Brain Research, 109 (1976) 311-322

311

© ElsevierScientificPublishingCompany,Amsterdam- Printed in The Netherlands

GLIAL CELLS AND EXTRACELLULAR POTASSIUM: THEIR RELATIONSHIP IN MAMMALIAN CORTEX

K I N J. F U T A M A C H I * AND T I M O T H Y A. PEDLEY**

Department of Neurology, Stanford University School of Medicine, Stanford, Calif. 94305 (U.S.A.) (Accepted November 3rd, 1975)

SUMMARY

Simultaneous recordings were made of glial cell potentials and the extracellular potassium concentration ([K+]o) in cat cortex in an attempt to provide more quantitative information about the sensitivity of mammalian neuroglia to changes in [K÷]o. A penicillin epileptogenic focus served to generate both transient and sustained elevations in [K+]o, thus allowing measurement of glial membrane potential (Vm) at both resting and increased [K÷]o levels many times during the same experiment. Resting Vm averaged --92.6 ± 10.9 mV for 33 glial cells. With each surface interictal spike, glial cells exhibited slow depolarizations averaging 18.4 -4- 6.5 mV which mirrored rises in [K+]o in many respects. Several discrepancies were found, however, between transient and focal rises in [K~]o and the associated glial cell depolarizations which made it difficult to determine accurately the effect of changes in [K*]o on glial Vm. For example, the amplitude of the glial depolarization caused by a single interictal discharge showed no constant relationship to depth below the cortical surface in contrast to the consistent laminar profile recorded by the K + electrode. Thus, large glial membrane depolarizations could be recorded at times when there was little or no increase in measured [K+]o. Agreement between changes in [K+]o and glial cell depolarizations was closer to that predicted by the Nernst equation during sustained elevations in [K+]o such as occurred during ictal episodes ('seizures'). These findings may be related in part to methodology as a consequence of the different spatial relationships which exist between glial membrane, K+-electrode tip and released K +. In addition, though, they may indicate the presence of a functional glial syncytium.

INTRODUCTION

Glial elements comprise the vast majority of cells within the mammalian central * Present address: Laboratory of Neurophysiology, NINCDS, NIH, Bethesda, Md., U.S.A. ** To whom reprint requests should be made at the above address.

312 nervous system, outnumbering neurons by as much as 8 to one s . Experiments in leech and amphibia have shown that these cells share a number of properties m common including (1) high resting membrane potentials which appear to be exclusively determined by the membrane potassium concentration gradient (i.e., their resting membrane potentials are at Ek, the potassium equilibrium potential); (2) membrane depolarizations which obey the Nernst equation for increases in the extracellular potassium concentration ([K~]o); (3) absence of spike generation; and (4) low resistance interconnections resulting in the possibility of electrical interaction within a glial syncytium (see Kuffler and Nicholls for reviewl~). Though mammalian glia have been less thorough investigated, largely because of technical difficulties, a number of studies have confirmed their physiological similarity to gila in simpler models. Recent experiments have suggested, however, that mammalian glia may be less sensitive to changes in [K~]o than those in lower animals. Thus, a 10-fold change in [K~]o has been reported to result in a glial depolarization of 30 mV (ref. 18) and 38 mV (ref. 22) in cat cortex, and 42 mV in rat optic nerve 4. If these estimates are correct, the resting membrane potential of mammalian glial cells would appear to depend partially on the transmembrane gradient for ions other than potassium. The use of ion-sensitive microelectrodes has made it possible to measure cortical K + activity directly. In the experiments reported here, we simultaneously recorded glial cell potentials and extracellular potassium activity to provide more quantitative information about the sensitivity of cortical glial cells to changes in [K ~]o. MATERIALS AND METHODS

Twenty-two adult cats were lightly anesthetized with a single dose of thiopental (28 mg/kg). A tracheotomy was performed and the skeletal musculature paralyzed with gallamine triethiodide (5 mg/kg). Bilateral thoracotomies were made and the cisterna magna drained to reduced brain pulsations. Body temperature was maintained at 37 °C with a heating pad, and 5 ~ dextrose in normal saline was constantly infused at 3 ml/kg/h through a catheter placed in the femoral vein. The bone and dura overlying pericruciate cortex were removed, and warmed, artificial cerebrospinal fluid (CSF) was perfused over the surface of the exposed brain. All wound edges and pressure points were liberally infiltrated with a long-lasting local anesthetic. K ~ electrodes with time constants of 2-3 msec were constructed as described by Prince et al. ~°. The ion-sensitive electrode was glued to a reference microelectrode so that the tips were parallel and less than 50 #m apart. Differential recording between these pipettes using a very high-input impedance differential amplifier (MetaMetrics model AK-47, input impedance 10la ~ ) eliminated contamination of the K ÷ signal by local EEG transients or DC shifts. Intracellular microelectrodes with unusually long shanks were filled with 3 M potassium acetate which contained 1 0 ~ potassium chloride. Usable electrodes had impedances of 40-60 MD. A 3-electrode array was then assembled as follows: the glued K + reference pair was fixed in a holder which allowed independent manipulation of a third electrode. Using a dissecting microscope fitted with an ocular micrometer, an intracellular microelectrode was then positioned

313

Fig. 1. A: photograph of electrode set-up used in these experiments. The chlorided silver wire protudes from the reference pipette; the potassium-sensitiveelectrode is glued to it and inserted into a sealed electrode holder. The intracellular microelectrode (smaller diameter pipette in foreground) is positioned in the groove formed by the other two electrodes. B: diagrammatic representation of the set-up shown in A.

in the groove formed by the shanks of the other two pipettes, so that the 3 shanks were parallel and snugly apposed, and the microelectrode tip was 30-70/~m below the tip of the K + electrode. This arrangement permitted replacing broken or clogged microelectrodes without necessarily discarding good K + reference pairs. The electrode set-up is shown in Fig. 1. The tips of the 3 electrodes penetrated the brain normal to its surface through a 2-mm opening in a small lucite pressor plate placed lightly on the pial-cortical surface. A silver-silver chloride ball in the base of the pressor foot recorded the DC-coupled E E G activity. A bipolar stimulating electrode made from a pair of stainless steel wires with tip separation of 1 mm and insulated except for 0.5 mm at the tips was stereotaxically placed in the ipsilateral ventral lateral (VL) nucleus of the thalamus to provide a specific orthodromic input to pericruciate cortex. Stimulus pulses were delivered through a constant current isolation unit. Recorded signals were displayed on an oscilloscope and stored on magnetic tape (bandpass DC - - 5 kHz). Data was reproduced on a Brush model 260 inkwriter. As an alternative to K + superfusion, we elected to use a penicillin epileptogenic focus produced by pial application of buffered, isotonic sodium penicillin G (80,000 I.U./ml) as a means of generating both transient and sustained elevations of [K+]o5,7,17. This had the advantage of allowing measurements ofglial Vm at both resting (3 mEq/l) and increased [K+]os many times during the same experiment.

314

A

B

EEG

25mV

GL,A

~--'"-'-~ ] i mEq~

[K+Jo REF ~

'p"'--

~

~

5 SEC

~

15mV

1 SEC

C EEG

-

I

GLIA

~

. ~ ~ J rllllll

J]rnV

l

..... +.100mV ............

rK o

I I tL till t ittltt t t | , , , , , , . ~ [llflrlTU[lU++uulrr~+ ..... ". . . . . ~ '

20mV +',

'. . . . . . . . .

.

"

.

.

.

J3

+

5 SEC Fig. 2. Simultaneous [K+]o and glial cell responses. Polarities: EEG, glial cell and local DC reference potential (REF), positivity up; [K+]o, increase up. Surface EEG is DC-coupled; reference signal is recorded against ground. A: individual interictal events, one of which is shown in B at a faster paper speed to demonstrate the differences in rise times. D e p t h - 1750k tin. C: summating triggered interictal responses in a cell (depth = 600 Fm) illustrating sustained 'plateau' of peak K ÷ response and glial cell depolarization. Dots below traces in C indicate stimuli to ipsilateral VL thalamic nucleus.

RESULTS

Glial cells and [K+]o Glial cells were identified by their high resting membrane potentials; lack of injury discharge on impalement; absence of spontaneous spiking; absence of spiking during intracellular depolarizing current pulses; and the presence of slow depolarizing responses during interictal epileptogenesis 19. We arbitrarily excluded from analysis cells with resting membrane potentials less than 70 mV. The basis for this was the assumption that cells with lower membrane potentials had probably been injured by the electrode and were likely to have both spuriously low internal potassium concentrations ([K+]i) and falsely high permeabilities to ions other than K +. Thirty-three glial cells were available for detailed study. Resting Vm averaged --92.6 ± 10.9 mV (range --70 to --110 mV). Characteristic slow depolarizations averaging 18.4 ± 6.5

315

A

B ~

EB3

~

~

i

~

~

.

:-¢-

aa

GLIA

i ~:'-J 2mY

_a L -

[K+l, i

__:.~

REF

~

'

r--/"-

~

'

Y~l

1°my

~SSEC Fig. 3. Summating interictal responses and subsequent ictal episode in a glial cell 800 #m below the pial surface. Interval between A and B is 20 sec. Following repolarization, the glial cell hyperpolarizes in association with an undershoot of the [K+]o below the resting level. The sharp downward deflection seen at the beginning of some of the glial cell responses is due to current flow in the extracellular fluid. Dots below traces indicate stimuli to ipsilateral VL thalamic nucleus. EEG is RC-coupled in this example.

mV occurred in the glial cells coincident with each interictal spike discharge recorded f r o m the cortical surface. Paralleling these events were changes in local [K+]o (Fig. 2A). During trains o f triggered epileptiform discharges or ictal episodes ('seizures'), both glial cells and the simultaneous K + responses rapidly reached plateau levels which were sustained for varying periods o f time (many seconds) before recovery (Figs. 2C and 3). The extracellular potassium concentration often fell transiently below baseline levels following the sustained elevations which accompanied ictal episodes. Glial

A [K+]o

i

,

i, !N!x~i'~i"i

j x,J

~'

~, ,,

'i:

3

20 18 17 13 13 12 14 10 8 10

1.6 1.4 1.5 1.4 1.4 1.8 2.2 1.9 1.2 1.5

5 SEC

Fig. 4. Summating triggered interictal responses in a cell 2000/~m below the cortical surface. The table to the right of the traces gives the actual magni*ude of the numbered incremental changes in Vm and [K+]o.

316

A 4.0

2

i z~ [K+] o VS DEPTH

3.6

36

3.2

32

2.8

28

2.4

24

z~ V m ~/$ DEPTH

•

A .J

O

•

•

ILl

20 I:

2.0 O

>E

•

,~,i.8

:

•

-I"~ 16

•

C

"

•

•

:

•

1.2 0.8

8 t-

• .

•

]

•

0.4

I

•

0

i

[

0.4

• t

I

I

0.8

I

1.2

I

I

1.6

l

I

2.0

I

I

2.4

I

4

I

t

0

2.8

L

0

I

0.4

I

i

I

0.8

I

1.2

I

~

1.6

I

I

2.0

I

~

2.4

I

I

I

J

2.8

DEPTH (MM)

B

50

[K']o----

500

1000

1500

2000

-/ 20 mV

5 SEC Fig. 5. A: plots of simultaneously recorded A[K+]o (1) and AVm (2) v e r s u s depth for interictal responses in 33 glial cells. Each dot may represent more than one point. B: laminar profile showing the marked discrepancy between the size of the glial cell depolarization and simultaneous rise in measured IK+]o. Numbers above each response indicate the depth in micra below the cortical surface.

cells simultaneously hyperpolarized with a time course similar to the measured change in [K+]o (Fig. 3). The magnitude of the glial depolarizations was poorly correlated with the resting I'm. The amplitude of successive glial depolarizations occurring during closely spaced interictal discharges showed successive decrements, while the [K+]o level during such a sequence rose in approximately equal steps (Fig. 4), This relationship would be

317 I

-I

I

I

I

A [ K I o vs V m SLOPE

,-I

tu

~0

"g

=~

10 8

e

•

° l@-

4

oo.,~o

o

/=" °=o° "~ 0

~" ¢~3

0

o

0

0

I

I

I

I

10

20

30

40

50

AVm (mY)

Fig. 6. Semi-log plot of individual interictal (circles) glial cell responses (A Vm) as a function of the simultaneously measured A[K+]os. Each circle may represent more than one point. The maximal glial depolarizations (A) for ictal episodes in 10 cells are also shown as a function of peak A[K+]oS. expected if changes in [K+]o were logarithmically related to the glial Vm and is in agreement with results reported by Ransom and Goldring 22. No change in membrane conductance occurred during slow depolarizations as judged by responses to injection of constant current pulses, confirming earlier observations that the glial cell membrane behaves in a passive manner to changes in external potassiumZl, 12. While the glial cell responses appeared closely related to the changes in [K+]o recorded by the ion-sensitive electrode (compare glial and K ÷ responses in Fig. 2A), we observed several important differences. Though there was no statistically significant difference in the fall times of [K+]o changes and glial cell depolarizations produced by a penicillin spike, the rise times were strikingly different (Fig. 2B). The glial cell response was over 4 times faster (1/2 rise time 49 4- 14 msec) than the associated increase in [K+]o (1/2 rise time 230 q- 80 msec) and appeared to be independent of the depth of impalement. The fall times, on the other hand, are not significantly different (glia I/2 fall time 870 :]: 300 msec and K + electrode 1/2 fall time 960 -I- 250 msec). Another major difference between changes in glial Vm and measured [K+]o is illustrated in Fig. 5. The amplitude of the A[K+]o during interictal discharge varied consistently with depth below the cortical surface, but no similar correlation could be determined for the glial responses. For example, large glial depolarizations could be seen superficially at times when no or minimal [K+]o changes could be detected. In order to examine the relationship of [K+]o to glial Vm, we made a semi-log plot of A[K+]o for representative individual interictal responses in 33 cells (Fig. 6)*. * The extracellular field potential recorded immediately after losing a glial cell was consistently found to be very much smaller and much faster than the glial response. Accordingly a correction for the extracellular field was not made for interictal events. During ictal episodes, however, a sustained DC shift of approximately 5 mV occurs and a correction for this must be made in determining the actual transmembrane potential.

318

A EEG

NEURON

5

t4mM J3

'

REF •

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

0•

•

• • • • • • o • o o

1 SEC

NEURON

[K+]o 250 mSEC

Fig. 7. A: intracellular recording from a neuron during interictal epileptogenesis with the simultaneously measured changes in [K+]o. An expanded trace is shown in B. Dots below traces indicate stimuli to ipsilateral VL thalamic nucleus. In both A and B it is apparent that each orthodromic stimulus generates an EPSP. Measurable [K+]orises and paroxysmal depolarization shifts occur only with surface interictal discharges.

I f the glial cell is behaving as a K + electrode, a given increase in [K÷]o should produce a change in membrane potential which, when plotted, would fall on a line having a slope of 61 mV per decade change of [K+]o. It is apparent from Fig. 6 that virtually all of the interictal changes (©) fall below this line (slope > 61 mV). However, if [K+]o and Vm are measured at plateau levels achieved during seizures or trains of interictal discharges, the points (Jk) relating AVm and A[K+]o cluster about a line with a slope of 61 mV (with one exception, see below). D a t a were available from 10 cells at depths from 400 to 2000 # m during such plateaus. The mean slope for the cells analyzed in this way was 63.4 -4- 17.5 mV. Underlying these determinations is the assumption that the [K+]~ of a glial cell remains constant, a necessary requirement if Vm is related to [K÷]o in a manner predicted by the Nernst equation. One of the 10 cells had a significantly lower slope (about 30 mV) which is partially responsible for the large standard deviation. This cell, while extremely stable, had a lower resting potential (--74 mV) than the others (average - - 9 8 mV). Whether this reflects un-

319 recognized damage to the cell, resulting in a smaller K+-induced depolarization, or whether this cell is representative of a population which has different physiological properties cannot be determined from these experiments.

Neurons and [K+]o During the course of these experiments, neurons were frequently and readily impaled (Fig. 7). Though these were not studied in detail, several observations were consistently made. Rises in [K+]o were associated with each paroxysmal depolarization shift (PDS). The peak of the [K+]o increase followed the onset of the PDS by 300-500 msec. Individual spikes and EPSPs were not accompanied by significant changes in [K+]o as recorded by the K + electrode.

DISCUSSION

Pape and Katzman is and Ransom and Goldring22 have reported that mammalian glial cells have slopes considerably lower than the theoretical 61 mV per 10-fold change in [K+]o predicted by the Nernst equation. In contrast, Lothman and Somjen15 have more recently recorded glial cell potentials and [K+]o simultaneously in cat spinal cord and have determined that measured changes in glial Vm are fully explained by the Nernst equation for K +. Since the earlier experimentsis,z2 relied upon perfusion of the cortical surface to change the [K+]o bathing impaled glial cells, it is possible that incorrect estimates of [K+]o in the depths led to errors in determining the effect of [K+]o on glial Vm. For example, the presence of a barrier to diffusion at the surface or the use of an incorrectly high intracortical diffusion coefficient for K + would result in overestimation of [K+]o and a spuriously low slope n. In addition, it is clear that superfusion does not mimic the profiles obtained when increases in [K+]o are internally generatedS, 17. The results of our experiments do not resolve these differences. Though fluctuations in glial membrane potentials are similar in many ways to changes in [K+]o, we discovered several significant discrepancies between direct measurement of extracellular [K+]o and the simultaneously recorded changes in I'm. The most striking observation in the work reported here was the apparent lack of relationship between A I'm and A[K+]o for single interictal events (Fig. 6, circles). This is also shown in Fig. 5 where it is clear that the amplitude of the glial depolarization is not related in a readily discernible way to depth below the pial surface, in marked contrast to the consistent laminar profile recorded by the potassium microelectrode. Though a definitive explanation for these discrepancies is not possible from our experimental data, several factors may be contributory, and each is considered in detail below. It is possible that a glial cell may reflect [K+]o changes at sites remote from its own cell body. This could occur if cortical neuroglia were linked through low-resistance connections similar to those which exist in leech, Necturus and frog optic nervO z,lz. In these lower animals, a glial syncitium allows passive and extensive current spread from one glial cell to another. Though direct physiological evidence

320 is not available for a similar syncytium in mammalian cortex, it is probable that such a network exists. It is now well established that depolarizing responses in glial cells, in contrast to neuronal intracellular potentials, are the major contributors to the DC shifts recorded at the cortical surface'-', 21. The presence of a glial syncytium may provide the necessary substrate for this phenomenon as it is unlikely that small giial cells, electrically independent, could generate sufficient current flow to produce voltage changes detectable at the surface 3,11. In addition, fused membrane contacts, the likely anatomical substrate for a functional syncytium, have been found between cortical glial cells 1,9. Thus, a spatially restricted increase in [K~]o occurring at some distance from the recording electrode pair might produce little measurable change in local [KF]o but still result in a significant membrane depolarization of the impaled glial cell. However, if electrotonic current spread between glial cells were a significant factor contributing to the poor correlation between A[K+]o and AVm, one would expect to see slower rise times and lower amplitudes of glial cell depolarizations tile greater the distance from the source of maximal K ~ release 2,H,1". In fact, we found that some of the largest glial responses occurred at the cortical surface where little or no rise in [K ~]o was measured. We considered that penicillin might be affecting the amplitude of the depolarizations in superficial glial cells in some way, but no information is available on this point. It is possible that variability in the interictal glial and [K+]o responses may reflect in part basic physiological differences in a heterogenous population of glial cell typesl0, ~5 (e.g., astrocytes versus oligodendrocytes), as to date electrophysiological data have not been adequately correlated with the different kinds of neuroglia. It has also been suggested that glial cells are sensitive either to ions other than K ~- or to some additional and unknown changes in the extracellular microenvironment 4,1s,22. This should result in glial cells responding as if they had slopes less than the theoretical 61 mV. To the contrary, we found that during interictal events, neuroglial cells depolarized more than would have been predicted by the Nernst equation (Fig. 6, circles; see below). Perhaps the most important factor contributing to the discrepancies between A[K+]o and A Vm is a technical one. The methodology used in these experiments will invariably introduce a systematic error since it is obvious that different spatial arrangements exist between glial membrane, K+-sensitive electrode tip, and released K +. The glial cell membrane, separated by only 10-20 nm from a nerve cell which is releasing K + into the intercellular space, is in an optimal position to reflect rapidly small rises in [K+]o. The 2-#m tip of the K + electrode, however, is many times the size of an intercellular cleft, and of necessity has an artificial extracellular space of damaged tissue around its tip. Diffusion across this 10-15/~m dead space must occur before the ion-sensitive electrode will register changes in [K+]o. Thus, transient increases in [K+]o lasting less than 1 sec may not have time to equilibrate by diffusion across the 10-15 #m dead space around the electrode tip 5,16. The latter problem should be minimized during longer-lasting events which presumably are less subject to measurement error due to dead space. Two of our observations are consistent with this interpretation. First, glial cells were consistently found to depolarize faster than the

321 associated rises in [K+]o. A diffusional delay would act to slow the rise time and attenuate the peak of a transient potassium response. This would also contribute to recording glial depolarizations apparently larger than would be predicted from the simultaneously measured [K+]o. On the other hand, the fall times have a longer time course and are similar. This would be expected, assuming diffusion as the principal method for removal of K + from the extracellular spaces 6,16. Second, peak A[K+]oS and AVms measured during seizures, or long trains of triggered epileptiform discharges, are related in a way that is much closer to what is expected from the Nernst equation for K + (Fig. 6, A ) . Ransom found that glial cells in epileptogenic cortex had significantly higher resting membrane potentials than glia in normal cortex 21. He suggested that this might reflect a lower baseline [K+]o during interictal periods because of a net loss of potassium from the extracellolar spaces as a consequence of paroxysmal activity. Despite a glial cell population with comparable resting membrane potentials, we found that baseline [K÷]o was normal (3.12 ± 0.33 mEq/l) between interictal discharges. Our average Vm 0 f - - 9 2 mV is thus near Ek, assuming that [K+]i approximates 100 mEq/ 1 (see ref. 14). Transient lowering of [K÷]o below resting levels associated with glial cell hyperpolarization occurred only following periods of intense neuronal activity as has been reported previously 21,23,24. We conclude from these experiments that while mammalian cortical neuroglia behave like glial cells in amphibia and invertebrates in many respect, the interaction between glial cells and [K+]o appears to be more complicated in neocortex. It is our speculation that this may be due less to fundamental differences in the membrane properties of glial cells in different animal models (particularly as regards sensitivity to changes in [K÷]o), than to technical problems encountered because of the complexity of nervous system organization associated with phylogenetic advancement. ACKNOWLEDGEMENTS We thank Professor David A. Prince for constant support during these experiments and assistance in preparing the manuscript. Drs. Denis Baylor and John Nicholls provided many helpful discussions, and Ms. Geraldine Pickering provided secretarial assistance. Supported in part by USPHS Grants NS 06477, NS 12151 and NS 11075 from the NINCDS.

REFERENCES 1 BRIGHTMAN,M. W., AND REESE, T. S., Junctions between intimately opposed cell membranes in the vertebrate brain, J. Cell BioL, 40 (1969) 648-677. 2 CASTELLUCCI,V. F., ANDGOLDmNG,S., Contribution to steady potential shifts of slow depolarization in cells presumed to be glia, Electroenceph. clin. Neurophysiol., 28 (1970) 109-118. 3 COHEN, M. W., The contribution by glial cells to surface recordings from the optic nerve of an amphibian, J. Physiol. (Lond.), 210 (1970) 565-580. 4 DENNIS, M. J., AND GERSCI-IENFELD, M. H., Some physiological properties of identified mamma-

lian neuroglical cells, J. Physiol. (Lond.), 203 (1969) 211-222.

322 5 FISHER, R. S., PEDLEY, T. A., MOODY, W. J., JR., AND PRINCE, D. A., The role of extracellular potassium in hippocampal epilepsy, Arch. Neurol. (Chic.J, 33 (1976) 76 83. 6 FISHER, R. S., PEDLEY, T. A., AND PRINCE, D. A., Kinetics of potassium movement in noJ'mal cortex, Brain Research, 101 (1976) 223-237. 7 FUTAMACHI, K. J., MUTANI, R., AND PRINCE, D. A., Potassium activity in rabbit cortex, Braitt Research, 75 (1974) 5-25. 8 GLEES, P., The neuroglial compartments at light microscopic and electron microscopic levels. In R. BALAZS AND J. E. CREMER (Eds.), Metabolic' Compartmentation in the Brain, Halsted Press, New York, 1973, pp. 209-231. 9 GRAY, E. G., Ultra-structure of synapses of the cerebral cortex and of certain specializations of neuroglial membranes. In J. D. BOYD AND J. D. LEVER (Eds.), Electron Microscopy in Anatomy, Arnold, London, 1964, pp. 54-73. 10 KELLY, J. P., AND VAN ESSEN, D. C., Cell structure and function in the visual cortex of the cat, J. Physiol. (Lond.), 238 (1974) 515-547. 11 KUFFLER, S. W., AND NICHOLLS, J. G., The physiology of neuroglial cells, Ergebn. Physiol., 57 (1966) 1-90. 12 KUVVLER, S. W., NICHOLLS, J. G., AND ORKAND, R. K., Physiological properties of glial cells in the central nervous system of amphibia, J. Neurophysiol., 29 (1966) 768-787. 13 KUFFLER, S. W., AND POTTER, D . D . , Gila in the leech central nervous system: physiological properties and neuron-gila relationship, J. Neurophysiol., 27 (1964) 290-320. 14 LEES, M. B., AND SHEIN, H. M., Sodium and potassium content of normal and neoplastic rodent astrocytes in cell culture, Brain Research, 23 (1970) 280-283. 15 LOTHMAN, E. W., AND SOMJEN, G. G., Extracellular potassium activity, intracellular and extracellular potential responses in the spinal cord, J. Physiol. (Lond.), in press. 16 Lux, H. D., AND NEHER, E., The equilibration time course of [K+]o in cat cortex, Exp. Brain Res., 17 (1973) 190-205. 17 MOODY, W. J., JR., FUTAMACHI, K. J., AND PRINCE, D. A., Extracellular potassium activity during epileptogenesis, Exp. Neurol., 42 0974) 248-263. 18 PAPE, L. G., AND KATZMAN, R., Response of glia in cat sensorimotor cortex to increased extracellular potassium, Brain Research, 38 (1972) 71-92. 19 PRINCE, D . A . , Cortical cellular activities during cyclically occurring inter-ictal epileptiform discharges, Electroenceph. clin. Neurophysiol., 31 (1971) 469484. 20 PRINCE, D. A., LUX, H. D., AND NEHER, E., Measurement of extracellular potassium activity in cat cortex, Brain Research, 50 (1973) 489~-95. 21 RANSOM, B. R., The behavior of presumed glial cells during seizure discharge in cat cerebral cortex, Brain Research, 69 (1974) 83-99. 22 RANSOM, B. R., AND GOLDRING, S., Ionic determinants of membrane potential of cells presumed to be gila in cerebral cortex of cat, J. Neurophysiol., 36 (1973) 855-868. 23 RANSOM,B. R., AND GOLDRING, S., Slow hyperpolarization in cells presumed to be glia in cerebral cortex of cat, J. Neurophysiol., 36 (1973) 879-892. 24 SYPERT, G. W., AND WARD, A. A., Unidentified neuroglial potentials during propagated seizures in neocortex, Exp. Neurol., 33 (1971) 239-255. 25 WATANABE, S., MATARAI, G., AND TAKENAKA, S., The glial cell in the cerebral cortex of the cat, In Proc. int. Union Physiol. Sci., 24th Int. Congr., 7 (1968) 459.