Effects of insulin-induced hypoglycemia on intracellular pH and impedance in the cerebral cortex of the rat

Brain Research, 221 (1981) 129-147 Elsevier/North-Holland Biomedical Press

129

EFFECTS OF I N S U L I N - I N D U C E D H Y P O G L Y C E M I A ON I N T R A C E L L U L A R pH A N D I M P E D A N C E IN T H E C E R E B R A L C O R T E X OF T H E R A T

DALE PELL|GRINO, LARS-OLOF ALMQUIST and BO K. SIESJO* Laboratory for Experimental Brain Research, Research Department 4, E-blocket, University Hospital, S-221 85 Lund (Sweden)

(Accepted February 12tb, 1981) Key words: intracellular pH - - specific impedance - - insulin - - hypoglycemia - - extracellular fluid

space - - intracellular buffer base changes

SUMMARY In order to evaluate changes in extra- and intracellular p H in the brain during hypoglycemia rats were injected with insulin and p H changes evaluated when the EEG showed a slow-wave-polyspike pattern ('precoma'), or when E E G activity had ceased for 5, 15 or 30 min ('coma'). Extra- and intracellular acid-base changes were evaluated from pCO2 and HCO3- concentrations. In order to allow calculation of intracellular pH (and HCOa- concentrations) changes in extracellular fluid volume were estimated by measurements of cortical tissue impedance. The main results were as follows. (1) At constant arterial pCO2 the CSF HCO3- concentration either rose (15 min of coma) or remained unchanged (all other groups). However, since the cerebrovenous (and tissue) pCO2 fell, all groups except one (30 min of coma) showed a significant increase in extracellular fluid pH. (2) During severe hypoglycemia, and especially when EEG activity ceased, cortical impedance increased markedly. Calculations with the help of the Rayleigh and Maxwell equations showed that the extracellular fluid volume was reduced to about 50 ~'~,of control. (3) lntracellular p H increased significantly in precoma and in coma of 15 and 30 rain duration. However, p H in the 5 rain coma group was significantly lower (but no different from control). (4) In general, the increase in intracellular pH is consistent with previous findings that hypoglycemia is associated with oxidation of endogenous acid metabolites. However, the data suggest that in the initial period of coma acids accumulate by some unidentified mechanism.

*To whom correspondence should be addressed. 0006-8993/81/0000-0000/$02.50 © Elsevier/North-Holland Biomedical Press

130 INTRODUCTION A tight control of intracellular pH (pHi) is required for normal cellular function. Some of the pathology resulting from a variety of insults to the brain may, at least in part, be due to a failure of H + control mechanisms. During insulin-induced hypoglycemia, brain tissue metabolic acid levels become progressively reduced as blood glucose levels declineZ°,2L with a modest further reduction over time following the loss of spontaneous EEG potentials 3. Specifically, severe hypoglycemia is associated with reductions in lactate and pyruvate, most TCA cycle intermediates (except succinate and oxaloacetate), and a net reduction in amino acids plus a substantial increase in NH4 ~. This, therefore, should lead to a net loss of H ~ in the intracellular compartment. One can use the changes in the concentrations of the metabolites, from the studies cited above, to calculate the increase in cellular buffer base concentration as. Thus calculated the buffer base concentration is increased (up to 2.5 mM/1) in animals exhibiting a slow-wave EEG pattern (blood glucose ~ 2 /zmol/g) el and substantially more so (up to 8 mM/1) in animals with an isoelectric EEG (blood glucose < 1 #mol/g) zl. One, therefore, might expect to find an increased pHi in severe hypoglycemia, especially in the period following cessation of spontaneous EEG activity ('isoelectric EEG'). Lewis et al. 2° reported no change in cortical tissue phi, in hypoglycemic rats, at any given level of reduced blood glucose. This included animals exhibiting both a slow wave and an isoelectric EEG pattern. Intracellular pH was estimated using both the CO2 method and the creatine kinase (CK) equilibrium 37. However, these calculations did not take some important factors into account. Firstly, no allowance was made for a possible reduction in the extracellular fluid space (ECS). The ECS is an important component in the calculation of pHi using the CO2 method. It is quite likely that during severe hypoglycemia the volume of the extracellular fluid compartment will shrink for the following reasons. During hypoglycemia, energy stores become profoundly reduced following the onset of an isoelectric EEG 3,21,~. This, then, must lead to a reduction in cell membrane ion pump activity, as evidenced by a large increase in extracellular fluid K + (ref. 7). The reduced ion pumping will result in a n e t gain of ions in the intracellular compartment (primarily in the form of Na ~-and C1--) causing water to enter the cell from the extracellutar fluid 17,18,42,48. If applied to the study of Lewis et al. 20, a decreased ECS in the isoelectric period would result in higher values for pHi using the COz method. Secondly, tissue pCO2 values were estimated by adding 6 mmHg to the arterial pCO2. One can no longer assume that arterial and tissue pCO2 relationships 2s would remain constant since cerebral blood flow increases in severe hypoglycemia1, '4. Finally, the reduction in cellular ATP levels should effect the free Mg 2+ inside the cell, which could introduce an error in the calculation of pH~ using the CK equilibrium 33. In the present study, with the CO2 method, we have measured pHi and [HCOz-]i changes during insulin-induced hypoglycemia using a better estimate of the tissue pCO2. We have also included an estimate of ECS changes during hypoglycemia, based on changes in cortical tissue impedence. With some limitations (see Discussion),

131 impedance changes should be able to provide a reasonable estimate of changes in the ECS during hypoglycemia. MATERIALS AND METHODS

Animalpreparation. Male Wistar rats (225-400 g) were fasted 18-24 h prior to use, but were allowed water ad libitum. Thirty minutes prior to operation, they were given an intraperitoneal injection of insulin (Actrapid) at a dose of 40 l.U./kg. The insulin was dissolved in 0.75 ml of Krebs Hensleit solution. Control animals were given only 0.75 ml of this solution. Anesthesia was induced with 3 ?~i]halothane and the animals were tracheotomized, immobilized with tubocurarine chloride (0.5 mg/kg) and maintained with 70o/0 N20 30~o 02 on a respirator. The electrocortigram was continuously recorded through gold-plated copper bolts inserted in the skull, one over each hemisphere in the frontoparietal region. Two femoral arterial and one femoral venous catheters were inserted for monitoring of blood pressure, blood sampling and supplementary injections (if needed). Periodic arterial blood samples were taken into a glass capillary for analysis of pH, pO2 and pCO2. Animals were maintained at a p~CO~ of 35 40 m m H g and a paOz .> 90 mmHg. Body temperature was maintained at around 37 °C through the periodic use of a heating lamp. Experimental groups. There were two experimental series. The pH~ series consisted of 5 experimental groups with one control group (n = 5), one group where the animals were sacrificed at the end of 30 rain of slow-wave EEG activity (n -- 5), and 3 groups of animals which were sacrificed at 5 (n -- 6), 15 (n -- 5) or 30 (n - 5) min following the onset of an isoelectric EEG. At the conclusion of each experiment, while respiration was still maintained, the atlanto-occipital membrane was exposed and punctured with a fine-tipped glass capillary and 75-150 /~l of CSF were anaerobically withdrawn. A sagittal venous and a femoral arterial blood sample were then taken. Finally, a plastic funnel was fitted over the skull and the brain was frozen in situ with liquid nitrogen. Tissue samples were stored at - - 8 0 °C until analysis. The impedence series was comprised of just one group of animals (n -- 7). A craniotomy was made on one side of the skull over the parietal cortex and between the two EEG bolts. The animal's head was immobilized via placement in a stereotaxic head-holder. The dura was then observed through a high-power operating microscope and a small slit was made to accommodate the passage of the impedance electrode. The electrode was stereotactically inserted to 0.75 1.0 mm below the pial surface. The electrode was then left in place throughout the course of the experiment and the impedance was continuously monitored and recorded on a chart-recorder (Toa). Analytical techniques and calculations. The arterial and venous pO2, pCO2 and pH were measured with microelectrodes at 37 °C (Radiometer, Copenhagen and Eschweiler, Kiel). CSF and tissue pCO2 was taken as the arithmetic mean of the arterial and sagittal venous values plus l mmHgZ,~,2s. The total CO2 content of the CSF and cortical tissue was measured using a Conway microdiffusion method ~7. Arterial HCO3- and CSF pH were calculated from the Henderson Hassetbalch equation using the appropriate pK1 and CO2 solubility (S) factors'~2, 35. CSF HCO3- was

132

A Opto-isolated Current source

:;~ ...... ~__i~ r Recorder output

IReferenee

~gnal

mp.

~ Tissue

B

Ref. electrode

"

I Insulation

T i

2 mm

100 p 1 mm - ~ - ~

Fig. 1. The impedence measuring system. A block diagram of the various components of the system is shown in A (arrows indicate direction of signal). The electrode design and dimensions are represented in B. For a more detailed explanation, see text. e s t i m a t e d using the t o t a l CO2 c o n t e n t in the C S F a n d the tissue pCO~. W e calculated i n t r a c e l l u l a r H C O z - a n d p H a c c o r d i n g to the following e q u a t i o n s (see ref. 37): [HCO3-]i ==

Tco2 - - P t c o 2 " S 1 - - VEcv [HCO3-]csF - - Vbl [HCOa-]b~ Vi

(l)

133

pHi -- pK'I -- log

[HCO3-]i Ptco2"s2

(2)

where S~ is 0.0292 and $2 0.0314 mmol/kg/mmHg, Ptco~ is the tissue pCO2 (see above) : and VECF, VbXand Vi denote the extracellular blood, and intracellular volumes, respectively. VECF was assumed to be 20 ~i or 15 °/ounder control conditions and in the slow-wave EEG group. In the isoelectric groups, we utilized the per cent change of the ECS from the assumed normal values (either 20"~i or 15 °,o) as derived from changes in the specific impedance. As will be described below, the impedance value were applied to both the Maxwell and Rayleigh equations for estimating the volume fraction of a suspension of cells (see Results). The blood volume (Vbl) was taken as 3 °~o and the intracel]ular volume (Vi) was calculated according to the following: Vi ~-- total tissue H20 - - (VEcF + Vbl)

(3)

where total tissue water is 0.80 (fraction, by weight). Impedance measurement To measure cortical impedance in the present study, we have used the 4electrode technique as originally described by Ranck 31. The electrode design and a diagram of the measuring system are shown in Fig. 1. The system has the following characteristics. An oscillator produces a sinusoidal signal with a frequency of I kHz. This signal drives a controlled current source which feeds the outer electrodes in a linear 4electrode configuration. Optical coupling and battery operation are used to reduce leakage currents to earth to vanishingly small values. The voltage-drop between the two inner electrodes is directly proportional to the impedance between these electrodes, provided that the stimulating current is held constant and the current through the voltage-electrodes is zero. If this current is not zero, an interface-impedance will arise between the metal electrode and the brain tissue, causing a false impedance reading. The voltage-electrodes are therefore connected to a pair of very high impedance (1013 ~) buffer amplifiers (Burr-Brown, model 3528) with a grid current of less than 3 ~ I0 -la A. Their outputs are fed into a differential amplifier with a gain of about 1000. After passing the amplifier-section the signal filtered by two active filters, one highpass- and one lowpass-filter, together making a bandpass-filter with a gain of I in the passband, and a damping factor of 80 dB/decade in the attenuation region. The filtered signal is amplified and brought on to the phase-sensitive detector. The internal oscillator supplies its signal to a phase-shifter. This provides a reference output whose phase, with respect to the input, is continuously variable through 360 ~. The phase-sensitive detector can be described as a synchronous rectifier whose output is proportional to the amplitude of the input signal and the cosine of the phase angle between the input signal and the reference signal. Thus, if either the magnitude or phase of the input is changed, the output changes in accordance with the following relationship : Eout -- Eincos0.

(4)

134 That part of the signal which occurs at some frequency other than the reference frequency, or with varying phase angle, is averaged out. The time constant of the lowpass filter is chosen for the best compromise between maximum noise rejection and speed of response to input signal variations. The final section of the system is a DC amplifier stage, which has a variable amplification for calibration adjustment and range selection. Some of the advantages with the designed brain impedance measuring system are the following. (1) There are negligible polarization artifacts due to the electrode. This is achieved by using a 4-electrode probe, where a sinusoidal current of constant low density and suitable frequency is injected into the outer electrodes, while the potential is measured differentially across the inner electrodes. (2) The potential-sensing electrodes are connected to a pre-amplifier with very high input impedance, so that no significant loading of the measured response occurs. (3) The current electrodes are driven from a floating source to ensure that the current enters and leaves the tissue only through the current electrodes, i.e. with no significant leakage to earth. (4) The current source is a constant-current source, so that the excitation is automatically maintained at a given level during the measurement. (5) The density of the sinusoidal current to the stimulating electrodes is less than 10-13 A/sq./~m which is within the safe limits recommended to avoid cortical stimulation. (6) One can relate the measured impedance to absolute impedance without knowing the exact geometric factor for the electrode-probe used. This is achieved by adjusting the impedance-reading on the meter with the probe in a solution of known resistivity. This can be done both before and after measuring to ensure that the electrode-positions have not been changed during measurement. One may also compensate for any external phase shift in the experiment by using the phase-shifter adjustment on the front panel, although in the present study we found this never to be necessary. RESULTS Blood and CSF parameters The mean arterial blood pressure (MABP), arterial pO2, pCOg, pH and [HCO3-] and sagittal venous pCO2 values in the different groups are shown in Table I. There are two points worth noting. Firstly, sagittal venous pCO2 declined in value over time, with most of the change occurring prior to 5 min into the isoelectric period, whereas arterial pCO~ remained relatively constant throughout. This narrowing of the arterial-venous pCOz difference from 8.5 (control) to 1.8 mmHg (30 min isoelectric EEG):is undoubtedly a result of an increased cerebral blood flow TM. Secondly, the arterial [HCOz-] was increased over control at 30 rain of slow-wave EEG activity and u p t o 5 rain of isoelectricity (P -< 0.05). The arterial [HCOz-] then declined sharply from the 5 min value as the isoelectric period was prolonged (P < 0.01, when comparing the 15 and 30 min [HCO3-] to that of 5 min). In addition, the arterial [HCOz ] in the control

134 5- 5

167 5- 12 153 ~ 15 150 5- 8

Slow-wave(n = 5)

lsoelectric E E G 5 min 15 rain 30 min

101 ± 8 107 ± 8 105 5- 7

102 5- 4

103 ± 4

paO2 (mmHg)

* P < 0.05, when c o m p a r e d to control. ** P ~ 0.01, when compared to control. *** P _~ 0.001, when c o m p a r e d to control.

140 5- 3

Control (n = 6)

MABP (mmHg)

Blood

38.1 5- 0.9 36.9 5- 1.2 37.7 ± 1.7

38.9 5- 0.9

39.3 5- 0.9

paC02 (mmHg)

7.393 ± 0.020* 7.298 5- 0.01§§ 7.275 ± 0.032§

7.384 ± 0.009*

7.313 ± 0.020

pHa

24.8 5- 1.1" 18.9 5- 0.6§§ 18.2 ~ 1.4§§

24.5 ± 0.8*

21.1 ± 1.2

7.451 5- 0.014" 7.455 ± 0.020* 7.448 5- 0.028

7.440 5- 0.009*

7.403 ~ 0.010

/HC03 /~ pH (mM/kgplasma)

CSF

§ P ~ 0.05, when c o m p a r e d to 5 rain isoelectric EEG. §§ P < 0.01, when c o m p a r e d to 5 rain isoelectric E E G .

41.0 4- 1.3"* 40.0 ± 1.1"** 39.5 ~ 1.7"*

44.2 5- 1.1 *

47.8 5- 1.1

pvCO2 (mmHg)

All values are mean 5_ S.E.M. M A B P = mean arterial blood pressure; pvCO2 = sagittal venous pCO2.

Physiological parameters in blood and CSF

TABLE 1

27.0 ± 1.0 27.1 ~ 0.1 26.2 ± 1.3

27.6 ± 0.7

26.3 ± 0.2

/HC03 ] (mM/kg)

136 animals was consistently below that in the CSF. However, this is not an uncommon finding in anesthetized, paralyzed animals maintained for periods of hours (for example, see ref. 37). The CSF pH was significantly increased from control in all groups except the 30 min isoelectric EEG group. However, although the first 3 experimental groups showed some rise in CSF [HC03 -], only the IS-min isoelectric EEG group had a [HC03 -] significantly greater than control. Impedance changes Fig. 2 shows the impedence changes occurring in 3 of the 7 animals studied. These 3 animals were selected to represent 3 groups of rats exhibiting a particular pattern of impedance changes. There were 3 factors used in defining the groups, which were: (I) the occurrence of one or two periods of impedance increase, followed by full recovery, prior to the onset of an isoelectric EEG: (2) the occurrence of a final impedance increase at least 10 min before the onset of an isoelectric EEG (maximum, 27 min): and (3) the occurrence of a final impedance increase within 3 min of the onset of an isoelectric EEG. Fig. 2A represents rats exhibiting factor 1 and 2 above (n =_7 2). B represents rats exhibiting factors 1 and 3 (n ,- 3). Finally, C represents rats exhibiting only the third factor (nee 2). It should be noted that no changes in impedance could be detected in any animal during the slow-wave period. If a transient impedance rise and recovery occurred, it was only during the period of pronounced 1600 1400 1200 1000

A

rU

BOO

E0 ~ ~

II)

g IV

1 .s ~ -~ III

600 400 1600 1400 1200 1000 800 600 400

B

I

-[

1400 1200 1000 800 600 400

f-----

: Isoelectric EEG I onset I

I

I

80

.

.

100

120

. 140

.

160

. 180

i~

200 2

5 15 time following injection -~·-II-·- time following onset of isoelectric EEG (min)

EEG Pattern, fnormalj- slow wave

30

---I

(min)

-l--- convulsive -_.+1-'- - - - - - isoeIectric - - -_ _-1.1 -polyspike

Fig. 2. Changes in cortical specific impedance related to time and the EEG pattern in 3 individual animals. The duration of the EEG patterns (Le. the dominant pattern seen) prior to the onset of isoelectricity represent an average value for the 7 animals in the impedance series. These 3 rats were chosen to represent 3 general patterns of impedance changes found following the injection of insulin and induction of anesthesia (see text).

137 TABLE II Changes in cortical impedance and the extracellular fluid space (n

=

7)

All impedance values have been corrected upward by 11 %, based on the estimated amount of current shunted through cortical vessels (see ref. 41). All values are mean ± S.E.M. ECS = extracellular fluid space; for explanation of Rayleigh and Maxwell equations, see text. Impedance (D.·cm)

ECS estimate Rayleigh equation

Prior to isoelectric EEG (normal) 547

±

14

18.3

±

0.4

Isoelectric EEG 5 min 15 min 30 min

± 89*** ± 42*** ± 64***

10.7 8.9 8.2

± ± ±

0.8 0.3 0.4

*** P

1016 1189 1302

% normal

58 49 45

± ± ±

4 I I

Maxwell equation

14.4

±

0.3

8.2 6.8 6.3

± ± ±

0.6 0.2 0.3

% normal

57 47 43

±4 ± I ±1

:s 0.001 when compared to normal.

polyspike and/or convulsive EEG activity - with the earliest occurring 45 min prior to the onset of an isoelectric EEG. The changes in impedance occurring following the onset of isoelectric EEG are shown in Table II. From an average of 547 ± 14 n cm, the impedance increased to 1016 ± 89 n cm at 5 min ofisoelectric EEG and rose further to 1189 ± 42 n cm at 15 min, and to 1302 ± 64 n cm at 30 min (P < 0.001 in all cases). The impedance increases between 5 and 15 min and 15 and 30 min of isoelectric EEG were significant (P

<

0.05).

Changes in the extracellular fluid space (ECS) Two equations were used to calculate the ECS from impedance values. The equations, taken from Cole et al.l!, are based on the passage of a current through a uniform suspension of spheres (Maxwell equation): P 2

I-

rl/r2

2-

rl/r2

(5)

or through a uniform suspension of cylinders normal to the axis of the cylinders (Rayleigh equation): I-

rl/r2

p=--­

I

+ rl/r2

(6)

where p = volume of the cells in the suspension, rl = specific impedance of the fluid surrounding the cells (the extracellular fluid) and has a value of 55 n cm at 37 °C13, and rz = specific impedance of the tissue as a whole. The ECS is then calculated as I-p. With both equations, one calculates the same reduction in the ECS in hypoglycemia at 5, 15 and 30 min following the onset of an isoelectric EEG (Table 11).

42.4 + 1.0

40.3 :~ 1.0* 39.4 -! 1.1"* 39,2 ± 1.8"

Slow-wave E E G (n = 5)

Isoelectric E E G 5 rain (n = 6) 15 rain (n = 5) 30 m i n (n = 5)

10.73 iz 0.20 10.91 ± 0.27** 11.69 % 0,27

13.27 L 0,28*

12.40 =: 0.22

* P ~ 0.05, when c o m p a r e d to control. ** P _< 0.01, w h e n c o m p a r e d to control. *** P - 0.001, when c o m p a r e d to control.

44.2 4~ 0.9

Control (n = 6)

6.956 i 0,017 7.016 ± 0.016"* 7,083 ~ 0,023***

7.000 L 0.013"*

6.934 i 0.011

20 % E C S

pHi

pC'O2 (mmHg)

TOO2 {mM/kg )

lntracellular fluid

Cortical tissue

in the isoelectric groups, see text.

6.993 j= 0.015 7.045 5= 0.013" 7,101 i 0.022**

7.057 ~ 0,011"*

6,997 5 0.010

15 o; E C S

8.67 ± 0.20 9.75 i 0,41 11.29 i 0.47**

10.12 - 0,30*

9.14 :t 0.30

20 % E C S

9.45 :i 0.19" 10.42 ± 0.39 11.75 ± 0.43*

11.53 > 0.29*

10,51 ± 0.29

15 % E C S

H C O :~ ~ (mM/kg(~li.c. H20)

All values are m e a n £ S.E.M. Tissue pCO~ calculated by adding I m m H g to the arithmetic m e a n of the arterial a n d sagittal v e n o u s pCO2. S y m b o l s denote statistical significance as in Table 1. A s s u m e d ECS value for control (and slow-wave E E G ) animals. F o r ECS values used in calculating pH~ a n d [HCO3]~

A cid-base parameters in cortical tissue and cortical intracellular fluid

T A B L E III

139

Because of certain limitations and objections to the use of these equations in calculating absolute values for the ECS (see Discussion), we used the per cent change in the ECS, from an assumed normal value of either 20~;'; or 15 %, to estimate the respective ECS values in each group. For a control value of 20 %, the ECS in each respective group would then be 11.6 % at 5 min, 9.6 % at 15 min and 8.8 % at 30 min. For a control value of 15%, the respective ECS values are 8.7% (5 min), 7.2% (I5 min), and 6.6 %(30 min). Because no changes in specific impedance were observed during slow-wave EEG activity, this group was also assigned an ECS value of either 20% or 15%. Acid-base parameters in brain tissue and intracellular fluid These parameters are shown in Table Ill. Tissue pC02 levels were significantly reduced from control following the onset of an isoelectric EEG (up to 5 mmHg at 30 min isoelectric EEG). This drop in pC02 will directly affect the pH values calculated for both the CSF and intracellular fluid compartments whereas it should have a negligible effect on [HC03-] (see Discussion). Regardless of whether a 20% ECS or a 15 %ECS was used in calculation, pHi increased significantly from control in the slow­ wave EEG group (P < 0.00 1 for a 20~'-;; ECS and P < 0.0 I for a 15 % ECS). The pHi then declined to a value not significantly different from control (P > 0.1) at 5 min of isoelectric EEG. When comparing the 5 min isoelectric EEG group to the slow-wave group we found no significant difference in the pHi values if a 20 % control ECS (thus 11.6 % at 5 min isoelectric EEG) was used in the calculation. However, the 5 min isoelectric EEG value was found to be significantly lower (P < 0.05) if a 15~;'; control ECS (thus 8.7 % at 5 min isoelectric EEG) was used to calculate pHi. The pHi at 15 and 30 min of an isoelectric EEG was significantly increased over control for calculated pHi values based on both a 20 % control ECS (thus 9.6 % and 8.8 %, respectively, at 15 and 30 min) and a 15 % control ECS (thus 7.2 % and 6.6 %, respectively). However, the magnitude of the pHi change was somewhat diminished in both groups when using a 15 % control ECS as opposed to a 20 % control ECS in the calculations. Thus for a 20% ECS, the pHi change from control in the 15 and 30 min isoelectric EEG groups were + 0.082 (P < 0.01) and + 0.149 (P < 0.001), respectively. For a 15% ECS these changes reduced to + 0.048 (P < 0.05) and + 0.104 (P < 0.01), respectively. The pHi values in the 15 and 30 min groups were significantly greater than that of the 5 min group (P < 0.05 and P < 0.01), respectively, regardless of the control ECS). The pHi in the 15 min group was not significantly different than that in the slow­ wave group (P > 0.1, regardless of the ECS). Calculations based on a 20 % control ECS give a pHi value in the 30 min group significantly higher than that in the slow­ wave (P < 0.05) and the 15 min isoe1ectric EEG (P < 0.05) groups, whereas calculations based on a 15 % control ECS somewhat diminish these differences thus giving no statistically significant differences (P > 0.1). The pattern of change for [HCO-3]i was similar to that for pH;, but with some important differences. The [HCO-3]i in the slow-wave EEG group was also significant­ ly greater than control, regardless of the assumed control ECS (P < 0.05). However,

140 the [HCOz-]i at 5 rain of isoelectric EEG was significantly less than that of the slowwave group for calculations based both on a 20 °/o and a 15 j°o control ECS (P < 0.01 and P < 0.001, respectively) and significantly lower than control for calculations based on a 15 % control ECS only (P < 0.05). Regardless of on which ECS the calculations were based, the [HCO3-]i at 15 min of isoelectric EEG was significantly elevated over that of the 5 min group (P < 0.05) but not significantly different from the [HCOa-]i in either the control or the slow-wave group (P > 0. I). The [HCOak ]i at 30 min of isoelectric EEG showed a further increase over that in the 15 min and the 5 rain groups (P < 0.05 and P < 0.001, respectively, regardless of the assumed ECS). 1~ ~'/,>control ECS gave a However, although calculations based on both a 20 '•b{iand a .v [HCOa-]~ in the 30 rain group, which was significantly elevated over control (P < 0.01 and P < 0.05, respectively), the ~ [HCO3-]i for a 20 ~.{, control ECS was somewhat greater than i f a 15% ECS was used ( i- 2.14 vs -i 1.24, respectively). DISCUSSION The present results have shown that in animals maintained at constant arterial pCO2, severe hypoglycemia is accompanied by increases in CSF and intracellular fluid pH. In the CSF, pH increased by 0.04-0.05 units, the change being statistically significant in all groups but one (30 min isoelectric EEG). Although intracellular pH increased by 0.15 units in the 30 min isoelectric groups, and the change was significant already in the slow-wave-polyspike group, the results suggest that a shift towards an acid pH occurred early in 'isoelectricity' (5 min group). The results thus demonstrate that hypoglycemia 'coma', a condition which leads to neuronal cell damage 1°,1'~, is associated with a rise rather than a fall in intracellular pH. However, to some extent this rise is due to the fal! in tissue pCOe. Since we are also concerned with the influence of hypoglycemia on non-respiratory acid-base changes, some emphasis will be put on intracellular [HCOz-] in the present discussion. Before any detailed discussion of results can be attempted, it is important to establish the inherent limitations of the methods used in the present study to calculate [HCOz-]i. The interpretation of present results is dependent on the assumptions regarding both the extracellular fluid [HCO3-] and, as stated in the Introduction, the size of the extracellular fluid space.

Extracellular l HCOa-] concentrations During severe hypoglycemia, decreasing tissue levels of labile metabolic acids (see Introduction) may lead to, over the short-term, a [HCO3-] in the extracellutar fluid which is not uniformly distributed a n d i s higher than that in the bulk CSF. Yet one might expect the CSF to at least reflect the direction of the [HCO3-] change in the extracellular fluid, although with some time delay. However, there is no indication of any large or progressive increase in CSF [HCO3-] over time in the present study (see Results). Furthermore, recent reports have pointed to a greater role for blood in influencing ECF [HCO3-] ~,1~. In the present study, we find a blood [HCO3-] which

141 first increases and then declines substantially during the isoelectric period (see Table II; see also ref. 3). There are, therefore, no indications of any appreciable increases in ECF [HCO-3] in severe hypoglycemia. Tissue impedance and the extracellular fluid space Results from the present study clearly show that severe hypoglycemia is associated with a marked rise in the cortical impedance (greater than 2-fold), which continues to increase as one prolongs the period ofisoelectric EEG. With some limita­ tions (see below), this is equivalent to an approximate halving of the ECS. In 5 of the 7 animals studied, however, we observed one or two periods of transient impedance increase and recovery prior to the final rise in impedance which occurred (in 5 of the 7 animals) around the time of the onset of an isoelectric EEG. This final rise in impedance is undoubtedly a result of the energy failure 3 ,21,24 and reduction in Na+-K+ pump activity7 which occurs at this time (see Introduction). The transient impedance changes resemble those one observes during spreading depression 46 . These could be a result of a local K + release in specific loci within the cortical tissue, where perhaps the glucose concentration has fallen below the critical level needed to support energy production, thus the Na+-K + pump, but where there is still sufficient pump activity in the remainder of the cells to take up the K + which is released. Two of the animals showed a final impedance increase 13 and 27 min prior to the onset of an isoelectric EEG. This observation is not easily explained, but may be due to tissue trauma as a result of operative procedures, including the craniotomy, or the presence of the electrode itself. Thus, tissue damage (e.g. impaired circulation) could render the tissue more 'sensitive' to reductions in blood glucose. The impedance values at 15 and 30 min of isoelectric EEG in these two animals were within the range of values found for the other 5. However, the 5-min isoelectric EEG impedance value in one of the two animals was somewhat higher (not shown). This may imply that we have made a slight overestimation of the change in the size of the ECS in the 5-min group. However, there are other factors present at this time which could cancel out this effect (see below). There are a number of factors which diminish one's ability to estimate an absolute value for the cortical ECS from the measurement of the specific tissue impedance. Firstly, the equations given by Cole et alP assume a uniform suspension of cells of a given shape. This of course, is not the case in cortical tissue, which is composed of a mixture of cell bodies and cellular processes. Secondly, one must assume that all the current passes through the ECS and not the cells or their processes. However, this may be dependent on the orientation of fibres within the current field, with a much greater amount of current entering fibres oriented in the same direction as the current flow 23 ,30,32. Finally, there may be some areas within the ECS which may not participate in the passage of current (cf. refs. 45 and 49). It is, therefore, more reasonable to use impedance measurements to estimate changes in the size of the ECS. By doing so one can essentially eliminate cell shape and fibre orientation as a source of error because: (I) both the Rayleigh and Maxwell equation gave the same change in ECS in the present study; and (2) both cell shape and fiber orientation should remain constant. In the present study we have a reasonable range for the cortical ECS under

142 normal conditions. This is a reasonable range of ECS values from the literature derived through conventional tracer methods or electron microscopy from freezesubstituted tissue6.8,13,a6,19,2~,5o,although the values reported in these studies suggest an ECS closer to 20'}~,. There are, however, some limitations in using impedance measurements to estimate changes in the size of the ECS. In severe hypoglycemia there are two factors which may cause us to underestimate the actual change in the ECS size. l=irstly, the increased cortical blood flow of 2-3-fold could increase the amount of current passing through cortical vessels (cf. refs. 41 and 45). Secondly, it is quite likely that cell membrane resistance is decreased as a consequence of the depleted energy reserves and reduced in pump activityZ,7,2°, 24, thereby allowing more current to pass through the cells. Both of the above possibilities would mask some of the decrease in the ECS size. One factor which could cause us to overestimate the change in the ECS is an increased tortuosity in the extracellular fluid space. Increased tortuosity is a possibility in conditions, like the present, where the ECS is shrinking (cf. ref. 31 ). This could lead to areas of the ECS being 'hidden' from the current field. Finally, it has been suggested that an intracellular polyanionic mucoprotein/mucopolysaccharide matrix could bind extracellular cations reducing their mobi|itiesz4 and that alterations within the matrix could lead to changes in impedence independent from changes in the ECS 2,49. On the other hand, there is some indirect evidence that changes in the matrix are directly associated with a change in the size of the ECS 47. However, there are no data regarding how severe hypoglycemia would affect the matrix. Therefore we are not able to predict whether changes in the matrix would cause us to over- or underestimate the change in the ECS. There is information in the literature which allows some direct comparisons between impedance techniques and tracer or EM (freeze-substitution) techniques in measuring the magnitude of the reduction in the ECS following the onset of asphyxia and ischemia, which, like hypoglycemia, is associated with a disruption of membrane ion pumps. Using tracer or EM methods, the ECS was found to decrease to about I/4 of normal in a variety of brain regions, including the cortex (8 min to several hours following onset of asphyxia) a,zg,44. Impedance changes would give an ECS reduction to about 1/3 of normal in the cerebellum or cortex at about 10 min following the onset of asphyxia41,4a or 30 min of complete ischemia39. Although these comparisons are not precise, it appears that impedance methods may underestimate the ECS change following asphyxia or ischemia. The error is made even larger if one subtracts the effect resulting from collapse of intracerebral vessels. This same blood effect could not be present in severe hypoglycemia, however (see above). This leaves one with the possibility that a decreased cell membrane resistance during asphyxia (and severe hypoglycemia) may have some effect on the impedance measurement. If one were to take this possibility (together with the increased CBF) into account in the present study, it would only make the [HCO3-]i and pHi values higher in the isoelectric groups and increase the difference between them. We would therefore conclude that most, if not all, of the impedance changes reported in this study, are direct result of a decrease in the size of the ECS and, if anything, underestimate the true ECS change.

143

Changes in /HC03-]i during hypoglycemia As mentioned earlier, oxidation of anions of metabolic acids during hypoglycemia would be expected to remove H + ions from the intracellular fluid, thereby increasing [HCO3-] and buffer base (BB) concentration (see ref. 38). The approximate changes in [BB] and [HCO3-] can be estimated by an equation which simulates the buffer capacity of cerebral intracellular fluidsa4, 3s. Thus, if values for pCO2 and pH~ in the control situation are inserted, the simulated system will give a (fictituous) BB concentration. For the hypoglycemic groups, one can then derive the expected [BB] and [HCO3-] from Pco2 and A [BB], the latter being calculated from the concentration changes of major metabolic acids3, 20,24. So calculated,/k [BB] values for the slow wave and the 5, 15 and 30 min isoelectric groups were 2.5, 3.8, 3.8 and 7.8 mM/l, respectively, giving expected increases of [HCO3-]i of 1.1, 1.8, 1.7 and 4.1 retool/l, respectively. As Fig. 3 shows, with the exception of the slow-wave group, there were appreciable differences in expected and measured [HCO3-]i, especially in the 5-rain group, suggesting that, at that time, other factors must have caused acidification of the 14.0x Control o Predicted from ~ BB • Measured value

13.0-

/

/

,2

// /

/

/

//

0

12.0-

/l

/

/

/

E

b"

jo ........../ / / / t

11.0-

/T\

10.0

9.0' Control

/I

30 rain. I-

Nl~-w~ EEG

~

15

I ~ e l ~ t r i c EEG (mln.)

~)

Fig. 3. Evaluation of intracellular [HCO,~ ] changes during hypoglycemia. The measured changes in intracellular [HCO3 ] from control (x) at 30 min following the onset of a slow-wave LEG pattern and at 5, 15 and 30 min following the onset of an isoelectric LEG are represented by the closed circles. The open circles represent the predicted [HCO3 ]~ for each group using a computer model which takes into account physicochemical buffering and change in the concentration of metabolic acids (A B B)38. The ± BB values used in the calculation were taken from Lewis et al. 20, Norberg and Siesj624, and Agardh et al. 3. The calculations are based on a control ECS of 2 0 ~ (where applicable).

144 intracellular fluids. At present, noting is known about the mechanism(s) responsible but two alternatives can be envisaged. Before these alternatives are discussed we wish to emphasize that it is very unlikely that an error in the estimate of the ECF [HCO3- ] could explain the differences illustrated in Fig. 3. Thus, it would require a substantial decline in ECF [HCOs-] between the slow-wave and the early isoelectric period to account for the decrease in [HCO3-]i. As discussed above, there is no reason to suspect that [HCO3-] decreases during this period. It should also be mentioned that an overestimation of the ECS by 3-4/°,,'i (actual ECS) in the 5-rain group could account for half of the [HCO3-]i difference between that group and the slow-wave one. However, it would require a decrease in the ECS to an impossibly low value ( ~ 2 '}~o)to produce a [HCO3-]i in the 5-min group which would correspond to the change in metabolic acid levels. It is possible that the explanation for the decline in [HCO3--]i between the slowwave and early isoelectric EEG periods lies in transient factors surrounding depletion of energy stores and impairment of membrane ion pumps. Under normal conditions, the [HCO3-]i one measures represents a balance between H + generation (from cellular metabolism) and H + buffering and transport out of the cell (or HCO:~ entry) 3(~,:~8. Conventional H ~ generation is certainly an unlikely possibility, considering the declining metabolic acid levels. However, since we cannot eliminate the possible occurrence of other H ~-generating sources within the cell, the first possible alternative is that metabolic events other than those considered here are associated with release of H +. Although such mechanisms must remain undefined it is of interest that during the first 5 rain of hypoglycemic coma about 5 #mol/g of phospholipids are broken down ~. It will remain for future research to find out whether these or associated events affect pHi. A second alternative for the ~acidification' of intracellular fluids is an interference with the transmembrane 'transport' of H~/HCO3 - when energy stores are depleted and Na~-K + pump activity is reduced 9,4°. Furthermore, movement of C1 into the cell, when pump activity is suddenly diminished 18,19,42,~8, could force some HCO3- ions out of the cell 5. Finally, the loss of K + from the cell could provide a gradient for H movement in. At present, we cannot be certain as to the precise mechanisms responsible for the apparent tack of stoichiometry between the reduced metabolic acid levels and the [HCO~-]~ at 5 min of isoelectric EEG. Between 5 and 15 min of isoelectric EEG activity we observed a significant rise in [HCO3-]i. This particular time period is not, however, associated with any significant further reductions in metabolic acid levels :3.It is very unlikely that an error in the ECS estimate could account for this (discussed earlier). An error due to misestimation of the ECF [HCO3-] is also unlikely in that the ECF [HCO3-] would have to be increased by more than 3 mM/l in order to account for just one-half of the [HCO3 ]~ difference between the 5- and 15-min groups. One explanation could be an increase in the electrochemical potential for HCO~- movement into the cell. We have recently found in hypoglyeemic rats (unpublished observations) that ECF [K +] progressively increases over 30 min from the initial increase following the onset of an isoelectric EEG, up to 20 mM/1 greater than that found by Astrup and Norberg v. In support of this, the large

145 decrease in arterial bicarbonate occurring after 5 min of isoelectric E E G (see also ref. 3) at a time when arterial lactate and pyruvate levels are falling 21 suggests that HCO3is moving into the ceils in other tissues. However, any implication of a HCO3- shift from the cortical extra- to intracellular fluid awaits a direct microelectrode measurement of [H ] changes in the ECF. The intracellular [HCO3-] at 30 rain of isoelectric EEG was higher than in any other group. The further rise in [HCO3-]~ when going from 15 to 30 rain of isoelectric EEG is not explicable on the basis of errors in the estimate of the ECS or ECF [HCO3-]. It is possible that those factors discussed above could have had some influence in producing an additional increase in [HCO3-]i. However, calculations from the data of Agardh et al. 3 show an increase of 4.0 mM/l of the buffer base concentration between 15 and 30 rain of isoelectric EEG, due primarily to a further reduction of glutamate. This is certainly more than sufficient to account for the 1.3-1.5 mM/I difference in [HCO3-]i found between the two groups. A bit of caution must be exercised, however, when discussing stoichiometric relationships in the present study especially when one takes deviations from precise stoichiometry as suggesting that factors other than changes in metabolic acid levels must be used to explain significant but rather small (1-2 raM/l)[HCO3-]t differences among groups. We also cannot discount the influence of changes in labile acids not measured in the studies of Lewis et al. 2°, Norberg and Siesj62~ and Agardh et al. ~ as it affects the intracellular buffer base concentration and the intracellular bicarbonate concentrations of the present study. in summary, the present experiments demonstrate that insulin-induced hypoglycemia is accompanied by an increase in cerebrospinal fluid pH. Furthermore, although the quantitative values for intracellular pH are based on a series of assumptions and although the cause of the changes observed cannot be accurately defined, the data obtained show that also the intracellular pH increases above control. ACKNOWLEDGEMENTS This study was supported by grants from the Swedish Medical Research Council (Project No. 14X-263), and from USPHS (No. 5 R01 NS07838). The authors wish to thank Barbro Asplund and Kerstin Beirup for skilful technical assistance.

REFERENCES Abdul-Rahman, A., Agardh, C.-D. and Siesj6, B. K., Local cerebral blood flow in the rat during severe hypoglycemia, and in the recovery period following glucose injection, Acta physiol, seand., 109 (1980) 307-314. 2 Adey, W. R., Bystrom, B. G., Costin, A., Kado, R. T. and Tarby, T. J., Divalent cations in cerebral impedance and cell membrane morphology, Exp. Neurol., 23 (1969) 29-50. 3 Agardh, C.-D., Folbergrovfi, J. and Siesj6, B. K., Cerebral metabolic changes in profound, insulininduced hypoglycemia, and in the recovery period following glucose administration, J. Neurochem., 31 (1978) 1135 1142. 1

146 4 Agardh, C.-D., Chapman, A. G., Nilsson, B. and Siesj6, B. K., Endogenous substrates utilized by rat brain in severe insulin-induced hypoglycemia, J. Neurochem., 36 (1981) 490-500. 5 Ahmad, H. R., Loeschcke, H. H. and Woidtke, H. H., Three compartment model for the bicarbonate exchange of the brain extracellular fluid with blood and cells, Advanc. exp. Med. Biol., 99 (1977) 195-209. 6 Arieff, A. I., Kerian, A., Massry, S. G. and Delima, J., lntracellular pH of brain: alterations in acute respiratory acidosis and alkalosis, Amer. J. PhysioL, 239 (1976) 804-812. 7 Astrup, J. and Norberg, K., Potassium activity in cerebral cortex in rats during progressive severe hypoglycemia, Brain Research, 103 (1976) 418-423. 8 Bondareff, W. and Pysh, J. J., Distribution of extracellular space during postnatal maturation of rat cerebral cortex, Anat. Rec., 160 (1968) 773-790. 9 Boron, W. F., Active control of intracellular pH, Re~pirat. Physiol., 33 (1978) 59-62. 10 Brierley, J. B., Brown, A. W. and Meldrum, B. S., The nature and time course of the neuronal alterations resulting from oligaemia and hypoglycemia in the brain of Macaca mulatta, Brain Research, 25 (1971)483-499. 11 Cole, K. S., Li, C.-L. and Bak, A. F., Electrical analogues for tissues, Exp. neurol., 24 (1969) 459-473. 12 Cragg, P., Patterson, L. and Purves, M. J., The pH of brain extracellular fluid in the cat, J. Physiol. (Lond.), 272 (1977) 137-166. 13 Fenstermacher, J. D., Li, C.-L. and Levin, V. A., Extracellular space of the cerebral cortex of normothermic and hypothermic cats, Exp. Neurol., 27 (1970) 101-114. 14 Folbergrovfi, J., MacMillan, V. and Siesj6, B. K., The effect of hypercapnic acidosis upon some glycolytic and Krebs cycle-associated intermediates in the rat brain, J. Neurochem.. 19 (1972) 2507-2517. 15 Kalimo, H., Agardh, C.-D., Olsson, Y. and Siesj6, B. K., Hypoglycemic brain injury : II. Electronmicroscopic findings in rat cerebral cortical neurons during profound insulin-induced hypoglycemia and in the recovery period following glucose administration, Acta neuropath. (Bed.), 50 (1980) 43-52. 16 Korobkin, R. K., Lorenzo, A. V. and Cutler, R. W. P., Distribution of C~4-sucrose and llZ~-iodide in the central nervous system of the cat after various routes of injection into the cerebrospinal fluid, J. Pharmacol. exp. Ther., 164 (1968) 412-420. 17 Leaf, A., On the mechanism of fluid exhange of tissues in vitro, Biochem. J., 62 (t956) 241-248. 18 Leaf, A., Regulation of intracellular fluid volume and disease, Amer. J. Med., 49 (1970) 291-295. 19 Levin, V. A., Fenstermacher, J. D. and Patlak, C. S., Sucrose and inulin space measurements of cerebral cortex in four mammalina species, Amer. J. Physiol., 219 (1970) 1528-1533. 20 Lewis, L. D., Ljunggren, B., Norberg, K. and Siesj6, B. K., Changes in carbohydrate substrates, amino acids and ammonia in the brain during insulin-induced hypoglycemia, J. Neurochem., 23 (1974) 659-671. 21 Lewis, L. D., Ljunggren, B., Ratcheson, R. A. and Siesj6, B. K., Cerebral energy state in insulininduced hypoglycemia, related to blood glucose and to EEG, J. Neurochem., 23 (1974) 673-679. 22 Mitchell, R. A., Herbert, D. A. and Carman, C. T., Acid-base constants and temperature coefficients for cerebrospinal fluid, J. appl. Physiol., 20 (1965) 27-30. 23 Nicholson, P. W., Specific impedance of cerebral white matter, Exp. Neurol., 13 (1965) 386--401. 24 Norberg, K. and Siesj6, B. K., Oxidative metabolism of the cerebral cortex of the rat in severe insulin-induced hypoglycemia, J. Neurochem., 26 (1976) 345-352. 25 Pelligrino, D. A. and Dempsey, J. A., Effects of hypocapnia on Pcoe gradients between CSF and cerebral capillary blood, J. appl. Physiol., 43 (1977) 480-486. 26 Pelligrino, D. A., Musch, T. I., Dempsey, .1. A. and Nagle, F. J., Brain pHi: interregional differences, comparison of methods, and effects of hypoxic hypocapnia, Fed. Proc., 37 (1978) 532 (Abstr.). 27 Pont6n, U. and Siesj6, B. K., A method for the determination of the total carbon dioxide content of frozen tissues, Acta physiol, scand., 60 (1964) 297-308. 28 Pont6n, U. and Siesj6, B. K., Gradients of CO2 tension in the brain, Actaphysiol. stand., 67 (1966) 129-140. 29 Rall, D. P., Oppelt, W. W. and Patlak, C. S., Extracellular space of brain as determined by diffusion of inulin from the ventricular system, Life Sci., 2 (1962) 43-48. 30 Ranck, J. B., Jr. and BeMent, S. L., The specific impedance of the dorsal columns of cat: an anisotropic medium, Exp. Neurol., 11 (1965)451-463. 31 Ranck, J. B., Jr., Electrical impedance of the subicular area of rats during paradoxical sleep, Exp. Neurol., 16 (1966) 416-437.

147 32 Ranck, J. B., Jr., Electical impedance changes in many sites of brain in paradoxical sleep, anesthesia, and activity, Exp. Neurol., 27 (1970) 454-475. 33 Rose, I. A., The state of magnesium in cells as estimated from the adenylate kinase equilibrium, Proc. nat. Acad. Sci. (Wash.), 61 (1968) 1079 1086. 34 Schettini, A. and Moreshead, G., Effects of balothane and sodium thiopentone on surface brain pressure and brain electrical impedance in dogs with normal intercranial tension, Brit. J. Anaesth., 50 (1978) 1003-1011. 35 Severinghaus, J. W., Blood gas concentrations. In W. O. Fenn and H. Rabn (Eds.), Handbook ~ f Physiology, Vol. 2, Waverly Press, Baltimore, 1965, Ch. 61, pp. 1475-1487. 36 Siesj6, B. K. and Messeter, K., Factors determining intracellular pH. In B. K. Siesj6 and S. C. Sorensen (Eds.), hm Homeostasis o f the Brain, Munksgaard, Copenhagen, 1971, pp. 244-262. 37 Siesj6, B. K., FolbergrovS., J. and MacMillan, V., The effect of hypercapnia upon intracellular pH in the brain, evaluated by the bicarbonate carbonic acid method and from the creatine phosphokinase equilibrium, J. Neurochem., 19 (1972) 2483 2495. 38 Siesj6, B. K., Metabolic control of intracellular pH, Seand. J. Clht. Lab. Invest., 32 (1973) 97-104. 39 Sobotka, P., Impedance changes of the brain cortex during and after complete ischaemia, Activ. Nerv., Suppl., 13 (1971) 250-252. 40 Thomas, R. C., Comparison of the mechanisms controlling intracellu[ar pH and sodium in snail neurons, Respirat. Physiol., 33 (1978) 63-73. 41 Van Harreveld, A. and Ochs, S., Cerebral impedance changes after circulatory arrest, Amer. J. Physiol., 187 (1956) 180-192. 42 Van Harreveld, A. and Schad6, J. P., Chloride movements in cerebral cortex after circulatory arrest and during spreading depression, J. cell. comp. Physiol., 54 (1959) 65-84. 43 Van Harreveld, A., Asphyxial changes in the cerebellar cortex, J. eell. comp. Physiol., 57 (196l) 101 110. 44 Van Harreveld, A., Crowell, J. and Malhotra, S. K., A study of extracellular space in central nervous tissue by freeze-substitution, J. Cell Biol., 25 (1965) 117-137. 45 Van Harreveld, A., Brain Tissue Electrolytes, Butterworths, London, 1966, pp. 50-94. 46 Van Harreveld, A. and Khattab, F. I., Changes in cortical extracellular space during spreading depression investigated with the electron microscope, J. Neurophysiol., 30 (1967) 911-929. 47 Van Harreveld, A., Dafny, N. and Khattab, F. I., Effects of calcium on the electrical resistance and the extracellular space of cerebral cortex, Exp. Neurol., 31 (1971) 358 367. 48 Van Harreveld, A., The extracellular space in the vertebrate central nervous system. In G. A. Bourne (Ed.), The Structure and Function o f Nervous Tissue, Vol. IV, Academic Press, New York, 1972, pp. 499-51 I. 49 Wang, H. H. and Adey, W. R., Effects of carious and hyaluronidase on cerebral electrical impedance, Exp. neurol., 25 (1969) 70-84. 50 Woodward, D. L., Reed, D. J. and Woodbury, D. M., Extracellular space of rat cerebral cortex, Amer. J. Physiol.. 212 (1967) 367-370.