two electrodes, which were also used to record extracellularly the activity of the nerve fibres. The cerci were stimulated by a puff of air and the corresponding activity in the nerve cord was continuously recorded by means of a kymograph camera (Grass C4 Camera). The advantage of this procedure is that it limits the area of injury. Unfortunately, such a preparation cannot be used for long-term experiments because of the very fast desiccation of the nerve cord on the electrodes and the frequent contractions of the abdomen of the insect which soon break the microelectrode tip.

The second procedure involved more extensive dissection but allowed prolonged intracellular recordings to be carried out. The head, legs and wings were cut away and the insect was pinned, dorsal side uppermost, on a piece of cork. A large window was cut in the dorsal integument and the underlying digestive duct was gently pushed aside by means of the indifferent electrode until the nerve cord was made visible. The two stimulating electrodes where placed under the connectives between the 3rd and the 4th ganglia and the microelectrode was introduced between the 2nd and the 3rd ganglia into a giant axon.

Experiments were carried out at room temperature ranging from 20 to 26° C. on adult male cockroaches reared in the laboratory on a mixed diet (lettuce, wheat flour, milk powder and tap water).

The magnitude of the action potential, resting potential, ‘sheath potential’,^* overshoot and also the rates of rise and fall of the action potential of giant axons have been measured in intact nerve cords bathed in the insect’s own haemolymph (Table 1). Typical tracings of the action potential and its derivative are given in Fig. 1. It is seen that very low resting potentials (43·0 ± 4·8 mV.) were associated with large action potentials (105·1 ± 6·8 mV.), thus giving a very high overshoot of 62·1 ±7·0 mV.

The ‘sheath potential’ reached 28·2 ± 3·9 mV. The mean maximum rates of rise and fall were respectively of 1161 ± 168 V./sec. and 386 ± 65 V./sec. (Table x).

The preceding results have shown that membrane potentials of the giant axons of the cockroach measured in situ, in nerve cords bathed in the insect’s own haemolymph, are very different from those reported for de-sheathed preparations studied in vitro in Ringer solution (Yamasaki & Narahashi, 1959; Boistel, 1960; Narahashi & Yamasaki, 1960), for a mean resting potential of 43·0 mV. is associated with large action potentials (105·1 mV.) so as to give a proportionally large overshoot of 62·1 mV. These results are in agreement with previously reported experiments on sheathed nerve cords studied in vitro (Pichon & Boistel, 1967). The interpretation of the previous results is, however, not altogether clear at the time: two different explanations might be valuable which differ essentially by the signification given to the ‘sheath potential’.

The simplest explanation is that the true resting potential (RP) of the giant axons recorded in sheathed nerve cords would correspond to the algebraic difference between the measured resting potential (RPm) and the measured ‘sheath potential’ (SP). Thus for the present experiments:

This value is somewhat higher than that in de-sheathed nerve cords bathed in Ringer solution (67·4 mV., Pichon & Boistel, 1967), but is still lower than the value of 77 mV. reported by Narahashi & Yamasaki (1960) for the same preparation. This corrected value of the resting potential would correspond to a mean overshoot of 33-9 mV. The latter value, which is appreciably higher than that reported for desheathed preparations bathed in Ringer solution (18·5 mV., Pichon & Boistel, 1967; 22 mV., Narahashi & Yamasaki, 1960), is consistent with an increased sodium concentration in the extracellular fluid suggested by Treherne (1962b).

This explanation which seems at first quite satisfactory is unconfirmed for at least two reasons. First, it is difficult to concile this corrected value of the resting potential with a high extracellular K⁺ concentration demonstrated by Treherne (1962b). It has been shown, for example, that the resting potential of the cockroach giant axons depends upon external K⁺ concentrations (Yamasaki & Narahashi, 1959). According to the curve given by the latter authors a resting potential of 71·2 mV. would correspond to an external K⁺ concentration lower than 4 mM/1. Now the estimated extracellular concentration in the sheathed nerve cord of Periplaneta is as high as 17·1 mM (Treherne, 1962b).^* The second reason is that the ‘sheath potential’ may be an artifact determined by variations of the microelectrode tip potential in different ionic solution (Pichon & Boistel, 1967). This latter explanation must not be definitively discarded for no further information concerning the nature of this ‘sheath potential’ is available, and new experiments, including the study of the isolated nerve sheath, are needed. ^†

The sodium equilibrium potential may thus be elevated in situ by two means: first, because of a low intracellular Na⁺ concentration (resulting from a better functioning of the sodium pump) and, secondly, because of a higher extracellular Na⁺ concentration related to the integrity of the nerve sheath. In these conditions, the occurrence of large overshoots in our experiments is quite explicable without postulating any possible error in the resting membrane potential measurement induced by the existence of the ‘sheath potential’. Moreover, the large value of this ‘sheath potential’ in the present experiments (28·2 mV.) may be easily related to an enhanced difference between the ionic compositions of the extracellular fluid and the haemolymph in nerve cords kept in situ.

The value of 1161 and 386 V./sec. for the maximum rates of rise and fall of the action potential are very near the values of 1100 and 380 V./sec. reported by Yamasaki & Narahashi (1959) for de-sheathed nerve cords of Periplaneta bathed in Ringer solution. It therefore seems likely that the net ionic inward current during the rising phase of the action potential and the net ionic outward current during the falling phase of the action potential are equivalent, whether the mean resting potential was 43·0 mV. in intact preparations or 70·3 mV. in de-sheathed ones.

It is thus demonstrated that the nervous system of the cockroach may function quite satisfactorily despite a relatively high K⁺ concentration in the haemolymph. This phenomenon seems at the time to be related to the specialized ionic composition of the extracellular fluid rather than to special properties of the giant axons.

We should like to express our gratitude to Dr J. E. Treherne for reading and correcting this manuscript.