Journal of Cerebral Blood Flow and Metabolism
7:379-386 © 1987 Raven Press, New York
Hydrogen Ions Kill Brain at Concentrations Reached
in Ischemia
Richard P. Kraig, *Carol K. Petito, Fred Plum, and William A. Pulsinelli
Departments of Neurology and *Pathology (Neuropathology), Cornell University Medical College, New York,
New York, U.S.A.
Summary: Elevation of brain glucose before the onset of
nearly complete ischemia leads to increased lactic acid
within brain. When excessive, such acidosis may be a
necessary factor for converting selective neuronal loss to
brain infarction from nearly complete ischemia. To ex­
amine the potential neurotoxicity of excessive lactic acid
concentrations, we microinjected (0.5 ILl/min) 150 mM
sodium lactate solutions (adjusted to 6.50-4.00 pH) for
20 min into parietal cortex of anesthetized rats. Intersti­
tial pH (pHo) was monitored with hydrogen ion-selec­
tive microelectrodes. Animals were allowed to recover
for 24 h before injection zones were examined with the
light microscope. Injectants produced brain necrosis in a
histological pattern resembling ischemic infarction only
when pHo was � 5.30. Nonlethal injections showed only
needle tract injuries. Abrupt deterioration of brain acid­
base homeostatic mechanisms correlated with necrosis
since pHo returned to baseline more slowly after lethal
tissue injections than after nonlethal ones. The slowed
return of pHo to baseline after the severely acidic injec­
tions may reflect altered function of plasma membrane
antiport systems for pH regulation and loss of brain hy­
drogen ion buffers. Key Words: Glial cells-Hydrogen
ions - Infarction - Ion-selective electrodes -Ischemia
-Lactic acid.
Infarction occurs when brain glucose is elevated
above normal before the onset of nearly complete
ischemia (Myers, 1979; Kalimo et aI. , 1981; Pulsi­
nelli et aI., 1982), circumstances known to raise
neocortical lactate values during ischemia to 19 �
mmollkg (Pulsinelii et aI., 1982). Conversely, in
neocortex only small- to medium-sized neurons are
destroyed after nearly complete ischemia during
normoglycemia (Pulsinelli and Brierley, 1979),
which, in rat, is correlated with a neocortical lac­
tate content of up to 13 mmollkg (Pulsinelli and
Duffy, 1983). These considerations have led to the
postulate that infarction of brain from nearly com­
plete ischemia is thought to result from excessive
tissue lactacidosis (Plum, 1983).
Direct measurements of brain hydrogen ion con­
centration ([H+ ]) during ischemia have extended
our understanding of the mechanisms by which aci-
dosis can lead to infarction. Thorn and Heitmann
(1954), using surface H+ -selective electrodes, first
showed that brain becomes acidotic during isch­
emia. Twenty-seven years later Siemkowicz and
Hansen (1981) documented with H+ -selective mi­
croelectrodes that acidic shifts occurred in intersti­
tial pH (pHo) during nearly complete ischemia. Ne­
moto and Frinak (1981) also showed that a sec­
ondary acidosis occurs in interstitial space shortly
after reperfusion from nearly complete ischemia.
Kraig and colleagues (1985b) reported that this sec­
ondary acidosis can reach a pH as low as 5. 4 under
severely hyperglycemic conditions. More recently,
we used pHo and neocortical lactate content mea­
surements to provide the first evidence that some
cells become preferentially acidic during hypergly­
cemic and complete ischemia (Kraig et aI., 1985a).
Based on other data, obtained from simultaneous
and direct measurements of pHo and tissue carbon
dioxide tension, we suggested that the cells that
preferentially become more acidic than the re­
mainder of ischemic brain are glia (Kraig et aI,
1986a). Recently, direct measurements of intracel­
lular pH (Kraig and Chesler, 1987; Kraig and Ni-
Received February 16, 1987; accepted March 19, 1987.
Address correspondence and reprint requests to Dr. R. P.
Kraig at Department of Neurology, Cornell University Medical
College , 1300 York Avenue , New York , NY 10021, U.S. A.
Abbreviation used: FAM , 1: 1:8 (vol/vol/vol) solution of 37%
formaldehyde , acetic acid , and methanol , respectively.
379
380
R.
P. KRAIG ET AL.
.
cholson, 1987) have shown that glia become one to
two orders of magnitude more acidic than the inter­
stitial space during hyperglycemic and complete
ischemia.
The present study reports that injection of so­
dium lactate solutions directly into mammalian
brain produces coagulation necrosis when pHo
reaches or falls below 5.30 pH for 20 min. Prelimi­
nary reports of this work have appeared (Kraig et
aI., 1986b,c).
METHODS
Animal preparation and recording for acid injections
Male Wistar rats (250-400 g) were anesthetized with
halothane and ventilated spontaneously with an oxygen­
nitrogen mixture via an inhalation mask. Small cranio­
tomies were made over parietal cortex 2 mm lateral to the
sagittal suture and 2 mm posterior to the bregma. A tail
artery was cannulated. Animals were then mounted in a
stereotaxic head holder that was fitted with a water jacket
to keep body temperature at 37°C. Warmed (37°C) normal
Ringer solution (Kraig et aI., 1983) flowed through a su­
perfusion cup that surrounded the craniotomies. H+-se­
lective microelectrodes and a micro- ( 170 f.Lm) injection
needle (2- 136 1 fused silica needle from Supelco, Inc.)
were mounted on a Narishige MT-5 micromanipulator so
that the H+-selective microelectrode was at a point im­
mediately adjacent to the needle wall but 300 f.Lm above
the needle's tip.
The assembled needle-H+-selective microelectrode
array was advanced into the superfusion cup. First,
normal Ringer solution equilibrated with 95% oxygen5% carbon dioxide (7.35 pH at 25°C) flowed over the
brain. Second, normal Ringer solution equilibrated with
100% carbon dioxide (6. 1 1 pH at 25°C) replaced the first
superfusate to establish a two-point in situ pH calibra­
tion. Finally, 7.35 pH Ringer solution again filled the su­
perfusion cup and the array was advanced into the brain
so that the needle tip reached 800 f.Lm below the parietal
pial surface. Arterial pH, arterial oxygen tension (Pa02)'
and arterial carbon dioxide tension (PaC02) were stabi­
lized and monitored with a Corning 158 blood gas ana­
lyzer. Blood glucose was measured with a Glucometer
(Miles Laboratories, Inc.).
A constant infusion pump was used to inject sodium
lactate (Fluka Chemical Co.) solutions. Sodium lactate
solutions at 150 mM were adjusted to 4.00, 4.50, 5.00,
6.00, and 6.50 pH with concentrated hydrochloric acid
and injected at a rate of 0.5 f.Ll/min for 20 min. Recordings
of pHo were continued for 20-30 min after injections
were completed. The needle-H+-selective microelec­
trode was then withdrawn and the head incision sutured
closed. Animals were allowed to recover with food and
water provided ad lib. After 24 h animals were anesthe­
tized, killed, and their brains processed for histological
examination.
Histological techniques
Necrotic lesion diameters (Figs. 3 and 5) were
determined from frozen sections 02 animals with
an injection into each parietal cortex; n
24). For
frozen sectioning, animals were anesthetized with
halothane, decapitated, and their brains removed
and frozen in Freon over dry ice. Brains were
stored at - 80°C until processing, at which time
20-f.Lm-thick histological sections were cut from
frozen brains. Sections were stained with hematox­
ylin and eosin (Bancroft and Stevens, 1982). Ne­
crotic lesion diameters were measured with a com­
pound microscope that was calibrated with a stage
micrometer. Areas of tissue necrosis stained less
intensely with eosin and appeared pale with sharp
borders demarcating the area of necrosis from
normal brain.
A second group of seven animals was processed
by perfusion-fixation 24 h after acid injections to
provide more detailed light microscopic analysis of
the lesions. Animals were anesthetized with an in­
traperitoneal injection of urethane 0.6 g/kg; Sigma
Chemical Co.) and then killed by intracardiac per­
fusion-fixation with a I: 1:8 (vol/vol/vol) solution of
37% formaldehyde, acetic acid, and methanol, re­
spectively (FAM) (Pulsinelli and Brierley, 1979).
Animals were decapitated after 10-15 min perfu­
sion and heads stored at 4°C in FAM for 24 h.
Brains were then removed, serially dehydrated,
and embedded in paraffin. Seven-micron serial sec­
tions were stained with hematoxylin and eosin or
luxol fast blue and cresyl violet. Histological char­
acterization of lesions was done by one of the au­
thors (C.P.) who was blinded to the pH of the injec­
tants.
H+ -selective microelectrode construction
and calibration
H+-selective microelectrodes based on tridodecyla­
mine (Ammann et aI., 1981) were constructed, calibrated,
and used as previously described (Kraig et aI., 1983,
1985a, 1986a).
RESULTS
Preinjection physiologic variables
Blood physiologic variables were consistent with
those of other halothane-anesthetized and sponta­
neously ventilating animals (Kraig et aI., 1985a,b,
1986a; Kraig and Cooper, 1987) (Table O. Animals
were slightly hypercapnic probably because of a
TABLE 1. Physiologic variables
Brain pHo
Arterial pH
p.co2 (mm Hg)
p.o2 (mm Hg)
Arterial glucose (mM)
Hematocrit (%)
Systolic blood pressure (mm Hg)
Rectal temperature eC)
7.42
7.30
70
119
7.7
44
111
37.2
=
J Cereb Blood Flow Metab, Vol. 7, No.4, 1987
Values are means
±
SEM, n
=
38.
±
0.12
±
O.oI
±
1
3
0.2
0.0
1
0.1
±
±
±
±
±
HYDROGEN IONS KILL BRAIN
halothane-induced reduction in brainstem ventila­
tory drive stimuli (Smith, 1971). In addition, haio­
thane was the most likely cause for the mild reduc­
tion seen in arterial blood pressure (Deutsch, 1971).
Anesthetic levels of halothane (0. 75-1. 5%) were
associated with an absence of withdrawal to hind­
limb pin prick and an average Paco2 of 70 mm Hg
and an average systolic arterial blood pressure of
111 mm Hg. Otherwise, animals were normoxic
and normothermic and had normal hematocrits.
Animals were mildly hyperglycemic (glucose 7. 7 ±
0. 2 mM) presumably because they were not fasted.
Brain pHo was more alkaline (Fig. 1; Table 1)
than has previously been described. Brain pHo is
normally 0. 05-0. 10 pH more acid than arterial pH
in halothane-anesthetized and spontaneously venti­
lating rats (Kraig et aI. , 1985a, 1986a; Kraig and
Cooper, 1987). However, in experiments reported
here, pHo was 0. 12 pH on average more alkaline
(7.42 ± 0. 12 pH) than arterial blood (7. 30 ± 0. 01
---- -F
8.0
FIG. 1. pHo changes associated with insertion of injection
needle and H+-selective microelectrode. Penetration of the
microinjection cannula-H+-selective microelectrode array
into brain [seen as equivalent and parallel vertical deflec­
tions in direct current (DC) (bottom) and pHo (top) records]
often induced a spreading depression. DC and pHo changes
immediately after array penetration into brain were typical of
spreading depression (Kraig et aI., 1983): (a) slow negative
DC signal of �15 mV, which, after almost 5 min, was asso­
ciated with the return of spontaneous electrical activity (seen
as increased thickness of DC line) to involved parietal
cortex; (b) alkaline- then acid-going transient in pHo that
reached an acidic peak of �7.00 pH. After spreading depres­
sion pHo returned to a baseline of 7.42 ± 0. 12 pH (n
37).
Under normal circumstances brain is slightly (0.05-0. 10 pH)
more acid than blood (7.30 ± 0.01 pH). Why baseline pHo
was often more alkaline in these experiments is unknown
but probably reflects an effect associated with the blunt
trauma from penetration of the large (170 fLm) injection
needle into brain. For example, if bicarbonate remained the
sole H+ buffer of the interstitial space, then a reduction of
�20 mm Hg in tissue carbon dioxide tension at the H+-se­
lective microelectrode recording site would account for the
rise in pHo from 7.25-7.35 usually seen to 7.42 reported
here. Such a reduction in tissue carbon dioxide tension
might result from the increased diffusion distance from
normal brain to the H+-selective microelectrode recording
area created by the injection cannula.
=
381
pH). The cause of this alkalosis is unknown. If bi­
carbonate (HC03-) remains the sole H+ buffer of
the interstitial space, then the interstitial alkalosis
of 7. 42 could result from an increase in interstitial
HC03 - concentration or a fall in tissue carbon
dioxide tension. Changes in interstitial HC03concentration are unlikely to account for the ob­
served alkalosis since membrane transport pro­
cesses generally require minutes of activity to influ­
ence concentration changes on either side of a
membrane (Roos and Boron, 1981) and the alka­
losis was present as soon as the injection needle­
H+ -selective microelectrode array entered the
brain interstitial space. Instead, a reduction of
brain tissue carbon dioxide tension of 20 mm Hg is
a more likely cause for the rise in pHo from
7. 25-7. 35 that is usually seen (Kraig et aI. , 1983,
1985a, 1986a; Kraig and Cooper, 1987) to 7.42 that
is reported here (Table 1). Tissue carbon dioxide
tension might have been lowered near the H+ -se­
lective microelectrode recording tip by an in­
creased diffusion distance from the microelectrode
tip to brain that was created by the relatively large
injection cannula.
Penetration of the injection needle-H+ -selective
microelectrode array into parietal cortex often re­
sulted in a single wave of spreading depression
(Fig. 1), a gross, transient, perturbation in ion and
metabolite homeostasis of involved nervous tissue
(Bures et aI. , 1974). A single spreading depression
or even up to 40 repetitive spreading depressions in
neocortex does not result in detectable injury to
brain when involved areas are examined by light
microscopic techniques 24 h later (R. P. Kraig, un­
published observations). Thus, it is unlikely that
spreading depression associated with sodium lac­
tate injections caused any irreversible injury to
brain.
pRo changes associated with sodium lactate injections
Microinjection of sodium lactate solutions
quickly resulted in a steady-state pHo (Table 2; Fig.
2) that reflected the varying influence of brain
physicochemical H+ buffers. pHo plateaued 0. 240. 25 pH higher than injectants during introduc­
tion of the less acidic (6. 50 and 6. 00 pH) soluTABLE 2. Maximum pHofrom acid injections
Injectant pH
4.00
4.50
5.00
6.00
6.50
Values are means
4.18
5.16
5.46
6.24
6.75
±
±
±
±
±
±
0.06
0.14
0.14
0.11
0.12
(n
(n
(n
(n
(n
=
=
=
=
=
10)
9)
7)
6)
4)
SEM.
J Cereb Blood Flow Metab. Vol. 7. No.4. 1987
R.
382
P. KRAIG ET AL.
tions. The difference between plateau pHo and in­
jectant pH increased to 0.46 and 0.66 during intro­
duction of 5.00 and 4.50 pH solutions, respectively
(Table 2). Such disparity between injectant pH and
pHo is likely to reflect an increase in brain physico­
chemical H+ buffer content from acid-induced pro­
tein denaturation. This conclusion stems from the
observation (Kraig and Wagner, 1987) that the H+
buffer content of brain proteins can be increased by
exposure to acid. Brain proteins, like other selected
proteins (Hashemeyer and Hashemeyer, 1973), con­
tain H+ buffer groups that are normally available
for titration and others that are sequestered from
aqueous media. With exposure to acid, such pro­
teins begin to lose their normal three-dimensional
structure and thus expose more H+ buffer moieties
to the aqueous environment. Maximal exposure of
sequestered H+ buffer groups in brain homoge­
nates occurs at -4.5 pH (Kraig and Wagner, 1987).
Similar acid-induced denaturation of brain proteins
may also occur in vivo, since maximal H+ buff-
a
4.5
:\ ':�
�\
�
�
..
7.5
;0;-,'0'
FIG. 2. Composite of pHo changes from sodium lactate in­
jections. Typical recordings are shown of pHo changes from
intracerebral injections of 150 mM sodium lactate solutions
adjusted to 4.00 (a), 4.50 (b), 5.00 (c), 6.00 (d), and 6.50 (e)
pH. pHo was recorded with double-barreled H+-selective mi­
croelectrodes positioned adjacent to but 300 fLm behind the
tip of a microinjection needle. The latter in turn was posi­
tioned 800 fLm beneath the parietal pial surface. Constant
infusion (0.5 fLl/min for 20 min) began at the upward arrow
and stopped at the downgoing arrow. pHo fell to an approxi­
mately constant level (see Table 2) within 1-2 min after in­
jections began. pHo returned to baseline after injections. So­
lutions (6.50 and 6.00 pH) that did not result in brain necrosis
(e and d) were correlated with pHo records that returned to
baseline at analogous and rapid rates after injection of acid
stopped. Similarly, 5.00 pH sodium lactate (c) was correlated
with a pHo record that returned to baseline at the same rate
as after injection of solutions at 6.00 and 6.50 pH. Thus, al­
though injectant pH was 5.00 and brain was destroyed in a
volume with an approximate diameter of the injection
needle, pHo reached 5.55 only at the H+-selective microelec­
trode recording site in viable brain tissue and beyond the
edge of necrosis. On the other hand, sodium lactate injec­
tants at 4.50 and 4.00 pH (b and a) that caused pHo to fall
below 5.30 pH (see Fig. 3) produced a necrotic zone that in­
cluded the H+-selective microelectrode recording point.
After these latter two injections, pHo returned to baseline
more slowly. This suggests that H+ homeostatic mecha­
nisms are impaired immediately after exposure to H+, which
ultimately results in brain necrosis.
J Cereb Blood Flow Metab, Vol. 7, No.4, 1987
ering of injectants was seen at 4.5 pH (Table 2). Re­
duction of brain H+ buffer content through acid-in­
duced protein precipitation may account for the ob­
servation that plateau pHo was only 0.18 higher
than the injectant during introduction of 4.00 pH
sodium lactate.
The rate of return of pHo to baseline after mi­
croinjections correlated with the ultimate effect of
injectants on brain. Injectants (6.50, 6.00, 5.00 pH)
that failed to produce brain necrosis at the site of
the H+ -selective microelectrode recording resulted
in pHo changes that returned to baseline at rapid
and virtually identical rates. Alternatively, pHo re­
turned to baseline more slowly after injection of so­
lutions (4.50 and 4.00 pH) that produced brain ne­
crosis. These results suggest that H+ -related ho­
meostatic mechanisms are impaired immediately
after exposure to solutions sufficiently acidic to
produce brain necrosis (see Discussion).
Histological examination of injection zones
Injection of sodium lactate solutions at a rate of
0.5 !-LlImin caused a new steady-state pHo to be
reached that extended -300 !-Lm to either side of
the midline of the injection cannula (Fig. 3, left).
The size of the involved volume of brain provided
an easily observable area of necrosis with the more
acidic injectants (Fig. 3, right). Necrotic diameters
of acidic lesions were measured from frozen sec­
tions (i.e., Fig. 3, right) and cellular detail was ana­
lyzed from perfusion-fixed brains stained with he­
matoxylin and eosin (Fig. 4). Sodium lactate injec­
tions that resulted in pHo values at or below 5.30
pH (Fig. 5) were associated with a well-circum­
scribed zone of coagulation necrosis. The necrotic
zone was surrounded by tissue where neurons
showed eosinophilic necrosis, astrocytes had en­
larged nuclei, and monocytes were present. Injec­
tions producing brain pHo above 5.30 pH were as­
sociated only with damage from the injection
needle-H+ -selective microelectrode array inser­
tion into brain and alteration of the neuropil from
fluid injections. Such injections revealed a central
zone of necrosis (61 ± 5 !-Lm in diameter) that re­
sulted from needle tract damage that was sur­
rounded by an altered neuropil extending 100-300
!-Lm to either side of the injection tract, which was
vacuolated and contained occasional neurons
showing eosinophilic necrosis (Petito et al., 1987).
DISCUSSION
This investigation shows that injection of lactic
acid into brain sufficient to produce a pHo at or
below 5.30 for a duration of 20 min produced local
areas of tissue necrosis. This level of acidity closely
HYDROGEN IONS KILL BRAIN
383
4.5
5.5
A
(OlJm)
pH
6.5
10 min
_____
7.5
4.5
5.5
pH
6.5
__ ._. ___ ._._._
C
(340IJm)
-'-------=-
7.5
pH
-- --- C:�
FIG. 3. Sodium lactate injection profile. Three double-barreled H+-selective microelectrodes (left) were glued together and then
positioned with an injection needle so that they were 300 f.Lm above the needle's tip and 0 (A), 200 (B), and 340 (C) f.Lm away from
the needle's wall. Sodium lactate (150 mM at 4.00 pH) was injected at 0.5 f.Ll/min for 20 min at a depth of 800 f.Lm below the
parietal pial surface. Twenty-micron-thick frozen sections stained with hematoxylin and eosin (right) showed the resultant ne­
crotic lesion and the approximate H+-selective microelectrode recording points (black arrows). pHo reached 4.11 (A) and 4.66
(8) at the two recording sites that were within the necrotic zone (left and center arrows), pHo reached 7.19 (C) at the recording
point (right arrow) that was outside the necrotic zone. Note that signals for 8 and C records were filtered through 10-Hz active
filers while the signal for record A passed through a 200-Hz active filter. Thus, recording A shows more electrical noise (i.e.,
thicker line) than recording 8 or C.
approximates that proposed to occur in glia (i. e. ,
5. 2 pH) during nearly complete ischemia (Kraig et
al. , 1986a) and is less than that directly observed
with intracellular H+ -selective microelectrodes to
occur within these cells during complete ischemia
under hyperglycemic conditions (Kraig and
Chesler, 1987; Kraig and Nicholson, 1987). Fur­
thermore, such hyperglycemia is associated with a
pHo in brain of 5. 4 pH shortly after reperfusion
from nearly complete ischemia (Kraig et aI. ,
1985b). Thus, 5. 2-5. 4 pH is a level of acidity that
can occur in brain during ischemia and is shown
here to cause brain necrosis. The pattern of ne­
crosis observed here from sodium lactate injections
resembled that seen in ischemic brain infarction: A
zone of coagulation necrosis is surrounded by a rim
of necrotic neurons and reactive glial cells.
In these experiments, the first hint of irreversible
injury to brain consisted of a slower than normal
rate for pHo to return to baseline after sodium lac­
tate injections (Fig. 2). Adjustment of pHo toward
baseline after acid injections can be proposed to
operate by at least two mechanisms. First, passive
diffusion or bulk flow of excess acid away from or
H+ buffers toward the injection zones could raise
pHo toward baseline. After nonlethal injections,
such movement of proton equivalents would be re­
stricted to the interstitial space because of retained
plasma membrane integrity. Alternatively, if cell
membranes were immediately damaged during se­
verely acidic injections, proton equivalents would
be expected to diffuse into injured cells from the
interstitial space. Movement of proton equivalents
across leaky (and irreversibly injured) cell mem­
branes probably occurred after injection of the
most acidic (i. e., 4.00 pH) sodium lactate since this
J Cereh Blood Flow Metab, Vol. 7, No. 4, 1987
384
R.
P. KRAIG ET AL.
FIG. 4. Necrotic lesions from sodium lactate injections. Luxol fast blue-and cresyl violet-stained section is shown. Necrotic
lesion was produced by injection (0.5 fLl/min for 20 min) of 150 mM sodium lactate (4.50 pH) at a depth of 800 fLm below the
parietal pial surface. pHo was recorded as in Fig. 2 and reached 4.85 during the injection. A well-circumscribed area of coagula­
tion necrosis was produced. Most neurons within this area appeared as eosinophilic ghosts or showed changes similar to those
seen in ischemia with shrunken eosinophilic cytoplasm and pyknotic nuclei. Nuclear fragmentation was evident and nuclear
debris was scattered throughout the lesion. Arrow (left) points to an eosinophilic neuronal ghost, which is also shown at a
higher magnification (right). The area of necrosis was surrounded by a thin band of altered neuropil. Neurons with eosinophilic
necrosis were present in this band. In addition, perineuronal vacuolization, proliferation of microglial cells, swollen astrocytic
nuclei, and a mild proliferation of capillaries were evident in this border zone. Calibration bar in lower right-hand corner of each
figure is 100 fLm in length. (Modified from Kraig et aI., 1986b.)
level of acidity has previously been shown to per­
manently denature brain proteins (Kraig and
Wagner, 1987) and has been shown here to be asso­
ciated with exhaustion of brain H+ buffer capacity.
Other lethal injectants (i. e. , 4.50 and 5.00 pH) may
not have immediately destroyed cell membranes
since they were associated with the largest degree
of brain H+ buffering (i. e. , greatest disparity be­
tween injectant pH and plateau pHo during injec­
tions). After the introduction of nonlethal 4.50 and
5.00 pH injectants, pHo could have returned toward
baseline by a second mechanism: exchange trans­
port (antiport) of proton equivalents across plasma
membranes. For example, HC03-, the principal
H+ buffer of interstitial space, is thought to emaJ Cereb Blood Flow Metab, Vol. 7, No.4, 1987
nate from glia by antiport with interstitial chloride
(Kimelberg and Bourke, 1982; Kraig et aI. , 1986a).
After nonlethal injections an increased rate of
HC03 - delivery to the interstitial space might
hasten the return of pHo to baseline. Similarly, if
cell membranes remain intact, HC03- might be the
only H+ buffer added to the interstitial space
during and after the introduction of lethal injectants
at 4.50 and 5.00 pH. The intermediately slowed re­
turn of pHo to baseline seen with these latter injec­
tants (Fig. 2) may reflect a pH sensitivity for
HC03- delivery to interstitial space via CI-I
HC03- antiport. In red blood cells, CI-IHC03antiport is a pH-dependent process that is almost
completely inhibited at 5.0-5.5 pH (Cabantchik et
HYDROGEN IONS KILL BRAIN
,.
'.
pH
7�--�--�--�--�--�--__--�
4
5
3
o
diameter 01 necrosi. (�m x 100)
FIG. 5. Comparison of pHo during injections to resultant size
of necrotic lesions.. Maximum pHo values reached during in­
jection of sodium lactate solutions are compared to the di­
ameter of the resultant necrotic lesion measured from frozen
sections (black dots). The latter were prepared and stained
with hematoxylin and eosin (as in Fig. 3) after allowing an­
imals to recover from acid injections for 24 h. Eight points
below 100 fLm in diameter (61 ± 5 fLm SEM) represent diam­
eter of the altered neuropil from needle tract damage and
not necrosis. After these points, necrotic lesion diameter in­
creased in a sigmoidal fashion compared to pHo. The line
through these points is drawn by sight and emphasizes the
observation that brain necrosis began when pHo reached
and then fell below 5 30
�
.
.
aI., 1978). If Cl-IHC03 - antiport in glia has a sim­
ilar pH sensitivity, then the slowed return of pHo
after introduction of 4.50 and 5.00 pH injectants
could indicate an acid-induced inhibition of HC03 secretion by glia.
The cellular mechanisms by which acid injections
produce brain necrosis are unknown but seem
likely to accompany excessive cellular acidosis.
When pHo is lowered to ,,;;;5 pH, one can expect
the intracellular milieu of neurons and glia to be­
come progressively acidotic because of pH-related
effects on plasma membrane antiport systems. Bar­
nacle muscle, for example, becomes progressively
acidified when pHo is lowered below 6.0 pH (Boron
et aI., 1979), presumably by reversal of Na+/H+
antiport, so that interstitial protons enter the cell
instead of being expelled from cells in the normal
manner. A similar mechanism coupled to acid-in­
duced inhibition of CI- IHC03 - antiport could
acidify mammalian neurons and glia during suffi­
ciently acidic sodium lactate injections. Both
Na+IH+ and Cl-IHC03 - antiport are present in
mammalian astrocytes (Kimelberg and Bourke,
1982) but they have not been demonstrated in
mammalian neurons. Their documentation, how­
ever, in several other animal cell types (Roos and
Boron, 1981; Thomas, 1984) including vertebrate
neurons (Chesler and Nicholson, 1985) implies that
N a+ IH+ and Cl-IHC03 - antiporters probably
exist in mammalian neurons as well.
The molecular basis by which excessive cellular
acidity destroys brain is unknown. Preliminary ex­
periments by Pulsinelli and Petito (1983) using an
385
acid injection paradigm suggest that the acidic re­
duction of ferric to ferrous iron and release of
ferrous iron from organic stores in the tissue may
catalyze the generation of active oxygen radicals
with subsequent lipid peroxidation of cell mem­
branes (Pulsinelli et aI., 1985).1
Acknowledgment: This study was supported by the Na­
tional Institutes of Neurological and Communicative Dis­
orders and Stroke grant nos. NS- 19 108, NS-003346,
BRSG S07, RR05396 and a Teacher Investigator Develop­
ment Award (no. NS-00767) as well as a Redel Founda­
tion and Du Pont Foundation grant to R.P.K. The expert
technical assistance of D. Vecchio and C. Pike is grate­
fully acknowledged.
REFERENCES
Ammann D, Lanter F, Stevens RA, Schulthess P, Shijo Y,
Simon W (1981) Neutral carrier based hydrogen ion selec­
tive microelectrode for extra- and intracellular studies. Anal
Chern 53:2267-2268
Bancroft JD, Stevens A (1982) Theory and Practice of Histolog­
ical Techn iques. New York, Churchill Livingstone
Boron WF, McCormick WC, Roos A (1979) pH regulation in
barnacle muscle fibers; dependence on intracellular and ex­
tracellular pH. Am J PhysioI 237:C I 85-C I 93
Bures J, Buresova 0, Ktivanek J (1974) The Mechanism and
Applications of Leao's Spreadin g Depression of Electroen­
cephalographic Activity. New York, Raven P ress
Cabantchik ZI, Knauf PA, Rothstein A (1978) The anion trans­
port system of the red blood cell. Biochim Biophys Acta
515:239-302
Chesler M, Nicholson C (1985) Regulation of intracellular pH in
vertebrate central neurons. Brain Res 325:313-316
Deutsch S (1971) The pharmacodynamics of halothane. In: Text­
book of Veterinary Anesthesia (Soma LR, ed), Baltimore,
Williams & Wilkins, pp 68-74.
Hashemeyer RH, Hashemeyer AEV (1973) Proteins-A Study
of Physical an d Chemical Methods. New York, John Wiley
& Sons, pp 352-385
Kalimo H, Rehncrona S, Soderfe1dt B, Olsson Y, Siesjo BK
(1981) Brain lactic acidosis and ischemic cell damage. 2.
Histology. J Cereb Blood Flow Metab 1:313-327
Kimelberg HK, Bourke RS (1982) Anion transport in the
nervous system. In: Handbook of Neurochemistry, Vol 1:
Chemical and Cellular Architecture, 2nd ed. (Lajtha A, ed),
New York, Plenum, pp 31-67
Kraig RP, Chesler M (1987) Glial acid-base behavior in brain
1 Sodium lactate solutions were originally injected into brain
through stainless-steel needles (Pulsinelli and Petito, 1983).
These early injections produced brain necrosis when injectant
pH was �6.0. When the injection paradigm was changed to one
that used fused silica instead of stainle ss steel, the threshold for
brain necrosis dropped to 4.50-5.00 for injectant pH (Thble 2)
and 5.30 for pHo (Fig. 5). Acidity in the range of 6.50-6.00 pH
enhances the solubility of ferrous iron so that more acidic so­
dium lactate solutions that passed through stainless-steel
needles contained higher quantities of iron than less acidic solu­
tions (w. A. Pulsinelli, unpublished observations). Therefore,
the additional injection of iron in the early experiments may
have increased intracerebral ferrous iron; that in tum could have
catalyzed the formation of active oxygen radicals with subse­
quent lipid peroxidation of cellular membranes (Pulsinelli et aI.,
1985). Because of this possibility, all solutions used in the
present experiments were never exposed to iron-containing ma­
terials.
J Cereb Blood Flow Metab, Vol. 7, No.4, 1987
386
R.
P. KRAIG ET AL.
ischemia [Abstract). J Cereb Blood Flow Metab 7(supp\
\):S 126
Kraig RP, Cooper AJL (1987) Bicarbonate and ammonia changes
in brain during spreading depression. Can J Physio/ Phar­
macoI65:1099-1104
Kraig RP, Nicholson C (1987) Profound acidosis in presumed
glia during ischemia. In: Cerebrovascular Diseases, 15th
Princeton-Williamsberg Conference (Raichle ME , Powers
WJ, eds), New York , Raven Press , pp 97-102
Kraig RP, Wagner RJ (1987) Acid-induced changes of brain pro­
tein buffering. Brain Res 410:390-394
Kraig RP, Ferreira-Filho CR , Nicholson C (1983) Alkaline and
acid transients in the cerebellar microenvironment. J Neu­
rophysioI49:831-849
Kraig RP, Pulsinelli WA, Plum F (1985a) Hydrogen ion buffering
in brain during complete ischemia. Brain Res 342:281-290
Kraig RP, Pulsinelli WA, Plum F (1985b) Heterogeneous distri­
bution of hydrogen ions and bicarbonate ions during com­
plete brain ischemia. In: Progress in Brain Research, Vol
63: Mechanisms of Ischemic Brain Injury (Kogure K , Hoss­
mann K- A, Siesj6 BK , Welsh FA, eds), Amsterdam , Else­
vier/North-Holland , pp 155-166
Kraig RP, Pulsinelli WA, Plum F (1986a) Carbonic acid buffer
changes during complete brain ischemia. Am J Physiol
250:R348-R357
Kraig RP, Petito C , Pulsinelli WA, Plum F (1986b) Lactacidosis
and brain irUury during severe or complete ischemia. In:
Acute Brain Injury: Medical and Surgical Therapy ( Batti­
stini W, Fiorani R , Coutbier R , Plum F , Fieschi C, eds),
New York , Raven Press , pp 35-40
Kraig RP, Petito C , Plum F , Pulsinelli WA (1986c) Hydrogen ions
and ischemic brain damage. Ann NY Acad Sci 481:372-374
Myers RE (1979) Lactic acid accumulation as a cause of brain
edema and cerebral necrosis resulting from oxygen depriva­
tion. In: Advances in Perinatal Neurology (Korbkin R ,
Guilleminault G , eds), New York , Spectrum , pp 85-114
Nemoto EM , Frinak S (1981) Brain tissue pH after global brain
ischemia and barbiturate loading in rats. Stroke 12:77-82
J Cereb Blood Flow Metab, Vol. 7, No.4, 1987
Petito CK , Kraig RP, Pulsinelli WA (1987) Light and electron
microscopic evaluation of hydrogen ion-induced brain ne­
crosis. J Cereb Blood Flow Metab (in press)
Plum F (1983) What causes infarction in ischemic brain? Neu­
rology 33:222-233
Pulsinelli WA, Brierley JB (1979) A new model of bilateral hemi­
spheric ischemia in the unanesthetized rat. Stroke 10:267-
272
Pulsinelli WA, Duffy TE (1983) Regional energy balance in rat
brain after transient forebrain ischemia. J Neurochem
40:1500-1503
Pulsinelli WA, Petito CK (1983) The neurotoxicity of hydrogen
ions [Abstract). Stroke 14:130
Pulsinelli WA, Waldman S, Rawlinson D , Plum F (1982) Mod­
erate hyperglycemia augments ischemic brain damage: a
neuropathologic study in the rat. Neurology 32:1239-1246
Pulsinelli WA, Kraig RP, Plum F (1985) Hyperglycemia , cerebral
acidosis , and ischemic brain damage. In: Cerebrovascular
Diseases, 14th Research Princeton-Williamsburg Confer­
ence (Plum F , Pulsinelli WA, eds), New York , Raven Press ,
pp 201-205
Roos A, Boron WF (1981) Intracellular pH, Physio/ Rev 61:296-
434
Siemkowicz E , Hansen AJ (1981) Brain extracellular ion com­
position and EEG activity following 10 min ischemia in
normo- and hyperglycemic rats. Stroke 12:236-240
Smith TC (1971) Respiratory effects of general anesthesia, In:
Textbook of Veterinary Anesthesia (Soma LS , ed), Balti­
more , Williams & Wilkins , pp 156-177
Thomas RC (1984) Experimental displacement of intracellular
pH and the mechanism of its subsequent recovery, J Physiol
354:317-338
Thorn W, Heitmann R (1954) pH der Gerhirnrinde vom Kanin­
chen in Situ wahrend perakuter, totaler Ischamie , reiner
Anoxie und in der Erholung. Pflugers Arch 258:501-510