Mechanisms of Ionic Homeostasis in the Central Nervous System of An Insect

We have used insect nervous systems to study three related aspects of glial-neurone interactions : the contribution of neuroglia to blood-brain barrier systems, the nature and functional significance of the neuronal microenvironment delimited by the neuroglia, and the homeostatic role of glial cells in regulating the ionic composition of this environment.

Insects provide convenient preparations for these investigations, for (unlike most invertebrates) they possess well-developed blood-brain barriers that differ from those of mammals in being associated with glial rather than vascular elements (Treherne & Pichon, 1972; Abbott & Treherne, 1977; Lane & Treherne, 1980). Furthermore, insects exhibit an extreme form of ionic homeostasis by maintaining ionic concentrations at the neuronal surfaces that can be very different from those of the blood plasma (see Treherne, 1974).

Our research has concentrated on the central nervous system of the cockroach, Periplaneta americana. This avascular organ is delimited by a connective tissue sheath, the neural lamella (Fig. 1). The neural lamella contains collagen embedded as a matrix of glycosaminoglycans which, in histochemical tests, are identified as a mixture of chondroitin, dermatan and keratan sulphates (Ashhurst & Costin, 1971a).

Beneath the ensheathing neural lamella is the perineurium, a layer of overlapping neuroglia attached to one another by septate and tight junctions. The tight junctions are found predominantly at the inner ends of the clefts between the cells. The perineurial cells are also linked to each other and to the underlying neuroglia by gap junctions (Lane & Treherne, 1972; Skaer & Lane, 1974; Lane, Skaer & Swales, 1977; Lane, 1981). The neuroglia are packed around the neurones with a separation of 10–20 nm, so that these cells lie in a complex of narrow interconnecting channels. As originally described by Smith & Treherne (1963) these extracellular channels contain electron-dense material (see Lane, 1981) which has been histochemically identified as hyaluronate in the larger glial lacunar spaces of the ganglia (Ashhurst & Costin, 1971a).

In this review we examine research upon the cockroach central nervous system that indicates both passive and active roles for neuroglia in maintenance of the ionic composition of the neuronal environment. We also present recent electrophysiological data that demonstrate the importance of the perineurium as a diffusion barrier, and radioisotopic evidence of an anion matrix which acts as an extracellular cation reservoir.

A critical factor in our analysis is the localization of the blood-brain barrier system. This system limits the movements of water-soluble ions and small molecules between the blood and the neuronal surfaces, as has been demonstrated in electrophysiological experiments (e.g. Hoyle, 1953; Twarog & Roeder, 1956; Treherne et al. 1970). It has been supposed to be located in the perineurium since extracellular tracers (peroxidase, microperoxidase and ionic lanthanum) permeate the neural lamella, but are unable to penetrate beneath the superficial layer of neuroglia. The tracers are capable of only limited entry into the clefts between the perineurial cells, which is suggested to be due to the intercellular junctional complexes, notably tight junctions (Lane & Treherne, 1972).

The ultrastructural studies do not necessarily indicate an impermeability to smaller substances. It is known, for example, that brief exposure of cockroach connectives to hypertonic urea increases access of sodium, potassium and lithium ions to the axon surfaces (Treherne, Schofield & Lane, 1973; Schofield & Treherne, 1978) but does not facilitate penetration of ionic lanthanum (Treherne et al. 1973) or microperoxidase (J. E. Treherne, P. K. Schofield & N. J. Lane, in preparation).

A superficial diffusion barrier, at the perineurial level, is also difficult to reconcile with the rapid fluxes of radiocations which have been observed between the plasma and the central nervous tissues (Treherne, 1961 ; Tucker & Pichon, 1972).

We have, however, recently provided further evidence that the blood-brain barrier is located in the perineurium, by recording with microelectrodes from perineurial cells, identified by injection of peroxidase (Fig. 2d).

The cockroach blood-brain barrier may be characterized electrophysiologically, by recording the diffusion potentials that are generated across the barrier when the ionic composition of the external saline is altered. These potentials have been measured within giant axons, in the extracellular spaces immediately outside them, and with sucrose-gap recordings (Treherne et al. 1970; Pichon & Treherne, 1970; Pichon, Moreton & Treherne, 1971). We have now shown that these potential changes are at their largest amplitude within the perineurial cells, and only slightly attenuated immediately below this neuroglial layer (Fig. 2b). These observations show that the potentials are generated across the outermost membranes of the perineurial cells, and the junctions, joining the cells, and that this interface may therefore be identified as the blood-brain barrier for monovalent cations.

Brief exposure of cockroach connectives to hypertonic urea, in the manner of Ussing (1968), greatly increases the access of water-soluble cations to the axon surfaces. This is deduced from the accelerated electrical responses of the axons to changes in external sodium, potassium and lithium concentrations (Treherne et al. 1973; Schofield & Treherne, 1978), by the reduction in the amplitude of extraneuronal potential changes (Treherne et al. 1973), and by the substantial increase in the fast component of ²²Na efflux (see Fig. 6).

The increased access induced by urea treatment does not appear to result from appreciable cellular damage, for extracellular tracers are excluded from the perineurial cytoplasm of urea-treated preparations (Treherne et al. 1973) and the uptake and efflux of ¹⁴C-xylose and inulin are unaffected by urea treatment (J. E. Treherne, unpublished observations). Furthermore, brief exposure to hypertonic urea does not reduce the membrane potentials of perineurial cells, or the resting and action potentials of surgically exposed giant axons (Treherne et al. 1973).

Despite the increased intercellular access of monovalent cations induced by urea treatment, the conventional extracellular tracers, ionic lanthanum (Treherne et al. 1973) and microperoxidase (J. E. Treherne, P. K. Schofield & N. J. Lane, in preparation) still fail to penetrate into the subperineurial extracellular system. This suggests that the effect of urea treatment on intercellular cation movements may be to alter the charge characteristics of the perineurial junctional complexes in the absence of appreciable structural changes that may be required to affect the intercellular diffusion of larger charged particles.

The presence of the peripheral, neuroglial, blood-brain barrier ensures that the movements of sodium ions between the external medium and the extracellular fluid are largely transcellular. As suggested in the model illustrated in Fig. 3, the escape of sodium across the blood-brain interface is mediated by sodium pumps on the outer perineurial membranes. This accounts for the observed effect of ethacrynic acid in slowing the rapid, terminal, decline in amplitude of the action potentials in preparations exposed to sodium-free saline (Schofield & Treherne, 1975). Similarly, the net uptake of sodium appears to be mediated by pumping at the inwardly facing perineurial and glial membranes. This can be concluded from the effect of ethacrynic acid and dinitrophenol in slowing the net inward movement of sodium ions to the axon surfaces, following exposure of sodium-depleted connectives to normal, high-sodium saline (Schofield & Treherne, 1975). It is also indicated by the inability of lithium ions to gain access to the extracellular fluid in sodium-depleted connectives despite substantial accumulation of this cation within the nervous tissues (Bennett, Buchan & Treherne, 1975), for Na/K pumps appear not to accept lithium ions (Keynes & Swan, 1959; Baker, 1965).

According to the model illustrated in Fig. 3 the sodium content of the perineurial cytoplasm and that of the underlying glia (to which it is supposed to be linked by gap junctions) will be determined by a balance between the passive sodium permeabilities and sodium pumping. The pump on the inwardly facing perineurial and glial membranes will tend to re-cycle sodium ions in the vicinity of the axon surfaces. The decline in extracellular sodium concentrations (in sodium-free saline) will therefore primarily result from reduced sodium pumping as the intracellular concentration in the glial compartment falls.

If the sodium pumps on the outer perineurial membranes have a higher sodium affinity than those on the inwardly directed perineurial and glial membranes then at low intracellular sodium levels net pumping would tend to be largely outwardly directed. Such a system could give rise to the accelerated terminal decline in extracellular sodium concentration, on initial exposure to sodium-free saline (Fig. 4), which can be slowed by sodium-transport inhibitors (Schofield & Treherne, 1975).

The slow decline in extracellular sodium which occurs on initial exposure of cockroach connectives to sodium-free saline (Fig. 4) can be accounted for by the provision of sodium from a reservoir (Treherne & Schofield, 1978). This is indicated by the observation that connectives can lose around 50% of their total sodium content with no appreciable decline in sodium concentration at the axon surfaces (Fig. 5). The reservoir would not appear to be easily refilled since the return of sodium to the bathing medium results in a rapid recovery of extracellular sodium (Fig. 4) that is accompanied by only partial recovery of the sodium content of the connectives (Bennett et al. 1975). Furthermore, subsequent outward and inward movements induced between the external medium and the extracellular fluid occur at similar, rapid, rates (Fig. 4). The existence of a sodium reservoir (either extracellular or intraglial) has also been proposed by Fulpius & Baumann (1969) to account for the persistence of receptor potentials in drone photoreceptors during prolonged exposure to sodium-deficient saline.

The inability to recharge the postulated reservoir in the cockroach central nervous system following exposure to sodium-free saline could result from uncoupling of the glial compartment from the perineurial cytoplasm. Under these circumstances the glial compartment would fill only slowly by passive diffusion through the glial membranes. Such uncoupling in sodium-free media could result from an increase in intracellular calcium (cf. Baker, Hodgkin & Ridgeway, 1971) either by reducing the permeability of the junctional channels directly, as proposed by Rose & Rick (1978), or indirectly by changing intracellular pH (cf. Meech & Thomas, 1977), as proposed by Turin & Warner (1977) and, in the case of neuroglia, by Orkand, Orkand & Tang (1981).

The perineurial junctional complexes divide the extracellular system into two fractions: a superficial one, in the neural lamella and underlying perineurial clefts, and a subperineurial one, between the closely applied glial and neuronal membranes, which is effectively isolated from the blood plasma.

We have recently studied the ionic composition of these extracellular environments using radioisotopes. As previously shown (e.g. Treherne, 1962), the efflux of radiocations and ³⁶C1 can be represented as a two-stage process (Fig. 6a). In the case of sodium ions the fast component comprises 21·6% (6·3 ±0·4 mmol/kg tissue) of the total exchangeable sodium. This component does not exhibit the characteristics of an intracellular cation fraction. For example, both the rate and the size of the fast component of sodium efflux are unaffected by dinitrophenol and sodium transport inhibitors (ouabain, ethacrynic acid). Furthermore, cation/chloride ratios of the fast components are not as would be predicted for intracellular ion fractions. For example, the calcium/chloride ratio of the fast components is exceptionally high for an intracellular fraction (0·20/6·0 mmol kg⁻¹ = 0·033) and exceeds that for the external medium (2·0/131·7 mmol kg⁻¹ = 0·015). The ratio between the fast components of sodium and chloride efflux (5·7/6·7 mmol kg⁻¹ tissue = 0·85) is close to the sodium/chloride ratio for the bathing medium (120·0/131·7 mmol I⁻¹ = 0·91) J. E. Treherne, P. K. Schofield & N. J. Lane, in preparation).

The above observations imply that the fast component of sodium efflux is largely extracellular as was originally supposed (Treherne, 1962). This fraction must apparently be located above the level of the perineurial tight junctions (i.e. largely in the neural lamella and superficial perineurial clefts) because there is an effective intercellular diffusion barrier in the peripheral neuroglia (see above). There could, however, be a small intracellular contribution from the sodium contained within the relatively thin perineurial glia. It should be noted that the half-time for radiosodium efflux (t_0·5 = 63·5 s) into sodium-free saline (i.e. net efflux) approximates to that for the net sodium movements between the external medium and the axon surfaces calculated from electrophysiological observations (Fig. 4b, c). Disruption of the blood-brain barrier, by brief exposure to hypertonic urea, results in a substantial increase in the fast component of sodium efflux (Fig. 6). Chloride movements are less affected, so that the additional ion fraction released by urea treatment contains considerably less chloride than sodium ions: the Na/Cl ratio of this fraction is 9·6/3·6 mmol kg⁻¹ tissue = 2·7 as compared with a ratio of 120·0/131·7 mmol I⁻¹ = 0·91 for the bathing medium and 5·7/6·7 mmol kg⁻¹ tissue = 0·85 for the fast fraction in untreated cords. Such an excess of sodium over chloride would be expected if the fast fraction revealed by urea treatment consisted of ions which had been in Donnan equilibrium with fixed anionic sites within the extracellular system. The existence of such sites can be correlated with the nature of the sub-perineurial extracellular material. This material has been found to bind lanthanum ions (J. E. Treherne, P. K. Schofield & N. J. Lane, in preparation) and contains hyaluronate (Ashhurst & Costin, 1971a) which could thus constitute, or contribute to, an extracellular anion matrix. This system bears some similarities to the proposed Ca-binding matrix which has been postulated to overlie the plasma membrane of the squid axon and to influence Ca efflux into the bathing medium (Baker & McNaughton, 1978).

Regulation of the ionic composition of the immediate fluid environment of insect nerve cells appears to be achieved by a combination of passive and active processes involving the neuroglia and an extracellular anion matrix.

The intercellular diffusion barrier in the superficial layer of neuroglia, the perineurium, protects the underlying neurones from the immediate effects of the large fluctuations in ionic composition which occur in the blood plasma (Pichon & Boistel 1963 ; Pichon, 1970; Treherne, Buchan & Bennett, 1975 ; Lettau et al. 1977) and, also, from the adverse effects of extraneous pharmacologically active and toxic substances. However, it also renders them more susceptible to fluctuations in chemical composition of their immediate environment, which might result from electrical and metabolic activities. In insects, there are no large extracellular fluid compartments equivalent to the vertebrate cerebrospinal fluid. The neurones are thus confined in a very restricted microenvironment, frequently consisting of narrow clefts of between 10 and 20 nm in width. They will, therefore, be potentially vulnerable to rapid alterations in extracellular ion composition (resulting from neuronal activity), as well as from slower changes (resulting from variations in the blood).

The available evidence indicates that the neuroglia play an important role in regulating the sodium content of the brain microenvironment in the cockroach. We also have circumstantial evidence for the involvement of an anion matrix which could act as an extracellular cation reservoir.

The possible involvement of anion matrices in extracellular ionic homeostasis has been previously suggested on the basis of limited physiological evidence in some invertebrate nerves (e.g. Chamberlain & Kerkut, 1967; Treherne, Carlson & Gupta, 1969; Treherne & Moreton, 1970; Sattelle, 1973; Abbott, Pichon & Lane, 1977) and, as a consequence of histochemical or biochemical demonstrations of extracellular glycomolecules, in arthropod nerves (Lemire & Deloince, 1970; Ashhurst & Costin, 1970a, b, at the node of Ranvier (Langley & Landon, 1969; Langley, 1970; Landon & Langley, 1971) and in vertebrate brain (e.g. Szabo & Roboz-Einstein, 1962; Bondareff, 1967; Nicholson, 1980). There is, however, very little information to enable quantitative assessments to be made of the physiological roles of extracellular anion matrices in nervous tissues. An obvious possibility, in the cockroach nervous system, is that extracellular ions associated with the anion matrix contributes to the postulated sodium reservoir (see p. 66). Now the Na/Cl ratio from the sub-perineurial extracellular fraction is 2·7, as compared with 0·91 in the external medium. This suggests that for every free sodium ion there are two associated with the matrix which could be released as the concentration of free sodium ions decreases during initial exposure to sodium-free saline. Such a release would result in a threefold increase in the quantity of extracellular sodium available for re-cycling by glial sodium pumping.

The possibility also exists that the matrix could be involved in short-term homeostasis necessitated by the restricted extracellular environment. The effects of the tiny microenvironment can be illustrated in the case of the cockroach giant axons. In these the sodium entry which mediates the inward current of the action potential has been estimated to be 6·3 pmol cm⁻² impulse⁻¹ (Narahashi, 1963). Now with extra-axonal clefts of 15 nm in width containing 120 mmol Na+ (plasma concentration) the sodium immediately available to carry the increased current would be 180 pmol cm⁻² (i.e. enough to support only about 30 action potentials). The sodium within the extra-axonal space is unlikely to be rapidly augmented by that contained within the long, single, mesaxon cleft. Thus even if this cleft acted as an infinite reservoir it can be calculated, using the approach of Treherne et al. (1970), that with an axon of 50 μm diameter, it would take about 1·2 s for half equilibration with the ions contained in the extra-axonal space, which is long when considered in relation to stimulation regimes with spike intervals of only a few milliseconds (Parnas et al. 1969).

The above considerations indicate that the free sodium ions in the fluid immediately bathing the axon surfaces is likely to support only a limited number of action potentials. This situation is essentially similar to that calculated for the photoreceptor cells of the locust eye, in which a minimal estimate indicates that there is only about 20 times more sodium immediately available than is required for one full-sized response (Shaw, 1977). However, unlike the cockroach giant axon system, the fluid bathing the photoreceptor membranes communicate with additional extracellular spaces via channels of short path length (12 μm). This, it is postulated enables a rapid redistribution of sodium so as to recharge the fluid layer at the receptor surfaces (Shaw, 1977). In the cockroach giant axon system, on the other hand, there is only a single, long mesaxon cleft which could only slowly refill the extra-axonal clefts by diffusion. The abdominal giant axons can nevertheless conduct at 200 impulses s⁻¹ (Narahashi & Yamasaki, 1960), and can fire for some minutes at 50 s⁻¹ (Parnas et al. 1969).

In the cockroach central nervous system additional sodium (twice as much as in free solution) would be provided by that contained in the extracellular anion matrix. This sodium will increase the time (by up to three times) during which full-sized action potentials are maintained during rapid axonal firing. The presence of the matrix would also ensure that potassium ions released during neuronal activity would undergo a reduction in activity coefficient in the vicinity of the axon surfaces. Such an effect could, for example, explain why the decline in potassium activity (deduced from the decay of the negative after-potential) is more rapid at the surface of the cockroach giant axon (t_0·5 = 9·2 ms; Narahashi & Yamasaki, 1960) than that of the squid (t_0·5 up to 100 ms; Frankenhaeuser & Hodgkin, 1959) despite the relatively leaky glial covering of the latter (Villegas & Villegas, 1964, 1968).

Extracellular polyanions could thus function as a short-term homeostatic device to maintain the overshoot of the action potential: first, by releasing sodium ions in the vicinity of the axon surfaces and, secondly, by reducing axonal depolarization to reduce the degree of inactivation of the sodium channels.

An important feature of this mechanism is that the release of sodium from the matrix would tend to maintain the concentration of this cation at the axon surfaces whilst enabling an increase in sodium concentration to occur beneath the axonal membrane during excitation. This increase could stimulate linked Na⁺/K pumping (cf. Baker et al. 1969) which, consequently, would remove excess extracellular potassium and maintain the sodium concentration at the axon surfaces (see Varon & Somjen, 1979). Such axonal, rather than glial, cation pumping appears to play a significant role in ionic homeostasis in Necturus optic nerve (Tang, Cohen & Orkand, 1980).

Now in the case of a cockroach giant axon a firing frequency of 50 s⁻¹ would require the relatively large increase in sodium extrusion of 50×63 = 315 pmol cm⁻² s⁻¹ to maintain the sodium gradient across the axon membrane : a value which by itself exceeds the maximal rate of sodium pumping of around 150 pmol cm⁻² s⁻¹ k measured in squid axons with raised intracelllar sodium concentrations (Brinley & Mullins, 1968). In the insect central nervous system, therefore, it seems likely that the effects of axonal ion pumping must be augmented by movements of sodium ions induced by extracellular current flow during spatial buffering of released potassium (see Nicholson, 1980; Gardner-Medwin, 1981) as well as by sodium extrusion by linked Na/K pumps in the neuroglia, as suggested by the results of Coles & Tsaco-poulos (1981).

The available evidence thus suggests that ionic homeostasis of the brain microenvironment in this insect is achieved by a combination of passive and active processes involving the neuroglia and an extracellular anion matrix. The intercellular diffusion barrier in the superficial neuroglia protects the neurones from the large changes which occur in the ionic composition of the blood plasma. This passive role of the superficial neuroglia is reinforced by glial and axonal cation pumping which, it is postulated, tend to recycle sodium ions so as to maintain the extracellular concentration during prolonged exposure to sodium-deficient saline. This effect could be augmented by a sodium reservoir associated with an extracellular anion matrix. This matrix could also serve an important role in short-term ionic homeostasis by buffering extracellular sodium ions, to enable increases in intra-axonal sodium ions to occur and thus to stimulate the axonal Na/K pump, without depleting the extracellular concentration of free sodium ions.