Extracellular Potassium in Neuropile and Nerve Cell Body Region of the Leech Central Nervous System

Neuronal function in both vertebrate and invertebrate central nervous systems depend upon interactions of the nerve cell with its environment. The ionic content of the fluid immediately in contact with nerve cells may be actively controlled, providing an ionic environment that differs in composition from the blood. This is seen in some invertebrates in which the electrical behaviour of nerve cells, in the intact nervous system, can differ markedly from those which have been dissected out and placed in direct contact with solutions. Insect axons actually cease to function when bathed in artificial solutions with the ionic composition of the blood (cf. Treherne, 1974). The annelids have no such blood-brain barrier, and ion-regulatory processes like those of insects have not been found in the annelid central nervous system. In these animals, inorganic ions and small water-soluble molecules diffus! relatively rapidly from the blood to the surfaces of glial and nerve cells. In doing so they cross a cellular endothelium, as well as a capsule of connective tissue, before reaching the glial and neuronal membranes by way of intercellular clefts (cf. Kuffler & Nicholls, 1966; Kuffler, 1967).

The annelids have provided a great deal of data on the ion content of the milieu of neurones and on the function of nerve and glial cells, and much of this information has been obtained from electrophysiological and morphological studies of the central nervous system of the medicinal leech. The nervous system of this animal, like those of many other invertebrates, consists of a chain of ganglia joined by paired connectives. The peripheral zone of each ganglion is occupied by nerve cell bodies, surrounded by glial cells. The electrical properties of the nerve cell bodies have been used to monitor the composition of the fluid surrounding them (Nicholls & Kuffler, 1964); such measurements provide a method of following the movement of substances through preparations of leech ventral nerve cord. The results have suggested that the ion composition of the extracellular fluid in the leech nervous system is unaffected either by its surroundings or by the glial cells (cf. Kuffler & Nicholls, 1966).

The monopolar nerve cell bodies in a leech ganglion make no synaptic connexions; they surround the synaptic region, a dense network of axons with countless dendrites and synapses - the neuropile in the centre of the ganglion. Previous studies of the ion content of leech ganglia have not included the neuropile, even though this region must be particularly significant with respect to integrative processes. There is as yet no experimental evidence of unrestricted diffusion in the neuropile. It cannot be ruled out that the fluid in the extracellular space within the neuropile has a composition different from that of the blood, for the neuropile is surrounded by a special inner ganglion capsule. This capsule could (unlike the outer capsule) restrict diffusion of various substances. Under such conditions the ionic content in the neuropile might differ from that in the nerve cell body region and in the bathing medium.

The content of ions in the extracellular spaces of the leech central nervous system has previously been determined indirectly, and only in the regions of the somata or connectives (Nicholls & Kuffler, 1964). In the experiments described here, direct measurements were made by means of ion-sensitive electrodes, and the study was extended to the neuropile region. By changing the composition of the bathing medium, we have studied the rate of movement of potassium ions through the extracellular spaces in the leech central nervous system. We have compared the potassium level in the extracellular spaces with that of the external medium and observed the effect of the cardiac glycoside ouabain on the potassium content of the nerve cell body region and the neuropile.

The leeches (Hirudo medicinalis L.) were obtained from a variety of supply houses. Usually the smaller animals, 12 cm or less in length, were selected for experiments.

The leech was briefly anaesthetized in saline solution containing 0·2 % chloretone (1,1,1-trichlor-2-methyl-2-propanol) and then rinsed several times in physiological saline. After it was stretched and fixed in a dissection dish, ventral surface down, the dorsal wall of skin and muscle was slit with a scissors along the entire midline and pinned out at the sides. The gut was removed and the blood sinus that encloses all the ganglia, lateral nerves and connectives was cut open and dissected out with fine scissors. The lateral nerves were severed peripherally. At all stages of preparation the body of the leech was regularly moistened with physiological saline.

The CNS of the leech consists of the cerebral ganglion, a chain of 21 uniform ganglia (normal ganglia) and the anal ganglion. Three normal ganglia in the most anterior part of the body, and three in the most posterior part, are joined by connectives shorter than those between the remaining 15 normal ganglia. The latter ganglia were used for experiments, except for the ganglia 5 and 6 at the level of the sexual organs.

The ganglia were pinned by the connectives in a transparent chamber (volume ca. 0·2 ml) in such a way as to avoid excessive stretching. When fixed in this way the ganglia rarely moved by spontaneous contractions. In some experiments the outer ganglion capsule was torn open with sharpened forceps.

The experiments were performed at room temperature (22–25 °C).

The fluid bathing the preparation was continuously flowing and could be changed rapidly, with a flow rate of 15–20 bath volumes/min, without affecting recording conditions. The normal bathing medium (physiological saline; Baylor & Nicholls, 1969) had the following composition (mM): NaCl, 115, KCl, 4; CaCl₂, 1·8; Trismaleate (brought to pH 7·4 with NaOH), 10; glucose, 11. In solutions with altered potassium the sodium concentration was either reduced (with 10 and 40 mM-K+) or increased (with o, 0·2 and 2 mM-K+) to maintain osmolarity. When ouabain (G-Strophantin, Serva) was included in the solution, the cardiac glycoside was dissolved directly in normal saline to obtain a concentration of 5 × 10⁻⁴M. This amount of ouabain did not evoke a change in the potential of the reference barrel, or in the response of the potassium-sensitive barrel tested in the experimental chamber.

For some experiments single-barrelled microelectrodes filled with 3 M potassium chloride were used; their d.c. resistance, measured in the normal bathing solution, was between 20 and 30 MΩ.

The double-barrelled microelectrodes were prepared as described by Zeuthen & Monge (1975) and Berridge & Schlue (1978). Two glass capillaries differing in outside diameter (1·2 and 2·0 mm) were glued together; then, under heat, the thinner was bent around the thicker by 360°, and finally the two were drawn out together with an electrode puller. The inner surface of the thicker central barrel was siliconized with pure dimethyl-dichlorosilane vapour. The tip of the central barrel was filled with a column of liquid potassium ion exchanger resin (Corning 477317)- The remaining part of the tip and shank was filled with 0·5 M-KC1. The ther electrode barrel was filled with either 3 M sodium acetate or 1 M magnesium acetate. The tip diameters of the microelectrodes were usually less than 1 μm. After filling, the electrode tips were equilibrated in physiological saline for at least an hour. Thereafter the d.c. resistance of both electrode barrels and the slope and selectivity of the ion-exchanger barrel were measured.

The d.c. resistance measured in the normal bathing medium was 1·5 × 10⁹Ω in the ion-exchanger barrel and between 10 × 10⁶and 50 × 10⁶Ω in the reference barrel. The potential difference measured with the ion-exchanger barrel ranged between 45 and 52 mV for a tenfold change of potassium concentration, and averaged 49 mV (S.D. ± 1·8 mV; n = 34). The response was usually 90% complete within less than 2·5 s. The half-time of the response (i.e. when half of the change in concentration measured was achieved) was ca. 0·5 s for most electrodes, and always less than 2 s. The electrodes deteriorated within a couple of days of being filled, as indicated by a decreased slope and a slower time course of their response. Therefore only electrodes which had been filled the same day were used in any experiment.

Responses of typical double-barrelled microelectrodes in solutions with various concentrations of potassium are shown in Fig. 1A. The left-hand series (4, 10, 40 and 4 mM) was obtained from one electrode and the right-hand series (4–40 mM, 4 mM) from another. A calibration curve derived from the responses of all electrodes is given in Fig. 1B: a fit of Nicolsky’s (1937) equation to the data points based on a selectivity coefficient for sodium k = 1/62. The response of the potassiumsensitive barrel measured in nominal absence of potassium in the bathing solution was 75·6 mV (S.D. ± 2·2 mV ; n = 8); the Nicolsky equation predicts a value of 76·0 for zero potassium.

The continuous line in Fig. 1 B deviates from a straight line even at concentrations above 4 mM, by 2·5 mV, so that a ‘real’ mean slope of the potassium-sensitive barrel amounting to about 51·5 mV per decade was indicated. Although we did not correct our measurements for the unlinearity above 4 mM potassium, the error thus introduced was less than 5%. Because the electrode was strongly nonlinear in the concentration range below 4 mM, we obtained concentration measurements in this range by reference to the curve of Fig. 1 B.

The response of the potassium-sensitive electrodes is expressed in concentration rather than in activity (the parameter actually measured by ion-sensitive electrodes). Because the potassium levels measured in the great majority of these experiments were extracellular, we have compared those measured within the ganglia directly with the concentration of potassium in the bathing solution, in which the electrodes were tested before and after each experiment. It was assumed that the activity coefficient for potassium of the fluid in the extracellular spaces within the leech ganglia was the same as that of the external solutions. To convert the concentration values given into activity, the measurements must be multiplied by the activity coefficient. The activity coefficient for potassium in physiological saline can be derived from pure potassium and sodium chloride solutions, and is ca. 0·75 (Robinson & Stokes, 1959).

The ion-exchanger barrel and the reference barrel of an electrode were connected to two inputs of a differential electrometer (WPI F-223A) by chlorided silver wires; the input resistance was 10¹⁵Ω and the leakage current was 10⁻¹⁴A. The reference electrode was a calomel electrode, communicating with the solution in the bath by way of a polythene tube filled with 3 M-KCI in agar. The difference between the potential of the potassium-sensitive barrel and that of the reference barrel gave the pure potassium signal. The potassium signal (K⁺) and the reference potential (E_ref) were recorded by two independent channels of a penrecorder. The responses could also be displayed on an oscilloscope.

As in other invertebrates, the central nervous system (CNS) of the medicinal leech is surrounded directly by blood. It consists of a chain of ganglia, enclosed by an endothelium comprising a continuous layer of cells. These ganglia are delimited by a superficial connective tissue sheath (the outer capsule; Fig. 2B) that is underlaid by a thin layer of small glial cells. Most of the nerve cell bodies, subdivided into six ‘packets’ formed by glial cells, are located between the outer ganglion capsule and an inner capsule enclosing the neuropile. The nerve cell bodies send their axons through the inner ganglion capsule into the neuropile (Gray & Guillery, 1963; Coggeshall & Fawcett, 1964; Kuffler & Potter, 1964). The inner ganglion capsule also encloses some cell bodies - single nerve cell bodies in the periphery of the neuropile, many microglial cells scattered through the neuropile, and two especially large glial cells lying next to the neuropile in the anterior and posterior regions of each ganglion, the neuropile glial cells (Fig. 2B), which send out tapering projections.

The extracellular spaces are delimited by the membranes of nerve and glial cells that lie between the outer and inner ganglion capsules to form clefts that may be as narrow as 150–200 Å. Extracellular spaces also invade the packet glial cells, forming a complex system of communicating channels. These vary considerably in extent, with diameters ranging from several hundred Å to a few μm. Some larger extracellular spaces (Type II, Gray & Guillery, 1963) contain smaller glial cells (microglia). The extracellular space between the numerous axons and the projections of the glial cells in the neuropile is of about the same dimensions as that between the nerve cell bodies and the packet glial cells.

In each leech ganglion there is an extensive compartment lying between the outer and inner ganglion capsules. This compartment contains packet glial cells and the somata of the neurones; it will be called the ‘nerve cell body region’. A second compartment includes the neuropile in the centre of the ganglion, consisting of axons, dendritic arborizations and glial cells. This neuropile is bounded by the inner ganglion capsule, which separates it from the nerve cell body region.

All the ganglia used for measurement of potential and potassium content of the extracellular spaces in the nerve cell body region and neuropile had Retzius cell bodies of the normal size. In Fig. 2 A a typical leech ganglion is shown from the ventral aspect, to indicate the site of electrode insertion; the sites of insertion in all other preparations were like that in this example. The electrode penetrated the ganglion as illustrated after being advanced perpendicular to its ventral surface at about the point of intersection of the two dashed lines. A simplified drawing of a histological cross section through the anterior part of the ganglion is shown in Fig. 2B. As the electrode tip was advanced from the ventral surface into the median ganglion packet a characteristic profile of potential and potassium concentration was recorded (Fig. 3). During passage of the tip of a single-barrelled electrode through the endothelium and the outer ganglion capsule, or immediately thereafter, a small-amplitude negative potential (– 4 to –14 mV) usually appeared. Occasionally this potential was also recorded by the reference barrel of a double-barrelled electrode; the potential change can be seen, for example, in the recordings of Figs. 3(a) and 7, and is accompanied by a transient rise in potassium concentration. This event was always (whether single- or double-barrelled electrodes were used) followed by a potential step representing the membrane potential of a Retzius cell body, with superimposed spontaneous action potentials of the cell (Figs. 3, 5 and 7; cf. Fig. 2B or orientation). The concentration profile recorded by the double-barrelled electrode shows an associated increase in potassium concentration, to about 100 mM (Figs. 3(c), 5). To ensure that the electrode tip was actually in an extracellular space within the nerve cell body region, it was withdrawn from the Retzius cell (Figs. 3 (c), 5, 7). In this position the electrical potential was –19 mV and the potassium concentration was 5·8 mM. Taken together, these data define the first recording site in the nerve cell body region. To obtain recordings from a second extracellular site in this region, the electrode was advanced until the tip emerged from the other side of the cell. This emergence was signalled initially by injury potentials; shortly thereafter the potential stabilized, again at an average of –1·9 mV. The potassium concentration in this differently defined part of the extracellular space was 5·8 mM, as in the first site.

From this position the electrode tip could be advanced further into the ganglion, for extracellular recording from the neuropile (Fig. 3(g); for orientation cf. Fig. 2 of Occasionally an anterior neuropile glial cell was encountered (NG in Fig. 2B), as dicated by a potential shift representing its membrane potential (ca. – 60 mV for an uninjured cell). In such cases the electrode tip soon passed through the cell and the potential dropped; at this stage it was certain that the electrode tip was in the neuropile (Fig. 3(g), (h)). Whether or not the glial cell was represented in the potential profile, the advancing electrode tip repeatedly recorded injury discharges as axons were encountered (cf. Fig. 5 B). In the neuropile the measured potential averaged –3·1 mV and the potassium concentration was 6·3 mM.

At all the recording sites the extracellular potassium concentration in the intact ganglion was distinctly higher than that in the bath. With the normal potassium concentration in the bathing medium (4 mM) the concentration at both sites in the extracellular space of the nerve cell body region was 5·8 mM (S.D. ± 0·6 mM; n = 27). In these parts of the ganglion the potential was between + 2 and – 8 mV and averaged – 1·9 mV (S.D. ± 2·4 mV ; n = 36). The average potassium concentration in the extracellular space of the neuropile was 6·3 mM (S.D. ± 0·7 mM; n = 15). The extracellular potentials measured in the neuropile also ranged from + 2 to – 8 mV, but here the mean was – 3·1 mV (S.D. ±3·0 mV ; n = 20).

Both intact leech ganglia and those with the ganglion capsule opened were exposed to bathing solutions containing more than the normal 4 mM potassium. The time course of concentration change, and the final concentration reached, were followed by recording from the extracellular spaces. With intact ganglia, the concentration of potassium ions in the bath was first increased to 40 mM and then returned to 4mM; recordings like those of Fig. 5 were used to measure the change of potassium concentration in the nerve cell body region and in the neuropile. The change of potassium declined exponentially (Fig. 4) in both compartments of the ganglion, as would be expected for a first-order diffusion process. The average half-times (t₀.₅) for the change of potassium concentration under the various conditions are shown in Table 1, together with the associated latencies (the delays between the beginning of solution exchange and the onset of the recorded concentration change within the ganglion).

Fig. 5 shows sample recordings from such experiments. The recording in A was from the nerve cell body region; the lower trace represents potential (F_ref) and the upper trace the associated concentration (K⁺). The downward deflexion in the potential recording indicates penetration by the electrode tip of a Retzius cell body that could be seen in the microscope. This penetration is also reflected in the concentration recording, by a simultaneous change to a high potassium level (ca. 100 mM), the intracellular potassium concentration of this cell. After a brief period the electrode tip was withdrawn from the Retzius cell body, into the extracellular space of the nerve cell body region. This event was accompanied by a reduction in potential from about –40 mV to almost zero. The concentration recording indicates an extracellular potassium concentration of 6·1 mM, rather than 4 mM as in the bath, when the bath potassium concentration was increased to 10 mM and subsequently to 40 mM. With some delay after the exchange of solution was begun the potassium concentration in the nerve cell body region began to change in both cases (Fig. 5 A). The change occurred gradually. In Table 1 it can be seen that the mean half-time of potassium increase in the nerve cell body region was 26·2 s; the measured latencies averaged 6 s. When the test solution was replaced by normal saline, the extracellular potassium concentration gradually returned to the initial level (Fig. 5 A). The mean half-time for this return in the nerve cell body region was 21·8 s and the mean latency was 7·4 s (Table 1). The highest extracellular potassium concentrations observed in the experiment of Fig. 5 A were 11 mM (with 10 mM in the bath) and 45 mM (40 mM in bath). Table 2 gives the means obtained in other experiments on intact leech ganglia.

Fig. 5B shows a sample recording with the electrode tip in the neuropile. Near the beginning of the potential recording, injury discharges of several pierced axons are visible (arrows). These discharges, accompanied by changes in the potassium concentration, provided a reliable criterion by which one could be sure that the electrode tip was located in the neuropile. The stationary (measured after a certain waiting time) potassium concentration measured in the extracellular space of the neuropile was 6 4 mM in this experiment. When the potassium concentration in the bath was rapidly increased to 10 mM, and later to 40 mM, the concentration in the neuropile extracellular space also changed, though much more slowly (t_0·6 of the increase 27 s, mean latency 10 s; cf. Table 1). After return to normal saline, with 4 mM potassium, the extracellular potassium concentration also returned to the initial level, 6 ·4 mM (with t0·6 20 ·5 s, and mean latency 9 ·53; cf. Table 1). The highest potassium concentrations reached in the extracellular space with increased potassium in the bath are given as mean values in Table 2.

The measurements of magnitude of change, latency and half-time were repeated with the ganglion capsule opened. This operation usually made it difficult to localize the electrode tip precisely, but was presumed to be in the neuropile in most experiments. In bathing media containing either the normal, 4 mM, or increased potassium concentrations (10 or 40 mM), the extracellular concentrations within the ganglion were identical to those outside. Moreover, the latencies and half-times of concentration change were distinctly shorter with the ganglion opened than with the capsule intact (Table 1). The difference between the mean half-times in the two cases represents the amount of time needed for potassium to move through the outer ganglion capsule and the immediately adjacent tissue barriers inside the ganglion.

When the comparison was made with the data for the nerve cell body region, the difference for inflow of potassium into a ganglion was 19 ·6 s and for outflow from the ganglion, 16 ·4 s. The corresponding figures for the neuropile were 20 ·4 s in the case of inflow, and 15 ·1 s for outflow.

The external potassium concentration was reduced to 2, 0 ·2 and o mM, with both intact and opened leech ganglia, and the potassium concentrations in the extracellular spaces of the nerve cell body region and the neuropile were measured.

Sample recordings of experiments on intact ganglia are shown in Fig. 6. The potassium concentration in the nerve cell body region gradually declined (Fig. 6 A in all three bath solutions, whereas the potential was unchanged in all three. Return to the normal bath concentration was in each case accompanied by a return of the extracellular potassium concentration to the resting level. The mean final extracellular concentrations are given in Table 2, and in all cases represent a smaller change than that in the external medium. For example, when the potassium concentration of the bath was o mM, the extracellular concentration in the nerve cell body region fell only to 1 ·6 mM.

Fig. 6B shows sample recordings from the neuropile. Here, too, the potassium concentration in the extracellular space changed in all low-potassium solutions, with no change in the extracellular potential. Again the change was gradual, as was the return to the initial level when the low-potassium solution was replaced by normal solution. In no case did the extracellular potassium concentration fall sufficiently to equal the concentration in the bath, even when the time of low potassium concentrations in the bath was prolonged up to 30 min ; the mean final concentrations in the extracellular space are summarized in Table 2. An unusual feature appeared in the response to reduction of external potassium to o and 0 ·2 mM. After these solutions were replaced by normal bathing medium, the potassium concentration in the extracellular space declined transiently, so that the overall minimum concentration was somewhat lower than the stationary level reached during exposure to the low-potassium bath (Fig. 6B). This potassium ‘undershoot’ might be due to the reactivation of Na/K pumps in neuronal and/or glial membranes.

These measurements were repeated with opened ganglion capsules. With this barrier removed, the concentrations in the ganglion were identical to those in the bathing medium, whether the external concentration was 2, 0 ·2 or o mM. Moreover, we plateaus were reached much more rapidly than when the ganglion was intact.

Both intact and opened leech ganglia were studied with regard to the influence of the cardiac glycoside ouabain. Ouabain was added to the bathing medium in a concentration of 5 × 10⁻⁴M.

Sample recordings from such experiments on intact ganglia are shown in Fig. 7, with the electrode tip in the nerve cell body region (Fig. 7A) or in the neuropile (Fig. 7B). The potential profile (E_ref) obtained as the electrode tip was advanced into the nerve cell body region reflects passage through the endothelium and outer ganglion capsule, penetration of a Retzius cell body, and finally withdrawal from to Retzius cell body into the extracellular space. The potassium concentrations accom-panying this potential profile are shown in the upper trace. In a solution containing ouabain, the extracellular potassium concentration in the nerve cell body region increased by an average of 1·2 mM (S.D. ± 0·4 mM; n = 5). The concentration then began to decline from this maximum, even though ouabain was still present in the bathing medium. This decline persisted after return to the normal bathing medium, until the initial level of extracellular potassium concentration was reached. As the recording of Fig. 7 B shows, ouabain also brought about a transient rise in potassium concentration in the neuropile. Here the mean increase was 3·8 mM (S.D. ± 1.0mM; n = 6), a greater change than that in the nerve cell body region. The extracellular potential was unchanged in the presence of ouabain (Fig. 7B, E_ref).

Every ganglion studied in this series of experiments exhibited only a transient increase in potassium under the influence of ouabain, whether in the nerve cell body region or in the neuropile. The potassium concentration began to return toward the original level 3–6 min after exposure to ouabain was begun, as a rule, and after returning to normal saline without ouabain up to 15 min were required before the initial level was regained.

The effect of ouabain on the extracellular potassium content was also studied in ganglia with opened capsules. Here the potassium concentration in the ganglion also rose by ca. 3·8 mM in the presence of ouabain, though it began to decline more rapidly - with a shorter delay after the onset of ouabain exposure - than in intact ganglia.

The preceding section describes direct measurements of the potassium content of the extracellular spaces in the leech central nervous system, as compared with that in a bath of normal physiological saline and of saline with reduced or increased potassium; the effect of ouabain was also examined. Most previous estimates of extracellular ion composition and ion movements in invertebrate central nervous systems have been indirectly, estimated from the electrical responses of neurones (cf. Kuffler & Nicholls, 1966; Treherne, 1974). Experiments of this sort have employed not only annelid nervous systems (Hirudo:Nicholls & Kuffler, 1964), but also those of insects (Carausius:Treherne & Maddrell, 1967; Treherne, 1972; Periplaneta:Treherne et al. 1970) and molluscs (Limnaea:Sattelle, 1973; Anisodoris:Mirolli & Gorman, 1973). The problem has also been approached, though more rarely, by measurements with radioactive isotopes (e.g. Periplaneta:Treherne, 1961; Hirudo:Nicholls & Wolfe, 1967).

In resting conditions, with the ganglion in normal saline, the potassium concentration in the extracellular spaces of a ganglion with outer capsule opened (as measured by potassium-sensitive microelectrodes) was identical to that in the bathing medium. But if the outer ganglion capsule was intact, the extracellular potassium concentration was higher, by 45% in the nerve cell body region and by 58% in the neuropile. That is, with 4 mM potassium in the bath the concentration in the nerve all body region was 5·8 mM and in the neuropile, 6·3 mM. This result differs from those of other studies of the leech central nervous system. Nicholls & Kuffler (1964) concluded from their analysis of the ion content of extracellular-space fluid that the potassium content could not deviate by more than 15–20% from that of the bathing medium. This conclusion was based chiefly on the similarity between the dependence on bath potassium concentration of the membrane potential of nerve cells in situ and that of exposed nerve cells.

More recently, Coles & Tsacopoulos (1979) measured the potassium activity in extracellular spaces in the retina of the drone bee, and found them to be considerably higher than in the bathing medium. These experiments, like those described here, employed potassium-sensitive double-barrelled microelectrodes. The potassium activity in the superfused retina in the dark was 6·3 mM, even though that in the bathing medium was 2·2 mM.

The rate of change of potassium concentration and the latencies in intact leech ganglia following increase and decrease in concentration of potassium ions in the bathing medium suggest that the nerve cell body region and the neuropile are accessible to potassium ions.

In the intact leech ganglion, at least, the half-times and latencies measured cannot result from delays within the measurement system itself. The saline solution in the bathing chamber was exchanged with a high flow rate, and the half-time of the response of the microelectrodes was always less than 2 s (cf. Methods). Surprisingly, the half-times measured in the neuropile were similar to those in the nerve cell body region (Table 1); the t test shows no significant difference between them. Nor are the differences in latency statistically significant, by the same test (Table 1). This consistency in half-times and latencies of the two ganglion compartments suggests that the inner ganglion capsule presents no appreciable barrier to the diffusion of potassium ions. The considerable reduction in the time required for movement of potassium that is observed after opening of the outer ganglion capsule supports this conclusion. Evidently the greatest delay is caused by the outer ganglion capsule, and by the intercellular clefts formed by glial and nerve cells in the nerve cell body region.

The estimates of ion diffusion rates available in the literature for leech central nervous system were mostly obtained by indirect measurements. Nicholls & Kuffler (1964) determined the time course of potassium action by monitoring the changes in membrane potential of nerve cells and converting these to the equivalent concentration changes. In Table 3 these data are compared with those found in the present experiments by recording from the nerve cell body region.

Almost all of our measured mean latencies and mean half-times of potassium exchange in intact leech ganglia are larger than the corresponding values in the literature (Table 3). A discrepancy is also apparent when previously published data for the responses of exposed cells are compared with those found here for extracellular spaces in opened ganglia. But because the sites of measurement differed, agreement need not have been expected. In the experiments of Nicholls & Kuffler (1964) the recording electrode was inside single somata in the nerve cell body region, and these were directly in contact with the bathing medium. In our experiment with opened ganglion capsules the electrode tip was in the extracellular space deep within the ganglion, usually in the neuropile. The outer ganglion capsule had not been completely removed, and the potassium ions had to pass tissue barriers before they reached the neuropile.

When ouabain was present in the solution bathing the leech ganglia, the potassium concentration in the extracellular space of the nerve cell body region was increased by 1·2 mM, and in the neuropile by 3·8 mM. In both compartments of the ganglion the increase was transient, even during prolonged exposure to ouabain. Under the influence of ouabain, potassium ions accumulated in the extracellular spaces; they came, presumably, from neurones and/or glial cells.

Mirolli & Gorman (1973), studying the nervous system of the marine snail Anisodoris, estimated the effect that pharmacological blocking of the sodiumpotassium-exchange pumps in neurones and glial cells could have on the potassium content of the extracellular spaces. Their estimate was based on the results of permeability measurements and on observations of the membrane potential changes in substitution experiments. They calculated that the potassium that accumulates in extracellular spaces under the influence of ouabain comes chiefly from glial cells, only a small amount originating in neurones. The most interesting aspect of their calculations, with regard to the present results, is that the increase in potassium in the Anisodoris nervous system in the presence of ouabain would be expected to be transitory.

In the experiments described here, the potassium concentration within intact leech ganglia increased to an average of 10 · 9 mM in the nerve cell body region, and 11 · 8 mM in the neuropile, when the concentration in the bath was increased from the normal 4 mM to 10 mM (Table 2). Moreover, when the bath potassium concert tration was lowered, the concentration in the extracellular spaces did not decrease to the same level. With a bath concentration of 0 · 2 mM, for example, the concentration in the nerve cell body region was 2 · 2 mM and that in the neuropile, 3 · 1 mM.

Coles & Tsacopoulos (1979), who found differences between the potassium concentration in the extracellular space within the drone retina and that in the surrounding medium, evidently inferred that the underlying mechanism was not a passive process. Probably the strongest evidence for this view was that in their experiments the extracellular potassium activity of 6·3 mM changed only slightly even when the potassium concentration in the bathing medium was increased from 2 · 2 to 7 mM. The potassium activity in the extracellular space was not proportional to that in the bath. In the leech central nervous system, however, the extracellular potassium concentration changed in the same direction as the imposed changes in concentration of this ion in the bathing medium.

In the case of the leech preparation we cannot yet decide, even with the data available at present, whether the higher potassium concentration within the ganglion as compared with that in the bathing medium depends entirely on passive events, or entirely on active processes. Certain candidate mechanisms, both passive and active, will now be discussed.

Electron-microscope studies of the leech central nervous system have shown that the extracellular spaces in the nerve cell body region and in the neuropile can be no more than 150 – 220 Å in width. Extracellular spaces with diameters ranging from several hundred Å to a few μm also invade packet glial cells (Coggeshall & Fawcett, 1964). The double-barrelled microelectrodes used in our experiments to record from single leech ganglia had tip diameters somewhat less than 1 μm. That is, the electrode tip was usually larger than the width of the intercellular gaps. The tissue surrounding an inserted doublebarrelled microelectrode, then, must in almost all cases have been damaged, so that an artificial intercellular gap was created. The potential profiles recorded from single ganglia make clear that in the nerve cell body region at least one packet glial cell and one Retzius cell must be entered or passed through by the electrode, before it is certain that the recording is from the extracellular-space system of the region. Within the neuropile, single axons are also encountered, as the injury discharges indicate. Because of the damage done by the electrode to neurone somata, axons and glial cells, these structures release potassium into the extracellular spaces, so that the concentration of potassium could be at least temporarily increased. A certain time would be required for the normal potassium concentration to be restored - by active uptake into nearby intact neurones and glial cells and/or by diffusion into the immediate surroundings or the bath. Such transient increases in potassium concentration could be observed in the intercellular gaps, but a short time (ca. 2–5 min) after the insertion of the microelectrode the potassium activity always returned to a stationary level, which was distinctly higher than the concentration in the bathing medium. We conclude that the greater amount of potassium in the ganglion as compared with that in the bath is not predominantly due to potassium depletion of injured neurones and glial cells in the vicinity of the electrode.

The potassium levels maintained in a leech ganglion and the symmetry of inward and outward potassium movements suggest that passive mechanisms might explain the observed elevation in extracellular potassium. Two closely related mechanisms that might be involved here have been discussed elsewhere (Coles & Tsacopoulos, 1978), as possible explanations of the increased potassium concentrations in extracellular spaces within the drone retina. First, the extracellular clefts in the leech CNS could contain polyelectrolytes which would affect the distribution of small ions in such a way as to set up a Donnan equilibrium. With negatively charged polyelectrolytes the potassium concentration in the extracellular spaces could be higher than that in the bath. That the potential in the extracellular spaces of the ganglion is negative with respect to the bath is consistent with this hypothesis.

The distribution of cations in these extracellular spaces could also be affected by surface charges of the neuronal and glial membranes (cf. Liittgau & Glitsch, 1976). If these fixed charges were negative potassium ions could accumulate in their vicinity. It should be noted, however, that bivalent ions are much more strongly influenced by the surfaces of charged membranes than monovalent ions (McLaughlin, Szabo & Eisenman, 1971); the concentration of potassium would have to be many times higher to displace, for example, calcium ions.

Still another mechanism could bring about the elevated intraganglionic potassium concentration, at least when the concentration in the bath is raised or lowered (Table 2). When the bath potassium is reduced (2, 0·2, o mM), there could be leakage of potassium from intact glial and nerve cells into the narrow intercellular clefts. The complex, extensive system of channels formed by the extracellular spaces, and perhaps the ganglion capsule with the endothelium, could restrict to some degree the intercellular movement of potassium ions from the extra-neural fluid to the bathing medium. Because of such restriction, concentration gradients could build up between the extracellular spaces in the ganglion and the bathing medium, so that elevated extracellular potassium concentrations could be maintained for considerable periods. During the times over which the present measurements were made, changes in this gradient would hardly have been noticeable. In the long term, however, the potassium concentration gradient should level out because of the gradual depletion of intracellular ion stores. With high external potassium (10, 40 mM), on the other hand, the extracellular concentrations would approach that of the bath as potassium diffuses inward, into the extracellular system.

But such considerations cannot apply to the elevated potassium concentrations found in the nerve cell body region and in the neuropile (5·8 and 6·3 mM, respectively) when the bath concentration is normal (4 mM), for under these conditions the system is surely in a steady state. This situation is more likely to be explicable by a Donnan equilibrium, and/or by an active process discussed in the next section.

The potassium content in extracellular spaces of the leech central nervous system could also be regulated by special ganglionic structures. Nicholls & Kuffler (1964) inferred from their experiments on the ion content of intercellular clefts in the leech nervous system that the amounts of potassium and sodium are not regulated by glial cells. In interpreting the results of the present experiments, we make the same inference about the outer ganglion capsule. This connective-tissue sheath consists of fusiform cells which are embedded in a dense network of fine fibrils presumed to be collagenous in nature. There is no cytological evidence of special metabolic activity in this outer ganglionic capsule.

However, the increased potassium concentration in the extracellular spaces of the nerve cell body region could arise from processes in the endothelium, a complete cellular lining of the inner surface of the outer ganglion capsule. The endothelium contains pinocytotic vesicles, and offers little resistance to the passage of cations (Nicholls & Kuffler, 1964). Cytological findings indicate that water and metabolites can move from the blood to the nervous system through this endothelial layer (Coggeshall & Fawcett, 1964). So far it has not been ruled out, at least for the intact leech, that the fluid within the extracellular spaces of the nervous system could have an ionic composition different from that of the surrounding blood, regulated by the endothelium. In the intact leech the central nervous system is completely enclosed by endothelium, with none of the damage that dissection entails. The fact that the intact ganglia used in the present experiments exhibited longer latencies and longer mean half-times for potassium exchange than did the preparations of Nicholls & Kuffler (1964) could indicate a lesser degree of damage to the CNS, including the endothelium, in the former case. In a situation with minimal injury any effect of endothelial activity on extracellular potassium concentration would be more easily detected. It is possible, then, that this cell layer maintains a level of potassium in the extracellular spaces of the leech CNS that is higher than that of the bath, and perhaps higher than that of the blood in the intact animal.

These elevated potassium levels might affect the function of the nerve cell body region and the neuropile in various ways, and could even determine some features of integration in the central nervous system of the medicinal leech. For example, there are reports in the literature that artificial changes in the potassium concentration of the surrounding fluid can alter the sensitivity of neurones in this nervous system.

Sensory P (pressure) cells have been shown to change their degree of accommodation, depending on the external potassium concentration (Schlue, 1976). Accommodation decreases when the external potassium concentration is increased ; the outward current through the membrane is reduced and is less effective in preventing action potential generation. Conversely, accommodation is more rapid when the external potassium concentration falls, increasing the outward current and thus counteracting the generation of an action potential more strongly. Another effect, on the firing pattern of leech CNS neurones, has been demonstrated by Baylor & Nicholls (1969). In their experiments a small increase in the potassium concentration of the bathing medium altered the frequency of spontaneous discharge and of synaptic potentials in the neurones.

Elevated intraganglionic potassium concentration can thus influence at least three features of leech neurone activity - accommodation, the discharge of spontaneous action potentials, and the occurrence of synaptic potentials.

This research was supported by a Fellowship (Heisenberg-Stipendium, Schl 169/5) and equipment grants (Schl 169/2, 169/4) to W. R. S. from the Deutsche Forschungs-gemeinschaft. J.W. D. was on leave of absence from the Ruhr-Universität Bochum