The optic nerve of Necturus has proved a useful preparation for the study of glial cell membranes in vivo and in vitro with anatomical relations to axons intact and isolated following axon degeneration. The glial membrane potential behaves as a selective potassium diffusion potential; there is no evidence of a significant permeability to other naturally occurring ions. The specific membrane resistance of the glial cells is high compared to that of neurones; there are low-resistance intercellular connexions among the cells which permit the passage of both ions and the dye Lucifer Yellow. The cells are readily and reversibly uncoupled by procedures which decrease the intracellular pH. There is no evidence for voltage-sensitive sodium channels in the membrane. Following sodium gain and potassium loss the membrane displays a potassium-dependent strophanthidin-sensitive electrogenic sodium pump. The glial membrane is depolarized by potassium released from active axons as well as by glutamate. The glial depolarization contributes to potentials recorded with surface electrodes. Depolarization by K+ plays a role in the redistribution of K+ which locally accumulates around active neurones and also affects glial metabolism and glucose uptake.
Many functions of cells have been shown to be linked to the behaviour and properties of their external membranes. Thus, transport phenomena, receptors for specific molecules and electric activity are studied in membranes for the insight they provide into the control of physiological processes. In our studies of astroglia the motivation is somewhat different. Measurements of the physiological properties of these cells were begun ‘as an initial step in testing various hypotheses and widely held assumptions concerning the role of neuroglia’ (Kuffler, 1967). The physiological role of these cells is still rather poorly defined ; we study their membrane properties to gain additional clues to possible functions.
The present study of glial cells in Necturus optic nerve began about 15 years ago in Stephen Kuffler ‘s laboratory in Boston (Kuffler, Nicholls & Orkand, 1966). The preparation was used because of its anatomical simplicity, large glial cells and capacity for survival when isolated from the animal. The physiological observations made in Necturus have provided a model for other vertebrate glial preparations (Somjen, 1979).
For a number of years some of his naïve students thought that Steve ‘s remarkable! knack for choosing pregnant biological problems had been transiently inhibited in the early 1960s. However, the varied and interesting studies summarized in the present volume provide ample evidence that this was not the case. The scope of likely glial functions is growing rapidly as work on the physiology of neuroglia increases.
The optic nerve of Necturus is a naturally isolated central-nervous-system tract surrounded by a pia-arachnoid layer containing the blood supply (Bracho, Orkand & Orkand, 1975 ; Kuffler et al. 1966). The nerve is about 100 μm in diameter and about 1 cm long. It consists of large glial cells and small axons (<2 μm diameter). Intracellular electrodes can readily be inserted to record the membrane potential of the glial cells either in situ with the blood circulation intact, or in the isolated desheathed preparation. Since blood vessels do not penetrate the nerve, the pia-arachnoid containing them can be dissected away leaving the glial cells in normal relation with the axons and with the external environment. As seen in Fig. 1, the glial processes course through the nerve forming thin sheets that separate groups of axons into bundles. At places where glial processes come close to one another they form frequent desmosomal type junctions and presumed gap junctions. Specialized junctions are not observed between glia and axons. At the surface of the nerve the glial processes form a glia limitons which is interrupted by narrow intercellular clefts. A basal laminal surrounds the nerve.
After eye removal the axons of the optic nerve undergo ‘Wallerian’ degeneration. During axon breakdown membrane-bound electron-dense inclusion bodies appear in the glial cytoplasm, suggesting that they phagocytize the axonal remnants. There is a striking hypertrophy of the glial cytoplasm whose constituents appear essentially normal except for the addition of some phagosomes and lipid droplets as shown in Fig. 1. In the period 2–3 months after enucleation the optic nerve consists of relatively normal appearing astrocytic glial cells (Orkand, Bracho & Orkand, 1973). This preparation has proved a useful one in which to study the properties of isolated neuroglia (Cohen, 1970; Orkand et al. 1973; Salem et al. 1975; Tang, Strichartz & Orkand, 1979). Three months to a year after axonal degeneration the glial cells show some signs of disintegration. Increasing numbers of dense inclusion bodies and lipid droplets appear, the peripheral glial processes become narrower and more tortuous, the perinuclear cytoplasm decreases in volume, and the whole nerve becomes smaller in diameter. The glial cells progressively atrophy. Thus the long-term maintenance of the astrocytes seems to depend on the presence of axons.
Morphometric studies of the Necturus optic nerve indicated that only about 10% of the glial membrane faces the pia-arachnoid space while the remainder is surrounded by a system of narrow, 100–200 Å wide, extracellular clefts separating glial membranes and glia from axons (Dolack, Tang & Orkand, 1979). Physiological experiments have demonstrated that the ionic composition of these clefts under rest conditions is the same as that in the surrounding space (Kuffler et al. 1966). That is, the system of extra-cellular clefts freely communicates with the pia-arachnoid space ; there is little if any hindrance to free diffusion for ions and small molecules such as sucrose. Fig. 2 illustrates an experiment which demonstrates the rapid exchange of sodium for sucrose through the cleft system. The dependence of the axon action potential on external sodium was used to indicate the exchange of extracellular sodium. Replacing sodium in the external solution by sucrose rapidly and reversibly blocked action potentials in the axons. Since the resting potential of the glial cells (which depends on a high [K+]i) was unaffected by this exchange it was concluded that the exchange of sodium occurred by diffusion through the extracellular clefts rather than through the glial crytoplasm. In later experiments (Tang et al. 1979) it was found that sodium channel blockers like saxitoxin rapidly and reversibly diffuse in to block the nerve action potentials.
Cohen, Gerschenfeld & Kuffler (1968) studied the question of the relative contribution of the blood plasma and the cerebrospinal fluid to the ionic environment of the glial cells. By maintaining Necturus in water with elevated [K+] for a few days it was possible to raise the plasma [K+] with only a minimal effect on cerebrospinal [K+]. Using the membrane potential of the glial cells as an indicator of [K+]o it was found that the [K+]o was determined by that of the cerebrospinal fluid rather than that of the plasma. The functional ionic environment of the optic nerve may differ significantly from that in the plasma. Apparently control mechanisms exist which maintain the [K+] in the environment of the neurones and glial cells within relatively narrow limits despite fluctuations in plasma [K+].
Passive electrical properties
Current injection into a glial cell through an intracellular electrode leads to glial potential change over a considerable length of the optic nerve. With increasing distance from the site of injection the potential decreases in amplitude and has a slower time course. The relation between the amount of current injected, hyperpolarizing or depolarizing, and the recorded steady-state change in voltage is linear for membrane potential changes of up to 100 mV (Kuffler et al. 1966; Cohen, 1970). This result indicates that over the range of voltages studied the glial membrane behaves as a linear resistance and capacitor in parallel. In marked contrast to excitable membranes of neurones, the glial membrane conductance is not voltage- or time-dependent. The passive electrical properties of the glial membrane have been determined from steady state measurements of the distribution of membrane potential change produced by intracellular current injection with intracellular potential recording at various distances between the two electrodes (C.-M. Tang, unpublished observations). The optic nerve was modelled as a finite one-dimensional cable. The electrical length constant, A, determined in this way is 800 ±200 μm. Preliminary estimates of the specific glial membrane resistance are of the order of 40,000 Ω cm2, which is considerably higher than that of neurones.
Dye-coupling and uncoupling
The measurements of the electrical length constant of the glial cells show that ionic current readily flows from one astrocyte to its neighbours. More recently, it has been possible to demonstrate that the fluorescent dye, Lucifer Yellow CH, mol. wt. 457, readily diffuses from one glial cell to another (Gutnik, Connors & Ransom, 1981; C.-M. Tang, P. M. Orkand & R. K. Orkand, in preparation). Fig. 3. illustrates the distribution of the dye a few minutes after intracellular injection. It can be seen that a number of glial nuclei are stained following intracellular injection. If the glial cells are perfused with Ringer ‘s solution made acidic (pH 6·5 or below) by either bubbling with CO2 in the presence of 15 mm-HCO3− or by replacing an equivalent amount of NaCl with 50 mm sodium propionate the cells uncouple (Spray, Harris & Bennett, 1981). As shown in Fig. 3, injected uncoupled glial cells have very much the same form as cells injected with the protein horseradish peroxidase which is known from other studies not to traverse intracellular gap junctions (Takato & Goldring, 1979). The uncoupling of glial cells in acidic media is of particular interest because it suggests that glial cell communication may be interrupted under conditions of severe respiratory acidosis.
Permeability to monovalent ions
The normal high resting membrane potential (about − 90 mV) of optic nerve neuroglia arises from the diffusion gradient for K+ across the cell membrane (Kuffler et al. 1966; Bracho et al. 1975). Over a wide range of [K+]o (2–110 mequiv/1) the membrane potential is accurately predicted by the Nernst relation assuming [K+]i about too mequiv/1 (Fig. 4A). Attempts to demonstrate a measurable effect of other naturally occurring monovalent ions (i.e. Na+ and Cl−) on the potential have not been successful. For other monovalent cations the permeability characteristics of the glial membrane are similar to that of the potassium channel of the neurone. The selectivity fingerprint for the glial membrane is shown in Fig. 4B. From the observation that T1+ is much more permeable than K+ and that NH4+ is less permeable than Rb+ it is clear that properties other than that of the diameter of the unhydrated cation determine the permeability of an ion in the glial membrane channel.
A study of the temperature dependence of the glial membrane potential is of interest not only because it serves as a test of the applicability of the Nernst relation for potassium but also because it can indicate if active transport processes contribute to the resting potential. In those systems where electrogenic transport contributes to the resting potential, a decrease in temperature produces a greater depolarization than that predicted from the Nernst relation (Carpenter & Alving, 1968), The slope of the relation between change in temperature and glial membrane potential is about 3 mV/10 °C for temperatures between 1 and 30 °C, i.e. precisely as predicted for a K+ diffusion potential (Bracho et al. 1975). Raising the temperature above 30 °C leads to a progressive decrease in membrane potential presumably because the membrane loses its selective permeability for K+ at these temperatures.
Attempts to demonstrate voltage-sensitive sodium channels in vertebrate glial cells have been fruitless. The current-voltage relation of the glial membrane is linear over a range of ±100 mV, indicating that the overall membrane conductance is not voltage dependent. Furthermore, although the normal optic nerve demonstrates a saturable uptake of saxitoxin, which binds selectively to sodium channels, as well as depolarization in response to low concentration of veratridine, which opens sodium channels, both of these responses disappear when the axons disappear 2 months following enucleation (Tang et al. 1979). In normal optic nerves there is a saturable uptake of saxitoxin which is equivalent to a density of sodium channels of about 25 μm−2 of axon membrane, a value similar to that found in other vertebrate unmyelinated axons. After the axons were allowed to degenerate, and the optic nerve consisted only of glial cells, there was no significant saturable component of saxitoxin binding. Normal nerves are depolarized by concentrations of veratridine of 10−8 or 10−5 M. This depolarization is tetrodotoxin or saxitoxin sensitive and appears to arise from the effect of veratridine to open axonal sodium channels (Fig. 5). All glial nerves are not depolarized by such low concentrations. High concentrations of veratridine (10−2 to 10−3 M) do depolarize isolated glial cells. However, this depolarization is not sensitive to the sodium channel blockers and its basis is unknown. If one records intracellularly from a glial cell in a normal nerve during application of veratridine in low concentration a glial depolarization results. This depolarization appears due to a release of potassium from the depolarized axons rather than to a direct effect on the glial membrane.
In addition to veratridine, optic nerve neuroglia in the normal nerve are depolarized by exposure to glutamate (10−6 M). This glutamate depolarization is still observed in the isolated glial cells. At this time we know little of the properties of glutamate receptors on glial cells or the nature of the permeability changes induced by glutamate in the glial membrane.
Electrogenic sodium transport
The sodium permeability of the glial membrane is apparently low compared to that in neurones (Kuffler et al. 1966; Bracho et al. 1975). Nevertheless, there is presumably some inward leakage of sodium which must be balanced by active transport. If a glial nerve is bathed for some hours at low temperature in a potassium- deficient medium the recorded resting potential does decline (Tang, Cohen & Orkand, 1980). This depolarization probably results from a loss of internal potassium (presumably in exchange for sodium) (Keenan & Niedergerke, 1967). When such a potassium-depleted glial cell is exposed to normal potassium the membrane potential will either increase or decrease less than that expected from the normal potassium sensitivity of glial membrane (Fig. 6). Pretreatment of a potassium-depleted glial cell with the sodium-pump inhibitor strophanthidin or replacement of sodium for lithium in the external bathing solution will restore the normal potassium sensitivity. These results suggest that the addition of normal potassium to a potassium-depleted glial cell activates a hyperpolarizing electrogenic pump. The electrogenic pump in glial cells, like that in neurones, appears to be stimulated by an elevated internal sodium and requires external potassium. Under more normal conditions, without pretreatment in potassium-depleted solutions, we have been unable to demonstrate an electrogenic component of the membrane potential either by the addition of strophanthidin or by varying temperature.
Depolarization with neuronal activity
Nerve impulses lead to glial depolarization as a result of an accumulation of K+, lost from active axons, in the narrow intercellular cleft separating axons from glial cells (Orkand, Nicholls & Kuffler, 1966). As shown in Fig. 7, with a single nerve volley, the glial depolarization rises to a peak of about 3 mV in 50–150 ms and decays with a half-time of about 2 s. Repetitive nerve volleys lead to a nonlinear summation of glial depolarization as expected from the nonlinear relation between glial membrane depolarization and [K+]o. The observed glial depolarization is consistent with a potassium efflux from the axons of about IpM/cm2 into a maximum intercellular cleft width of about 250 A. Since the introduction of potassium selective electrodes the increase in [K+]o with nerve impulses has been directly measured in the amphibian spinal cord (Sykova et al. 1976). Glial depolarization with neuronal activity is readily observed in the optic nerve under conditions of natural stimulation of the retina with light (Orkand et al. 1966). Glial depolarization by raised [K+]o may participate in the exposed role of glial cells to equilibrate [K+]o in various regions of the nervous system (Orkand et al. 1966; Gardner-Medwin, 1981).
Contribution to surface recording
When an optic nerve is placed across a sucrose gap a significant fraction (30–60%) of the change in a glial membrane potential produced either by passing current across the glial membrane or by changing [K+]o may be recorded extracellularly across the gap. (Fig. 8). The glial contribution to the surface potential is a function of the relative volume of neuronal and glial tissue and the ease with which current spreads through the glial syncytium (Cohen, 1970). This result is of interest for the interpretation of electrical recordings made from the surface of active nervous tissue. As indicated above, the glial membrane potential undergoes slow potential changes with changes in [K+]o due to K+ efflux from active neurones. With maximal axonal stimulation the potassium concentration might increase to a level of 20 mm leading to a glial depolarization of almost 50 mV (Orkand et al. 1966; Sykova & Orkand, 1980). During the recovery phase after such stimulation the axons undergo a marked hyperpolarization due to activation of their electrogenic pump while the glial depolarization slowly subsides (Tang et al. 1980). Under these conditions the surface recorded potential is a complex function of the opposing potential changes going on in both neurones and glia.
Control of glial metabolism
Hypotheses for a functional role of neuroglia in the behaviour of the nervous system usually require coordination between neuronal and glial activity. Therefore, it is of some interest that it has been possible to demonstrate that increases in [K+]o of the same order as that which occurs as a result of potassium efflux from active neurones are sufficient to alter metabolic activity in glial cells. In the all-glial optic nerve, it was possible to demonstrate (Fig. 9) that the levels of reduced pyridine nucleotide are decreased in intact cells by the addition of potassium to the external bathing solution (Orkand et al. 1973), and that glucose uptake is increased (Salem et al. 1975) What remains is the more difficult task of defining which metabolic reactions an effected and demonstrating how these changes contribute to the functional activity of the glial cells.
Supported by U.S.P.H.S. No. NS-12253.