Potassium Ions and Insect Nerve Muscle

A number of analyses of the haemolymph of insects have been made by different authors using direct chemical and colorimetric techniques or flame-photometry. These analyses show that the haemolymph from different types of insects varies enormously in ionic composition. Furthermore, the values obtained diverge considerably from the relatively constant compositions of the blood of vertebrate animals, and even from the figures commonly obtained from a variety of other invertebrates. The ratios of the principal ions present in both vertebrates and most invertebrates are similar in general proportions to those in sea water of the present day, as was first pointed out by Macallum (1910). For the purposes of the present paper interest is confined to the ions of sodium and potassium. These are present in sea water in the ratio Na:K, 50:1. In frogs the blood ratio is 40:1 and in man it is about 28:1. In most invertebrate animals the ratio falls between 25:1 and 50:1 (data from Prosser, 1950). In the insects, however, the ratio determined has often been much lower than these values and may be less than unity. Some extreme values are collected together in Table 1 ; further examples may be obtained from the sources referred to therein. Very low Na:K ratios are not found in all insects, and in some the ratio is nearer to that found in vertebrates. Two such examples are included in the table ; the values for frog blood are given for comparison.

In an extensive series of analyses Boné (1944), demonstrated that the low Na:K ratio in insects is associated with the herbivorous habit. The normal ratios are found only in carnivorous insects, and intermediate values are found in insects with an omnivorous diet. Attempts have been made to depress the ratio in omnivorous forms by feeding them on exclusively vegetarian diets containing a high concentration of potassium. Tobias (1948a) was able to increase the serum potassium of Periplaneta americana by 56 % after changing from an omnivorous diet to lettuce, which contained less sodium and more dry-weight potassium. He was not, however, able to depress the ratio below unity by this procedure. Tobias adds in a footnote a personal communication from Boné. The latter found no effect of diet on haemolymph Na:K ratio in Tenebrio molitor. He apparently concluded ‘that the ionic composition is a species characteristic, maintained by a mechanism of mineral regulation’. Successful depression of the ratio to a level just below unity was nevertheless achieved by Ramsay (1952), in the blood-feeding bug Rhodnius prolixus by adding potassium chloride to the blood diet. These data are interesting in themselves and raise many problems concerning mineral regulation and diet, but they present a much more serious problem to the nerve physiologist. The knowledge that vertebrates need to maintain a constant blood composition by active regulation has been known for a century and led to the famous Claude Bernard dictum ‘La fixité du milieu intérieure est la condition de la vie libre’. The insects referred to are capable of surviving considerable changes in blood composition. In Ramsay’s experiments normal responses were obtained from a bug whose haemolymph potassium concentration was five times the concentration which is lethal to vertebrates. Mellanby (1939) has regarded insect blood as a reservoir of water, quoting evidence that the blood may become reduced in volume by as much as 70 % by water loss. He makes no mention of the physiological consequences which one would expect to find in many animals following such a change.

The most pressing problem, however, concerns the irritable tissues, to which attention was drawn by Katz (1949). It is now well established that resting nerve and muscle membranes are in a state of polarization which disappears and partially reverses when the membrane is excited. The propagation of a wave of depolarization constitutes the nerve impulse, and local or propagated disturbances of similar type precede muscle contractions. The resting potential difference across the membranes of both nerve and muscle is maintained by active processes involving ion transfer (see review by Hodgkin, 1951, and Hodgkin and Huxley, 1952). A net result of this transfer is the maintenance of a difference of potassium-ion concentration between the inside and outside of the nerve and muscle membranes. The concentration is much greater on the inside of the membranes than in the surrounding body fluid of all cases studied. The resting potential is clearly associated with the difference in potassium-ion concentration, for an increase in the external potassium-ion concentration causes a progressive decrease in this potential. The decline is slight over a small range of potassium concentration starting at zero, but soon becomes logarithmic. The potential is zero at a concentration of about 130 mM. K/l. The curves giving the relationship between the resting potential and external potassium concentration are remarkably similar in squid giant axon, frog sartorius muscle and frog myelinated nerve (Hodgkin, 1951). Some of the recorded potassium concentrations in the haemolymph of herbivorous insects are so large (Table 1) that one would expect these nerve and muscle tissues to be almost completely depolarized in such a medium. Under these circumstances the nerve would not propagate an action potential and the muscle contractile mechanism could not be excited. A further and related issue concerns the sodium concentration. The maintenance of the action potential also demands the presence of a certain minimal quantity of sodium ions on the outside of the membrane (Hodgkin, 1951). In the haemolymph of the larvae of Sphynx pinastri and Pieris brassica, Brecher found only traces of sodium. Tobias found very little sodium in Bombyx mori larvae at pupation.

The potassium problem has been given two possible explanations. Heilbrunn (1943) proposed that the insect potassium is largely bound to organic molecules and is not, therefore, free to depolarize the membranes. Hodgkin (1951) suggested that insects may have evolved some other basic mechanism for producing the resting and action potentials than the classical one of the crab, squid and frog. The present paper describes investigations on the effect of potassium ions on nerves and muscles of Locusta migratoria migratorioides R. & F., with a view to establishing the extent to which the nerve and muscle membranes of a typical insect diverge from the classical examples in their relation to these ions. These insects are normally herbivorous and were reared and fed in the laboratory on common British grasses. They will, however, eat other foods when hungry, and become cannibals under conditions of extreme starvation.

In the earliest experiments ventral nerve cords were dissected out and mounted on silver electrodes. They were stimulated maximally at one end and the action potential was recorded near the other. These cords did not live long in any of the saline media used, and the action potential diminished quite rapidly when the potassium concentration in the bathing medium was raised to 70 mM./l. They survived much longer when simply exposed in situ, and it was early observed that survival time increased with decreasing extent of dissection. This was probably related to the necessity of maintaining the tracheal system intact, as was stressed by Pumphrey & Rawdon-Smith (1937), in regard to their cockroach preparations, although it is notable that Adrian (1931) kept Dytiscus isolated nerve cords alive for 2 days, and Roeder (1948) has more recently obtained continued spontaneous activity in isolated cockroach nerve cords over periods up to 60 hr. The changes in the multiple spike potential of the cord with changing potassium were not easy to interpret. It was clearly desirable to obtain a single nerve-fibre preparation in situ and one which could be obtained with very little damaging dissection. This proved quite easy to obtain, and depends on the observations that the muscles of arthropods are innervated by only a few fibres (Pringle, 1939; Wiersma, 1941). The fine branches of the motor nerves, usually carrying only two fibres, pass down the centre of each of the muscle units into which the muscles are divided and branch frequently to supply individual muscle fibres at several places. (Details of histology and nerve supply will be published later.) Thus, if finely tapering electrodes, insulated down to the tips (about to µ diameter), are inserted into a muscle unit, and finely graded square-wave stimuli are applied, a single motor fibre can usually be excited and fired antidromically. The action potential can be picked up from the whole nerve trunk some distance away. In the experiments described below, a unit of the flexor tibialis muscle was used to provide the appropriate nerve site.

Animals were decerebrated through a hole cut in the top of the head and laid, ventral surface uppermost, in a couch of plasticine which held them in position without the use of pins. The hole also served to make contact with a pool of saline which was earthed and used as a reference point. The abdomens were left completely free so that tracheal pumping was not interfered with (Fig. 1). A small piece of cuticle was removed from between the inner ventral ridges of the femur, exposing part of a unit of the flexor tibialis muscle. The stimulating electrodes were then placed in position. A piece of the ventral tergite of the metathorax was removed, together with the underlying epithelium, exposing the edges of two large air-sacs. These were separated to reveal the crural nerve which supplies the motor fibres of the flexor tibialis muscle below, and an insulated silver-hook electrode was placed under this. A pipette was inserted through the gap by the electrode to direct test salines for bathing the whole nerve trunk. Many preparations were ruined through autotomy, but an equal number survived. Stimuli were derived from a square-wave generator and were passed through an isolating transformer to reduce stimulus artefact. The action potentials were recorded with respect to earth via a direct-coupled amplifier. The potentials were recorded after first sucking away fluid surrounding the nerve with a pipette. They were triphasic, with the first phase positive as the impulses emerged from the leg which was at earth potential. The spike then swung negative as the impulse passed the recording electrode and then gave a final small positive phase. The negative phase was about 100 pV. and conduction velocity 2·2—2·7 m./sec. at 20 °C.

Various salines were tried in early experiments following the compositions of Chauvin (1941), Bělăr (1929), and Pringle (1938), both with and without added glucose. The ultimate procedure was to prepare two mixtures, each containing the same calcium, magnesium, phosphate, bicarbonate and chloride, but one with 140 mM. K/l. and 10 mM. Na/1., the other with 150 mM. Na/1. and no potassium. No glucose was used in the mixtures. These were mixed together in various proportions from stock so that a variety of potassium concentrations from o to 140 mM./l. could be obtained quickly. Each mixture had the same osmotic strength and none had less than 10 mM. Na ions per litre. Mixtures with specific Na/K ratios were kept in a thermostatic bath and fed to the pipette in the preparation. A basic mixture was used for the saline pool making contact with the body via the hole in the head. This mixture had the following composition per litre of water.

The experimental test salines had the same calcium, magnesium, phosphate, bicarbonate and chloride as this, but the Na/K ratio was varied as described above.

The stimulus strength was increased until a single spike just appeared, and the trunk was then bathed for some time in the normal saline. This was then replaced by the test saline and spike height plotted against time. The test saline was circulated gently round the nerve trunk. Some typical results are shown in graphical form in Fig. 2. Each curve was obtained from a single locust. With a potassium concentration of 70 mM./l. or less there is little change in the potential over 6 hr. With higher concentrations the spike height gradually diminishes, usually reaching a fairly steady level a few hours after the start of the experiment. The steady level is reached more slowly, and has a lower final value, as the potassium concentration is increased. Potassium ions do, therefore, reduce the nerve potential, at least in high concentrations. Alternatively, as Pringle suggested (personal communication), the nerve sheath, which has not been adequately described in insects, may simply be acting as a very efficient barrier to the passage of ions. To test this possibility, a microsyringe device was prepared so that salines could be injected through the sheath. The device consisted of narrow plastic tubing filled with oil and connected to a hypodermic syringe at one end and to narrow glass tubing at the other. Glass pipettes with 3-5 μ tips, ready filled with test saline, could be quickly attached by shellac to the glass tubing and mounted on a micro-manipulator. A small hole was first bored through the sheath just above the recording electrode with a fine glass needle. The tip of the pipette was then inserted and saline injected. 70 mM. K saline injected in this way reduced the negative phase of the spike to zero in a few seconds. 50 mM. K regularly produced an equal block and even 40 mM. K did so on one occasion. Saline with only 5 mM. K had no effect on the action potential. When the blocked region was subsequently bathed in low-potassium external saline the second phase of the action potential reappeared slowly. With very large in-jections it was possible to reduce the size of the positive phase of the spike in addition, even to obliterate it, and recovery was then very slow and sometimes incomplete.

The block developed more slowly when only small volumes of 70 mM. K were injected, and it was thus possible to follow stages in the building up of the block. After a few seconds, as potassium ions diffuse across to the active fibre, the local potassium concentration increases and partially depolarizes the nerve. This process produces a characteristic delay in the negative spike (see Bullock & Turner, 1950), which also appears reduced in height. The delay increases progressively until the blocking concentration is reached over a sufficient length of nerve, when the negative spike completely disappears. Later, as the potassium ions further diffuse down the nerve trunk and mix with the existing interaxonic fluid, the local concentration decreases again and the reverse process occurs (Hoyle, 1952). The delayed, small, negative spike potential suddenly reappears and increases gradually with reducing delay, until the action potential is normal again about 2 min. after the injection. It is evident, therefore, that the axons are depolarized by similar concentrations of potassium ions to those which depolarize crustacean (Cowan, 1934), squid (Curtis & Cole, 1942) and frog myelinated nerve (Huxley & Stämpfli, 1951), and that, as in these preparations, depolarization is very rapid when the saline is brought into direct contact with the axons. The sheath surrounding the locust crural nerve axons must act as a highly efficient diffusion barrier to potassium ions. This has been further confirmed by dissecting away small patches of the sheath (complete removal proved too difficult), in the vicinity of the recording electrode, and adding a local drop of high potassium saline. Blocking occurs almost immediately, and conduction is equally rapidly restored by washing the treated region in potassium-free saline (Fig. 3).

The question arises as to whether the sheath is an active or a passive barrier or both. It was noticed that in some preparations 70 mM. K saline applied to the whole nerve trunk did, in fact, reduce the nerve potential. This phenomenon appeared to be correlated with damage to the tracheal system, so the latter was deliberately damaged by cutting the tracheal trunks supplying the crural nerve, and the time course of reduction of the potential was measured. The potential now declined relatively rapidly in 70 mM. K saline (Fig. 2). This could be due to an effect on the axons themselves, owing to the depletion of their air supply, but this should also have affected their excitability, and this was not appreciably changed. It is also possible that the damage reduced the efficiency of the sheath. The latter covers the whole nervous system including the brain, the other ganglia and the peripheral nerves right down to the muscle insertions. It consists of three layers : an outer very thin tracheolated membrane, a thick structural non-cellular neural lamella which is laminated in places, and an inner single layer of flattened cells, the perilemma (Hoyle, 1952). The latter forms a continuous cytoplasmic cylinder lining the whole nervous system. The tracheolated membrane probably has some effect as a diffusion barrier, especially when undamaged and receiving fresh tracheal air (see muscle section below). Most of the protection against high external potassium must, however, come from the inner layers. 112 mM. K reduces the nerve spike slowly at first, but after about 2 hr. a steady plateau is reached showing that the sheath may be doing active work. Evidence that the sheath surrounding the fine distal branches of the nerves has similar properties has been obtained and is described in the section on muscle.

The simplest muscle preparations used have been recorded from intact extensor or flexor tibialis muscles using single electrodes inserted through small holes in the cuticle of the femur. Both muscles receive fast-fibre innervation from branches of the crural nerve. The threshold of the flexor fibres is lower than that of the extensor fibres, and hence ‘clean’ potentials can be obtained from the flexor muscles by graded stimulation of the whole crural nerve in the thorax. The stimulating and recording electrodes are placed in roughly the reverse positions to those shown in Fig. 1. Saline is admitted through a small hole bored at the proximal end of the femur and allowed to drain away through two holes bored at the distal end. These preparations are useful for some purposes, but they are complicated by the fact that each of these muscles is composed of several muscle units. The units themselves cannot be readily isolated for investigations, but use has been made of the fact that one of the muscles in the femur consists of only a single unit. This is the retractor unguis muscle which is inserted on the proximal margin of the femur. Its apodeme is 3-4 cm. long and passes down most of the femur and the whole length of the tibia and tarsus to attach to the claw of the tarsus. It is innervated by a single fast-fibre which can be stimulated by controlled stimulation of the appropriate branch of the crural nerve. The flexor and extensor tibialis muscles can be cut away almost completely, leaving this tiny muscle intact and functioning. A single stimulus via the fast axon produces a marked twitch. The effect of this is to bend the claw towards the tibia. The potentials from both types of preparation are usually monophasic, with the main spike positive with respect to earth. In some loci, however, a slight shift of the recording electrode changes the sign of the potential. Positive potentials are commonly recorded from crustacean muscles (Wiersma & Wright, 1947), and from some muscles of the cockroach (Roeder & Weiant, 1950).

The advantage of the extensor and flexor tibialis preparations is that damage to the tracheal system can largely be avoided. With these preparations the muscle spike potentials are reduced fairly rapidly by high potassium salines (Fig. 4). Some animals appear to ‘ hold ‘ 70 mM. K saline for a time, but never for more than 1-2 hr. When the spike-potential decline has started it proceeds fairly rapidly. In all instances, provided the 70 mM. K was not left for too long a period, the original potential height was restored in 5 mM. K saline. Following treatment with 140 mM. K there is an almost immediate contracture, at least of surface muscle fibres, and complete recovery of the original spike potential is not found. During the course of reduction of the extensor muscle spike potential following potassium treatment, the twitch tension which is produced by maximal single-shock stimulation of the crural nerve gradually decreases. It is quite small at 70 % original spike height, and just absent at 60 %, although the muscle still gives a small twitch to a pair of shocks at 50 msec, interval, and at 50 stimuli per sec. a vigorous contraction is present. The muscle is now entirely frequency operated. At 50 % original spike height 150 stimuli per second are needed to produce a noticeable contraction, and at 45 % there is no movement at all. The twitch reappears after prolonged washing with 5 mM. K saline.

Further light is thrown upon these observations by experiments with the retractor unguis muscle. Here there are only about fifteen fibres and the investing tracheolated membrane which surrounds each insect muscle unit can be torn away with glass needles without damaging the nerve. 70 mM. K saline reduces this muscle spike to a small fraction of the original potential in a few seconds. The potential is equally rapidly restored by washing the muscle in 10 mM. K saline. It is clear, then, that the muscle fibres are affected by high potassium just as rapidly as the nerve axons, and are also in line with crustacean and vertebrate muscles in this respect.

The slower decline of the whole intact muscle is due to two factors. First, the tracheolated investing membranes must act as partial diffusion barriers to potassium ions. Secondly, the outer fibres must prevent rapid mixing of the interstitial fluid with the externally applied saline. The retractor unguis muscle records show the nerve spike preceding the muscle spike (Fig. 5). This spike is not altered for some time after treating the preparation with high potassium saline, although the muscle spike potential is almost completely abolished. In 70 mM. K saline, which does not completely depolarize the muscle membrane, a small muscle potential is clearly present after treatment, and even in 140 mM. K some activity is apparent. It is evident, therefore, that the nerve axons are protected by the sheath even in the finer branches.

Direct measurement of the muscle membrane potential has been effected with intracellular micro-electrodes using the methods of Ling & Gerard (1949) and Fatt & Katz (1951). The technique used has been to remove a metathoracic leg, mount it in plasticine with the ventral side uppermost, and remove a small piece of cuticle with underlying epidermis from the proximal end of the femur between the main cuticle ridges. This exposes several parallel fibres of a unit of the flexor tibialis muscle without the necessity of any further dissection. The whole is covered in saline and a micro-electrode thrust into a fibre. The isolated legs start to deteriorate after about 20-30 min. so experiments have been done on a number of different animals and the results averaged. The graph of resting potential against saline potassium concentration (logarithmic scale) is given in Fig. 6. Each point represents the average resting potential from several fibres of each of three or more animals. Each animal was given a preliminary 5 min. in the normal saline and then taken to the test saline for 10 min. Different animals were used for each different saline. At the higher concentrations there is sometimes a marked step of a few millivolts as one first penetrates the tracheolated membrane before entering a fibre. Surface fibres reach a steady value in a few minutes but deeper fibres take much longer. Consistent potentials greater than 70 mV. have not been recorded at the lowest potassium concentrations, although the general shape of the curve would lead one to expect them. On the whole the curve obtained resembles very closely those obtained from both nerve and muscle fibres of other animals, except that it is displaced slightly to the right along the abscissa.

Sodium-free salines containing 10 mM. K and isotonic glucose to replace the sodium reduce the spike potential of the retractor unguis muscle in a few seconds. The potential was partly restored by replacing the glucose with isotonic choline chloride. Nerve stimulation then gave a prolonged muscle spike with a smaller peak potential than that obtained in normal saline.

The results of the experiments described above clearly show that the locust nerve and muscle membranes are affected by potassium ions in a similar way to the vertebrate and crustacean fibres studied by previous investigators. It is reasonable to suppose that other insects are not significantly different in this respect. Roeder (1948) found reversible blockage of conduction of Periplaneta nerve in 50 mM. K saline. The suggestion of Hodgkin (1951) that insects may have evolved a different method of producing nerve and muscle potentials can, in part, safely be ruled out. It is not necessary, at least, to invoke a different relationship to potassium ions. The problem of the high potassium in the haemolymph still remains. The suggestion of Heilbrunn (1943), that the potassium may be bound to organic materials and be effectively absent, may still apply. However, it is clear that even if the high potassium values reported represent free potassium ions, the herbivorous insects may be able to cope with them adequately at least for short periods, if not indefinitely. The nerve axons are all surrounded by a membrane which is partly impermeable to potassium ions and which may be capable of doing active work to provide the axons with their own stable environment independent of fluctuations in haemolymph concentration. The muscle fibres are not so well protected, although tracheolated membranes investing muscle units offer a temporary safeguard against a sudden change in blood concentration. The fibres are, nevertheless, ultimately at the mercy of blood potassium. They can, however, still produce contractions when partially depolarized by high potassium, principally because, like crustaceans (Wiersma, 1941), they do not rely on propagated action potentials for the normal operation of the contractile mechanism, but instead respond to the facilitating effect of successive impulses in a continuous series (Pringle, 1939; Roeder & Weiant, 1950).

These observations make it possible to explain the observations of Ramsay (1952), on the effect of feeding Rhodmus on blood containing potassium chloride. These animals survived for some hours when the haemolymph potassium was gradually raised from the normal level of about 10 mM./l. to values as high as 124 mM. following feeding on the artificial diet. These animals showed progressively weaker responses with time after feeding. As the blood concentration increased, the nerves were probably enjoying complete protection such as the sheath, which seems to be widely present in insects, could provide, as evidenced by the fact that they must still have been conducting action potentials for any muscle movements to have occurred at all. The weakened responses indicate that the muscle fibres alone were gradually being depolarized and hence capable of developing less tension, until eventually they did not respond at all.

The implications of the observations described on the effect of potassium ions on resting potential, to herbivorous insects in general, are great. The apparent blood potassium of many forms would make the muscle resting potential low, and occupy a position on the steeply sloping part of the resting potential/potassium concentration curve. Hence a change in blood potassium should tend to produce a marked change in resting potential and consequently a change in muscle properties, and perhaps also of the behaviour of the whole animal. The blood analyses described were made, as far as can be determined, on animals fed ad lib. on plant material. Indeed, only under such conditions can adequate haemolymph samples for analysis be obtained. Even when only partly starved, locust haemolymph becomes a thin paste which is present in surprisingly small volume. The range of qualitatively observed values of blood volume in a batch of animals reared under identical conditions is large. Potassium variations are therefore to be expected, and the values in the literature for potassium concentration may well be indications of the upper limit of these variations rather than the norm. If the mineral-regulatory machinery excretes excess potassium relatively slowly, then a heavy continuous intake of potassium by way of plant food which usually contains little sodium and much potassium (see Tobias, 19486) may readily lead to an accumulation of the latter. The observed values may not indicate the ‘basic’ blood potassium but only that of an insect whose potassium excretory mechanism is under stress. Although attempts have been made (with success) to raise the blood potassium of omnivorous insects, I know of no experiments in which attempts have been made to lower it in herbivorous ones.

Although insects have been shown to have blood which varies widely in composition, especially when subjected to special stress, it does not follow, as Tobias (1948 a) proposed, that the effective bathing fluid of the tissues must follow these changes. Many of the organs of insects are invested by tracheolated membranes (Wigglesworth, 1950) which may act as selective barriers between the blood and tissues. A thin film of fluid, such as must exist beneath the tracheolated membranes, containing adequate sodium and not too much potassium, is all that is necessary as a physiological medium for heart, nerve, muscle and perhaps other tissues. This may partly answer the question raised by Tobias (1948b), that there are only minute amounts of sodium in the pupae of Bombyx mori. The range of observed haemolymph potassium values described in the literature, from a variety of insects, does not include any which Locusta could not tolerate, at least for some time, although undoubtedly the higher values would seriously impair muscle efficiency.

Gratitude is expressed to Prof. B. Katz, F.R.S., and Dr G. P. Wells who initiated the work and gave constant encouragement; to Dr P. E. Ellis for considerable help with the experimental animals, and to the technical staff of the zoology department. Special thanks are extended to Dr J. Del. Castillo for his assistance with the microelectrode technique.