Reaccumulation of [K +] o
in the Toad Retina
During Maintained Illumination
HIROSHI SHIMAZAKI and BURKS OAKLEY 11
From the Departments of Electrical and Computer Engineering and Biophysics, and the Bioengineering Program, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801
. was measured in the
Using K''-selective microelectrodes, [K+]
subretinal space of the isolated retina of the toad, Bufo marinus. During
maintained illumination, [K''], fell to a minimum and then recovered to a steady
level that was ^-0.1 mM below its dark level. Spatial buffering of [K*] o by Miiller
(glial) cells could contribute to this reaccumulation of K+. However, superfusion
with substances that might be expected to block glial transport of K+ had no
significant effect upon the reaccumulation of K+. These substances included
blockers of gK (TEA', Cs', Rb+, 4-AP) and a gliotoxin (aAAA). Progressive
slowing of the rods' Na'/K-' pump (perhaps caused by a light-evoked decrease
in [Na+]i) also could contribute to this reaccumulation of K' by reducing the
uptake of K' from the subretinal space. As evidence for a major contribution
by this mechanism, treatments designed to prevent such slowing of the pump
reversibly blocked reaccumulation . These treatments included superfusion with
2 uM ouabain, or lowering [K+ ] o, P02, or temperature. It is likely that such
treatments inhibit the pump, increase [Na'']i, and attenuate any light-evoked
decrease in [Na+]i. The results are consistent with the following hypothesis. At
light onset, the decrease in rod gNa will reduce the Na' influx and the resulting
rod hyperpolarization will reduce the K+ efflux . In combination with these
reduced passive fluxes, the continuing active fluxes will lower both [K'']
. and
;, which in turn will inhibit the pump. In support of this hypothesis, the
[Na"]
solutions to a pair of coupled differential equations that model changes in both
[K+ ] o and [Na+]i match quantitatively the time course of the observed changes
. during and after maintained illumination for all stimuli examined .
in [K+]
ABSTRACT
INTRODUCTION
Illumination of the vertebrate retina evokes a significant decrease in the extracellular potassium ion concentration, [K +]o, in the subretinal space. This lightevoked decrease in [K+ ]o has been observed in all species examined, including
frog (Tomita, 1976 ; Oakley and Green, 1976), toad (Oakley et al., 1979),
mudpuppy (Karwoski and Proenza, 1978), skate (Kline et al., 1978), gecko (Griff
Address reprint requests to Dr . Burks Oakley 11, Dept . of Electrical and Computer Engineering,
University of Illinois at Urbana-Champaign, 1406 W. Green St ., Urbana, IL 61801 .
© The Rockefeller University Press - 0022-1295/84/09/0475/30 $1 .00
Volume 84 September 1984 475-504
J. GEN. PHYStOL.
475
476
THE JOURNAL OF GENERAL PHYSIOLOGY " VOLUME
84 - 1984
and Steinberg, 1984), rabbit (Dick and Miller, 1978), and cat (Steinberg et al .,
.seems to be
1980). In rod-dominated retinas, the light-evoked decrease in [K+]
produced primarily by the rod photoreceptors themselves (Matsuura et al ., 1978 ;
Oakley et al ., 1979 ; Steinberg et al ., 1980). With maintained illumination, [K'']o
falls to a minimum and then begins to recover back toward the dark-adapted
level (this latter process is termed the "reaccumulation" of K'''). With continuous
illumination, [K+]o eventually recovers by an amount equal to 65-85% of its
initial decrease at light onset (Steinberg et al ., 1980 ; Oakley and Steinberg,
1982 ; Oakley, 1983). Following termination of maintained illumination, there is
.above the dark-adapted baseline (an overshoot) that
a transient increase in [K']
.at light onset.
nearly is a mirror image of the transient decrease in [K+]
The reaccumulation of K+ during maintained illumination is likely to be of
physiological significance . For example, membrane voltage in retinal pigment
epithelial (RPE) cells is determined in part by [K'] o in the subretinal space (Miller
and Steinberg, 1977), and in frog, changes in RPE membrane voltage follow
[K'], both during and after maintained illumination (Oakley and Steinberg,
1982) . Any RPE transport processes that depend either on membrane voltage
or on [K *]e (e .g ., Miller and Steinberg, 1979) will be affected by the reaccumulation of K' . In addition, although retinal sensitivity is affected by changes in
[K + ] o (Dowling and Ripps, 1976), the reaccumulation of K* makes it unlikely
that maintained changes in sensitivity will be caused by [K''] o (Steinberg et al .,
1980). One of the most significant aspects of the reaccumulation of K' is that it
helps to re-establish a stable extracellular environment, that is, a [K*]o homeostasis.
Two quite different mechanisms could contribute to the reaccumulation of K'
during maintained illumination . One of these mechanisms is the spatial buffering
of [K+ ] o by glial cells, whereby the high K+ conductance, gx, of glial cells will
. and exit in regions of low
allow K' to enter glial cells in regions of high [K'']
[K'] o. In some tissues, such as drone retina (Coles and Tsacopoulos, 1979 ;
Gardner-Medwin et al ., 1981 ; Coles and Orkand, 1983), cat cortex (Dietzel et
al ., 1980), rat neocortex (Gardner-Medwin, 1983 ; Gardner-Medwin and Nicholson, 1983), and rat optic nerve (Yamate et al ., 1983), spatial buffering of [K *]o
by glial cells is involved in the regulation of [K'']o. In fact, during maintained
depolarizing stimulation of several of these tissues, spatial buffering of [K'],, by
glial cells can produce changes in [K']o similar to those observed in the vertebrate
retina during and after maintained illumination (but of opposite polarity).
Another mechanism that could contribute to the reaccumulation of K'' in the
vertebrate retina involves changes in the activity of the Na*/K' pump in the rod
membrane, or in other cells bordering on the subretinal space (Steinberg et al .,
1980 ; Oakley, 1983, 1984). In tissues as diverse as cat cortex (Heinemann and
Lux, 1975, 1977 ; Nicholson et al ., 1978), frog ventricular muscle (Kunze, 1977 ;
Martin and Morad, 1982 ; Kline and Kupersmith, 1982), guinea pig hippocampus
(Benninger et al ., 1980), and rat sympathetic ganglia (Galvan et al ., 1979),
changes in the activity of Na''/K' pumps are involved in the regulation of [K *]o.
During maintained depolarizing stimulation, an increase in the activity of the
; (Cohen et al ., 1982 ; Ballanyi et
Na''/K + pump, caused by an increase in [Na']
SHIMAZAKI AND OAKLEY
Retinal (K'J, During Maintained Illumination
477
al ., 1983), produces changes in [K']o similar to those observed in the retina
during and after maintained illumination (but of opposite polarity). By analogy
with these tissues, inhibition of the rods' Na'/K' pump during maintained
illumination, perhaps caused by a decrease in [Na']i, could contribute to the
reaccumulation of K+ (Oakley, 1983).
In previous experiments (Oakley, 1983), it was found that superfusion of the
isolated retina of the toad with 2.0 mM Ba 2+ reversibly blocked the reaccumulation of K+ . This result did not help to determine the relative contributions of
the two mechanisms that were suggested (above) to be involved in the reaccumulation process, since Ba 2+ could block gK in glial cells and thus block the
movement of K+ through glial cells, but Ba2+ also could have direct inhibitory
effects on the Na'/K' pump at the relatively high concentrations used (Ellory et
al ., 1983 ; Oakley, 1983).
The experiments reported in this paper were designed to affect the two
different mechanisms selectively, in order to provide information regarding the
relative contributions of these mechanisms to the reaccumulation of K+ during
maintained illumination . Experiments designed to affect the spatial buffering of
[K + ] o by glial cells had little effect on the reaccumulation process, while experiments designed to affect Na'/K' pumps blocked reversibly the reaccumulation
process. On the basis of these data, a model of the ionic mechanisms involved in
. was developed, and this model can be used to explain
the regulation of [K+]
quantitatively the kinetics of the observed changes in retinal [K+]
. during and
after maintained illumination under a wide variety of experimental conditions .
METHODS
Preparation
All experiments were performed on the isolated retina preparation of the toad, Bufo
marinas, as described in detail recently (Oakley, 1983). Briefly, the isolated retina was
pinned, receptor side up, in a small chamber (--0 .25 ml vol) that had a transparent bottom .
This chamber was placed on the stage of a compound microscope, and the retina was
viewed using infrared illumination (>850 nm) and an image converter. The experiments
were performed at room temperature (22-24°C), unless otherwise noted.
Solutions and Experimental Conditions
During an experiment, the retina was superfused (1 .5-2 .0 ml/min) with an oxygenated
Ringer's solution that had the following composition (in mM): 110 NaCl, 2 .4 KG, 0 .9
CaCl2, 1 .3 MgCl2, 0.22 NaH2PO 4, 2.78 Na2HPO4, 5 .6 glucose, 0.01 EDTA, and 0.003
phenol red . The pH of this solution was 7.8 . Various test solutions were prepared by
slight alterations in the control solution . The effects of Rb' were investigated by equimolar
substitution of RbCl for 2.4 mM KCI. The effects of Cs' or tetraethylammonium ion
(TEA') were investigated by equimolar substitution of 1 .0-10.0 mM CsCl or 5.0-10.0
mM TEA-Cl for NaCI . The effects of 4-aminopyridine (4-AP) were investigated by the
addition of 1 .0-2 .0 mM 4-AP to the control solution . Solutions having lowered K' were
made by equimolar substitution of NaCl for KCI. A solution containing DL-a-aminoadipic
acid (aAAA) was made by equimolar substitution of aAAA for 5.0-10 .0 mM NaCl .
Ouabain (1 .0-10.0 IM) was simply added to the control solution .
47 8
THE JOURNAL OF GENERAL PHYSIOLOGY " VOLUME 84 - 1984
In several experiments, the retinal temperature was lowered by bathing the retina in
cooled Ringer's solution . The temperature of the solution surrounding the retina was
measured with a miniature thermistor probe (model 514 ; Yellow Springs Instrument Co .,
Yellow Springs, OH). In other experiments, retinal P02 was lowered by bubbling the
Ringer's solution with N2 instead of with 0 2. Although this solution nominally had a P02
close to zero, it was likely that a small amount o£ 02 could diffuse into the solution when
it was exposed to the room atmosphere in the chamber. The exact P0 2 of the solution
surrounding the retina was not measured under these conditions .
Electrodes and Recording
Double-barreled, K+-selective microelectrodes were used to measure [K+]o. One barrel
was an ion-selective electrode and contained a K+-selective liquid (477317, Corning
Medical Products, Medfield, MA ; or 60031, Fluka Chemical Corp ., Hauppauge, NY) in
its tip ; the remainder of this barrel was filled with 0 .1 M KCI . The other barrel was a
reference electrode filled with 1 .0 M LiCl (Steinberg et al ., 1980) . The double-barreled
microelectrodes were beveled on a surface embedded with diamond dust (Brown and
Flaming, 1979) . After beveling, the resistance of the reference barrel was 25-40 MSl.
Each K+-selective microelectrode was advanced toward the receptor surface from above
under visual control, until the electrode tip just made contact with a rod outer segment .
The electrode then was advanced into the preparation in 2-Wm steps, using a piezoelectric
positioning system (Burleigh Instruments, Inc ., Fishers, NY) . In most experiments, the
electrode tip was positioned ^-40-60 km below the receptor surface, at a depth where
both the amplitude and the initial rate of change of the light-evoked decrease in [K+],
were maximal (Oakley et al ., 1979) .
Immediately after an experiment, each K+-selective microelectrode was calibrated in
solutions having varying [K + ] and a fixed background of 110 mM [Na'] . The calibration
data were fitted by an equation of the form :
VK + ~ A log io ([K - J. +
S
)
+ V.,
where VK* is the differential potential between the two barrels (K+-selective barrel positive),
A is the logarithmic slope, S is the selectivity coefficient for K+ over Na + , and Vo is a
constant (Walker, 1971 ; Oakley, 1983) . The value of A was 55-58 mV/decade, while the
value of S was 60-70 for Corning electrodes and was 1,000-2,000 for Fluka electrodes .
Once the electrode calibration curve was determined, it was possible to convert entire
. by rearranging Eq . 1 into the form :
digitized waveforms of VK, into waveforms of [K+]
[K + ]o = 10(VKi^
[Na+]o
S
.
The Corning K+ ion exchanger has a greater sensitivity for Rb+ than for K+ (Wise et
al ., 1970), so the Corning electrodes could be used as Rb+-selective electrodes to measure
[Rb + ] o in the retina . However, the Corning ion exchanger also has a much greater
sensitivity for TEA+ , Cs', and 4-AP than for K + (Wise et al ., 1970 ; Neher and Lux, 1973 ;
B. Oakley, unpublished observations), so it was not possible to use the Corning electrodes
. in the presence of these interfering ions . Instead, the Fluka K+ cocktail,
to measure [K+]
which is based on valinomycin (Oehme and Simon, 1976 ; Wuhrmann et al ., 1979), was
used . At the extracellular concentration of K + (2 .4 mM), the Fluka electrodes had little
response to the concentrations of TEA + and 4-AP used in these experiments . The Fluka
electrodes had a larger response to Cs', however, since they are nearly as sensitive to Cs'
as to K + (Oehme and Simon, 1976) . Under control conditions, no differences were
SHIMAZAKI AND OAKLEY
Retinal [K +]o During Maintained Illumination
47 9
.
observed between the light-evoked decreases in [K+]
measured by either type of electrode.
However, the light-evoked changes in V, c" usually were larger in amplitude when the
Fluka electrodes were used, since there was less interference from extracellular Na +; the
resistance of the Fluka electrodes also was larger .
All microelectrode voltages were measured using capacity-compensated preamplifiers
having input resistances of 10'5 St . These voltages were referenced to a Ag/AgCl electrode
that was connected electrically via a KCl-agar bridge to the solution bathing the retina .
The electrode voltages were amplified, displayed on both an oscilloscope and a chart
recorder, and recorded on an instrumentation tape recorder for off-line analysis . The
recordings were digitized using a laboratory computer system and plotted on a digital
plotter (Oakley, 1983).
Light Stimulation
The retina was stimulated with 500 nm light, delivered through the microscope condenser.
The stimulus diameter in the plane of the retina was -2 mm, and all microelectrode
recordings were made from the center of the stimulated region . The stimulus duration
could be varied using an electromagnetic shutter. The stimulus irradiance was attenuated
with calibrated neutral density filters. For all responses illustrated, the stimulus irradiance
was 1 .3 log quanta s-' 1Im-2 .
RESULTS
Experiments Designed to Affect Glial Cells
Spatial buffering of [K + ] o by glial cells could contribute to the reaccumulation of
K+ during maintained illumination . Glial cells are known to have a large 9K, and
thus glial cells may be able to regulate [K']o by providing a current-mediated
transcellular flux of K+ (e .g ., Orkand et al ., 1966 ; Varon and Somjen, 1979 ;
Gardner-Medwin, 1983) . The differential hyperpolarization of Muller (glial) cells
by the light-evoked decrease in [K+]
. could lead to current flow and the transcellular movement of K+ through Muller cells. Such movement of K' might
possibly lead to a reaccumulation of K' similar to that actually observed . If this
is the case, then experimental conditions designed to interfere with Miller cells
should block the reaccumulation process.
Specifically, blocking gK in Muller cells should abolish any fluxes of K'' through
these cells. In a variety of cell types, substances such as Cs', Rb +, TEA', and 4AP are known to block K+ conductances (Hille, 1967 ; Narahashi, 1974 ; Hagiwara
and Takahashi, 1974 ; Hagiwara et al ., 1976 ; Meves and Pichon, 1977 ; Edwards,
1982). It is likely that these substances also block the gK of Muller cells, since
they abolish the slow PIII component of the electroretinogram (Winkler and
Gum, 1981 ; B. Oakley, manuscript submitted for publication), which is thought
to be generated by the Muller cells' hyperpolarizing response to the light-evoked
.(Witkovsky et al ., 1975 ; Fujimoto and Tomita, 1979). These
decrease in [K+]
substances have little effect on the decrease in [K+ ] o evoked by brief flashes of
light (B . Oakley, manuscript submitted for publication) .
Moreover, the gliotoxin aAAA is known to disrupt Muller cell membranes
(Szamier et al ., 1981 ; Bonaventure et al ., 1981 ; Welinder et al ., 1982 ; Zimmerman and Corfman, 1984), and aAAA appears to disrupt the ability of Muller
48 0
THE JOURNAL OF GENERAL PHYSIOLOGY " VOLUME 84 - 1984
. (Karwoski et al ., 1982). If this is the case,
cells to respond to changes in [K+]
then application of aAAA also might be expected to disrupt the ability of Miller
cells to participate in spatial buffering of [K''], during maintained illumination .
In the first set of experiments designed to affect Muller cells, the retina was
bathed in various substances known to block gK in a wide variety of cells. The
results of these experiments are summarized in Fig. 1 . Each part of Fig. 1
illustrates the effect of a different test substance. With each test substance, a
control response to a 180-s period of illumination is shown, as well as a response
to the same stimulus recorded while bathing the retina in the test substance. In
Fig. 1 A, 4-AP was used . Even after bathing the retina in 1 .0 mM 4-AP for 26
min, there was no detectable effect upon the changes in [K']o. Similar results
were obtained with 2.0 mM 4-AP. In Fig. 1 B, TEA' was used . In 10 .0 mm
TEA-, the light-evoked changes in [K +]o were slightly different than under
control conditions, in that they were reduced to ^-75% of their control amplitudes. However, there still was a significant reaccumulation and overshoot in
10 .0 mM TEA' . In Fig. I C, Cs' was used. The change in VK+ was much smaller
during superfusion with 1 .0 mM Cs'; however, this voltage represented almost
the same change in [K + ] o, because of the significant response of the K''-selective
microelectrode (Fluka K' cocktail) to Cs'. Similar results were obtained with
concentrations of Cs' up to 5.0 mM. In Fig. 1 D, K'' was replaced completely by
Rb+. As observed elsewhere (B . Oakley, manuscript submitted for publication),
there was a light-evoked change in [Rb']o of essentially the same amplitude as
the change in [K +]o. Moreover, there was a reaccumulation of Rb' during
maintained illumination, and an overshoot of [Rb''']o following light offset .
Overall, the results illustrated in Fig. I can be summarized by stating that
substances known to block gK in a wide variety of cells had no significant effect
upon the reaccumulation of K+ during maintained illumination . The smaller
amplitude of the light-evoked decrease in [K'']o observed during superfusion
with TEA' or Cs' may have been due to effects of these substances upon Muller
cells or upon the rods themselves (e.g., Fain and Quandt, 1980). However, none
of these substances significantly affected the conductance in rods responsible for
the light-evoked decrease in [K']o, which is consistent with other findings (B .
Oakley, manuscript submitted for publication) .
In another series of experiments designed to affect glial cells, the retina was
bathed in various concentrations of the gliotoxin aAAA . The results illustrated
in Fig. 2 were obtained during an experiment in which 10 .0 mM aAAA was
used . In the upper trace of Fig. 2, a control response is shown. In the lower trace
of Fig. 2, a response is shown that was recorded after 3 min of superfusion with
. were reduced to
10 .0 mM aAAA . In 10 .0 mM aAAA, the changes in [K']
-75% of their control amplitudes . This result could have been produced by a
direct effect of aAAA upon the Muller cells or by neurotoxic effects of the DLisomer upon the rods (see Zimmerman and Corfman, 1984). Nevertheless, there
was a reaccumulation of K+ during maintained illumination, as well as an
overshoot at light offset . Qualitatively similar responses were observed even after
60 min of continuous superfusion with 10 .0 mM aAAA (data not illustrated) .
SHIMAZAKI AND OAKLEY
Retinal (K*Jo During Maintained Illumination
481
3.0 mM
Control
~
...
3 niV
60s
Control
23
40
.
[K*]o
2.5
2.0
LM_f
LM
C
3.0 mM
Control
2.0
3.0 mM
Control
2.5
60s
D
[K*]o
25
3 mV
60s
Rb*
2.0
.
(K*]
3.0 mM
2.5
2A
[Rb*]
o
1.
Effects of blockers of gx upon the light-evoked changes in [K+]o. Each
panel of this figure shows both a control response and a response recorded during
superfusion with a blocker of gx. All responses were evoked by light stimuli having
durations of 180 s, as indicated by the traces labeled LM (light monitor) . (A) Effects
of 4-AP. The lower response was recorded after 26 min of superfusion with 1 .0
mM 4-AP. (B) Effects of TEA'. The lower response was recorded after 4 min of
superfusion with 10 .0 mM TEA`. (C) Effects of Cs* . The lower response was
recorded after 10 min of superfusion with 1 .0 mM Cs' . The light-evoked decrease
in Vx+ was smaller during superfusion with Cs', because of the selectivity of the K+selective microelectrode for K+ over Cs* of only 1 .2 :1 . However, this change in
Vx " represented a change in [K+], of nearly the same amplitude as under control
conditions, after taking into account the electrode selectivity . (D) Effects of Rb' .
The lower response was recorded after 20 min of superfusion with a solution
in which K+ was replaced by Rb+ . Since the K+-selective microelectrode was
more sensitive to Rb* than to K+ , this waveform represents a light-evoked change
in [Rb+ ]o .
FIGURE
Experiments Designed to Affect the Rods' Na +/K + Pump
A decrease in the rate of the Na'/K' pump in the rods could contribute to the
reaccumulation of K + during maintained illumination . By analogy with other cell
; could progressively
types, it was suggested that a light-evoked decrease in [Na']
and
thus
decrease
the
active
uptake of K+ during this
inhibit the Na'/K' pump
r~
22°C,
illustrated
case,
passive
types,
retinal
should
the
period,
several
of
1973
and
Glynn
[Na']
and
smaller
[Na'']i
into
indicated
that
first
athen
aAAA
superimposed
in
[Na'']i
of
of
effects
response
if
temperature
not
influx
Nelson
low
the
the
effects
any
and
conditions,
2[K+]o
the
leading
account
set
the
the
isIn
where
in
superfusion
produce
dark
solution
of
temperature
of
pump
likely
stimulates
Karlish,
retina
light-evoked
by
Fig
on
magnitude
experimental
pump
of
following
experiments
K+
and
recorded
the
If
increased
to
the
3the
Na''
allto
the
already
A,
of
during
and
trace
to
the
was
The
Blaustein,
as
occurring
The
of
rate
1975)
increase
aAAA
pump
three
iswith
pump
22°C
large
the
the
observed
these
lowered
effects
light
reduced
than
labeled
after
(9°C)
initial
bymaintained
decrease
of
isresponses
adesigned
pump
lower
conditions
In
isactive
0aThese
inhibited
solution
isoffset
(Chapman
Cooling
the
effects,
relatively
Fig
23
of
decrease
1980
inhibited,
responses
during
slope
mM
LM)
aAAA
in
In
min
control
response
reaccumulation
Moreover,
responses
in
transport
3,
aduring
general,
In
are
containing
to
effects
of
The
Saito
[Na+]i
at
sigmoidal
itdesigned
illumination
the
affect
Fig
maintained
were
the
in9°C,
shown
less
seems
et
have
response
then
upper
[Na+]i
was
the
and
al3B,
retina
normalized
were
will
of
largely
sensitive
by
temperature
the
iffirst
and
lowering
recorded
been
the
10
1983)
alikely
Na'/K'
response
to
manner
Wright,
take
the
evoked
during
Na'/K'
control
a(Oakley,
abolished
few
inhibit
reduced
MM
irreversible
In
illumination
response
and
first
mM
pump
scaled
place
in
to
that
Fig
minutes
response
retinal
by
3pumps
aAAA
time
the
(Garry
also
was
changes
70
response
1982)
pump
the
is
min
3over
for
180-s
is
any
1983)
active
sC,
known
dark
recorded
recorded
reversibly
of
inhibited,
abolished
periods
pump
of
after
temperature
equal
the
The
progressive
(e
aand
in
the
light
Thus,
at
superfusion
should
in
region
recorded
efflux
the
If
Into
9°C
gradient
control
starting
level
should
Garra[Na']
amplistimuli
this
Skou,
under
when
other
rods,
after
have
itwas
the
the
be
of
of
is
48 2
THE JOURNAL OF GENERAL PHYSIOLOGY - VOLUME 84 - 1984
13.0
VKf
Control
FIGURE . Effects
(as
control
continuous
.
[K+]
with
.
time
the
have
Na'
the
level
cell
han,
;
possible
higher ;,
Taking
inhibition
minimized
In
the
significant
1965 ;
are
.
at
re-warming
reaccumulation
overshoot
(22°C)
tudes
slightly
.
.
.0
.2
.
.
.
.
.
.,
.
.
;
.
;
.
.
.
.
.g.,
.
:
.
.
.
.
.
.
SHIMAZAKI AND OAKLEY
A
Retinal [K+], During Maintained Illumination
3,0 mM
VK.
22° C
9°
1 2.0
B
K
12.2 mM
CA
483
20s
22°c
- 90C
LM _I
3.0 mM
_j
60s
C
WWidr°w
2 mV
L8
40 s
2.7 mM
2.4
2 .1
Effects of lowering retinal temperature. (A) These three responses
were evoked by identical stimuli having durations of 180 s (as indicated by the trace
labeled LM). The upper response was recorded under control conditions at 22 0 C,
the middle response was recorded 23 min after cooling the retina to 9°C, and the
lower response was recorded 28 min after warming the retina back to 22°C . The
level of [K+I° in the dark increased by 0.2 mM during the first few minutes of
cooling. The 2-mV calibration bar applies to the responses in parts A and C. (B)
The initial 70 s of the first control response (22°C) and the low temperature
response (9°G) have been scaled for equal peak amplitudes and superimposed . (C)
In another retina that had been maintained at 10°C for 10 min, the tip of the K+selective microelectrode was withdrawn into the bathing solution ^" 150 s after the
onset of a maintained stimulus .
FIGURE 3 .
of [K']° during maintained illumination was measured in another retina at 10 ° C .
Approximately 150 s after the onset of maintained illumination, the tip of the
K+-selective microelectrode was withdrawn into the bathing solution (having a
[K+] of 2.4 mM) . The change in electrode voltage showed that the steady level
of [K+]° in the subretinal space was below that of the bathing solution during
maintained illumination . The presence of this gradient indicated that an active
uptake mechanism was still functioning (e .g ., Martin and Morad, 1982); that is,
the Na+/K + pump was not completely inhibited.
In another series of experiments, the Na +/K + pump in the rods was inhibited
by bathing the retina in ouabain (e .g ., Skou, 1965 ; Frank and Goldsmith, 1967 ;
Glynn and Karlish, 1975). Various concentrations of ouabain were tried, in order
to find one that did not irreversibly abolish the light-evoked decrease in [K+]°
(Oakley et al ., 1979) . As shown in Fig. 4, a concentration of 2 /AM was found to
produce reversible effects on the changes in [K+] ° during and after maintained
illumination, similar to the effects of lowered temperature. It was not totally
unexpected to find that ouabain was effective at such a low concentration, since
in previous experiments, Torre (1982) found that the Na+/K+ pump in toad
rods was affected significantly by 3 AM strophanthidin, a cardioactive steroid
with effects similar to ouabain . In Fig. 4A, three responses are shown: a control
response, a response recorded after 17 min of bathing the retina in 2 juM ouabain
484
THE JOURNAL OF GENERAL PHYSIOLOGY - VOLUME 84 - 1984
/\
Control
V28M
K;
1 . M
u
2A
Ovoboin
W-6
LM
Control
Ouoboin
K+l
'O
40 s
2b ndvl
LM
2.8 mM
C
2.4
2 mV
B
withdraw
2.4
1 2.0
2.6 mM
_ ,2.4
60 s
LM
Effects of ouabain . (A) These three responses were evoked by identical
stimuli having durations of 180 s (as indicated by the trace labeled LM) . The upper
response was recorded under control conditions, the middle response was recorded
17 min after starting to superfuse the retina with a solution containing 2 UM
ouabain, and the lower response was recorded 95 min after returning to superfusion
with control solution . The level of [K + ]a in the dark increased by 0 .1 mM during
the first few minutes of superfusion with ouabain solution . The 2-mV calibration
bar applies to the responses in parts A and C. (B) The initial 104 s of the first control
response and the response in ouabain solution have been scaled for equal peak
amplitudes and superimposed. (C) In another retina that had been superfused with
2 uM ouabain for 19 min, the tip of the K+-selective microelectrode was withdrawn
into the bathing solution 150 s after the onset of a maintained stimulus .
FIGURE 4 .
solution, and a response recorded 95 min after switching back to control solution .
Ouabain reversibly attenuated both the reaccumulation of K' during maintained
illumination and the overshoot in [K+]
. at light offset . In Fig . 4B, the first 70 s
of the control response and the response in ouabain solution have been scaled
for equal amplitudes and superimposed . The initial slope of the normalized
response in ouabain was essentially the same as it was under control conditions.
In Fig . 4C, the gradient of [K']o was measured in another retina that had been
bathed in 2 uM ouabain for 19 min . As in Fig . 3C, the microelectrode was
withdrawn into the bathing solution ^-150 s after stimulus onset . The change in
the electrode voltage indicated that the level of [K+]
. in the subretinal space was
slightly below that of the bathing solution during maintained illumination . The
presence of this small gradient indicated that the Na'/K+ pump was not completely inhibited, since this active uptake mechanism could still decrease [K + ]o
slightly below the level of the bathing solution .
In yet another attempt to affect selectively the Na'/K' pump in the rods, the
retina was bathed in a solution having 0 mM K' (K +-free solution) . As with the
other treatments (lowering temperature, ouabain), low [K']o should inhibit the
Na'/K' pump (e .g., Garay and Garrahan, 1973 ; Saito and Wright, 1982). The
effects of K'-free solution are shown in Fig . 5. In Fig . 5A, three responses are
SHIMAZAKI AND OAKLEY
A
Control
K+-free
VK+
Retinal [K+]o During Maintained Illumination
3.0 mM
485
B
2.5
2 .0
[K+ 1o
20%
LM
Control
K+-free
C
withdraw
60 s
0.4 mM
1 0.05
FIGURE 5. Effects of K+-free solution . (A) These three responses were evoked by
identical stimuli having durations of 180 s (as indicated by the trace labeled LM).
The upper response was recorded under control conditions, the middle response
was recorded 20 min after starting to superfuse the retina with K'-free solution,
and the lower response was recorded 13 min after returning to superfusion with
control solution . The level of [K+]
. in the dark decreased to 0 .4 mM during the
first 20 min of superfusion with K'-free solution . The 2-mV calibration bar applies
to the responses in parts A and C. (B) The initial 70 s of the first control response
and the response in K'-free solution have been scaled for equal peak amplitudes
and superimposed . (C) In another retina that had been superfused with K'-free
solution for 27 min, the tip of the K+-selective microelectrode was withdrawn into
the bathing solution -170 s after the onset of a maintained stimulus.
shown: a control response, a response recorded after bathing the retina with K'free solution for 20 min, and a response recorded 13 min after restoring the
normal [K+]o. Bathing the retina in K'-free solution lowered [K'],, in the
subretinal space to ~0 .4 mM in 20 min . Lowering [K'']
.reversibly abolished the
reaccumulation of K+ during maintained illumination and also abolished the
overshoot of [K +]o following light offset . In Fig. 5B, the first 70 s of the control
response and the response recorded in K'-free solution have been scaled for
equal amplitudes and superimposed . The initial slope of the normalized response
in K'-free solution was essentially the same as it was under control conditions .
. during maintained illumination was measured
In Fig. 5 C, the gradient of [K+]
in another retina that had been bathed in K'-free solution for 27 min . As in
Figs . 3 C and 4 C, the microelectrode was withdrawn into the bathing solution
(nominally 0.0 mM [K+]) ^-170 s after the onset of maintained illumination . The
. in the
change in the electrode voltage indicated that the steady level of [K+]
subretinal space was higher than that of the bathing solution during maintained
illumination . The presence of this large gradient suggests that the retina was
continuously losing K+ to the bathing solution, as would be the case if the Na'/
K+ pump was very inhibited, so that there was little active uptake of K+.
48 6
THE JOURNAL OF GENERAL PHYSIOLOGY " VOLUME 84 - 1984
In one final series of experiments designed to affect the Na'/K' pump in rods,
retinal P0 2 was lowered. Since the pump has a high demand for 0 2 , the pump
should become inhibited under conditions of anoxia (Kimble et al ., 1980) . In
these experiments, several values of P02 were used. A solution bubbled with
. during and after
room air had a negligible effect on the changes in [K+]
maintained illumination . However, a solution bubbled with N2 (termed "low P02
solution") had significant effects on these changes in [K']o, as shown in Fig . 6. In
Fig . 6A, three responses are shown : a control response, a response recorded
after 22 min of bathing the retina in low P02 solution, and a response recorded
24 min after switching back to control solution . Low P0 2 reversibly attenuated
both the reaccumulation of K + during maintained illumination and the overshoot
in [K+], at light offset . In Fig . 6B, the first 70 s of the control response and the
response in low P02 solution have been scaled for equal amplitudes and superimposed . The initial slope of the normalized response in low P02 solution was
slightly less than it was under control conditions. In Fig. 6 C, the gradient of
. was measured in another retina that had been bathed in low P02 solution
[Kt]
for 14 min . The microelectrode was withdrawn into the bathing solution ^-150
s after stimulus onset . The change in the electrode voltage indicated that the
level of [K+]o in the subretinal space was essentially the same as that of the
A
VK+
Control
2.5
[K+]o
3 mV
60 s
N2
1.7
2.8 mM
1 2.4
3.0 mM
20 s
Control
N2
LMJ
C.
2.5
2.0
60 s
6. Effects of lowering P02 . (A) These three responses were evoked by
identical stimuli having durations of 180 s (as indicated by the trace labeled LM).
The upper response was recorded under control conditions, the middle response
was recorded 22 min after starting to superfuse the retina with a low P02 solution,
and the lower response was recorded 24 min after restoring the normal P02. The
. in the dark increased by 0 .2 mM during the first few minutes of
level of [K+)
supeffusion with low P02 solution . The 3-mV calibration bar applies to the responses
in parts A and C. (B) The initial 70 s of the first control response and the response
in low P02 solution have been scaled for equal peak amplitudes and superimposed .
(C) In another retina that had been bathed in low P02 solution for 14 min, the tip
of the K'-selective microelectrode was withdrawn into the bathing solution ^-150 s
after the onset of a maintained stimulus .
FIGURE
SHIMAZAKI AND OAKLEY
Retinal [K +], During Maintained Illumination
48 7
2.8 nVA
2b
2.4
100S
[K+]o
2.2
LM
FIGURE 7. Changes in VK. evoked by an increase in P02 and light. The retina had
been bathed in low PO 2 solution for 19 min prior to the start of this record . At the
time indicated by the arrow, the solution bathing the retina was switched to the
control (oxygenated) solution . Approximately 3 min later, a 180-s light stimulus was
given, as indicated by the trace labeled LM .
bathing solution during maintained illumination . This result suggests that the
Na'/K' pump was very inhibited, since this active uptake mechanism could not
decrease [K+]o below the level of the bathing solution .
In the experiments in which P02 was varied, complex changes in [K+]o were
observed when control P0 2 was restored after a period of time during which the
retina had been bathed in low P02 solution, as shown in Fig. 7. When the retina
suddenly was exposed to 02, the level of [K''] o fell abruptly, and then there was
a reaccumulation of K+ back toward the control level. This reaccumulation of
K+ was similar in time course to the reaccumulation of K+ during maintained
illumination, as shown by the response to a 180-s light stimulus, delivered as the
[K +
. were
level of
]o approached the control level. Similar changes in [K+]
observed when the retinal temperature was suddenly increased after a prolonged
period of low temperature (data not illustrated), and when normal [Ca 21]e was
restored after a prolonged period of I/10 normal [Ca2+]o (Oakley, 1984).
It is possible that the changes in [K+], illustrated in Fig. 7 were produced as
follows. Inhibition of the pump would cause an increase in [Na+]i, so that when
the normal P02 was restored, the pump would be stimulated by the high [Na']
;,
causing an increased uptake of K+ and a decrease in [K + ]o. As [Na +]i would
subsequently fall, the pump would slow and allow [K + ] o to reaccumulate . Similar
effects of altered [Na+]i were postulated to explain the effects of changing
[Ca2+10 observed previously (Oakley, 1984). When ouabain solution was used,
the level of [K +]o did not change as abruptly as in Fig. 7, presumably since the
effects of ouabain were slower to reverse. When 0 mM [K+] solution was used,
+]
any similar changes in [K o were not observed, since the level of [K']. was
increasing rapidly as the control conditions were restored .
DISCUSSION
The experiments reported in this paper investigated the relative contributions
of two quite different mechanisms to the reaccumulation of K+ during maintained
illumination . One mechanism involved the spatial buffering of [K + ]o by glial cells.
However, the reaccumulation of K+ during maintained illumination was not
affected significantly by conditions that were designed to block gK in glial cells.
48 8
THE JOURNAL OF GENERAL PHYSIOLOGY " VOLUME
84 - 1984
Moreover, a gliotoxin, which in other experiments seemed to disrupt the ability
of glial cells to respond to changes in [K + ] o (Karwoski et al ., 1982), had little
effect upon the reaccumulation process. Thus, these data seem to indicate that
the contribution of spatial buffering is minimal in this preparation . However,
any more definitive assessment of the contribution of spatial buffering must
await a direct demonstration that the glial cells actually were affected in the
assumed manner by the conditions employed in these experiments.
The other mechanism that was examined in these experiments involved
progressive inhibition of the rods' Na'/K' pump during maintained illumination,
and the experimental data are consistent with the idea that a major contribution
to the reaccumulation of K+ comes from this mechanism. Experimental conditions designed to affect Na'/K' pumps selectively had significant effects on the
. during maintained illumination . It is not likely that Na'}/K+
changes in [K'}]
pumps on cells other than the rods are participating in the reaccumulation
process in the isolated retina preparation . It will be shown (below) that the
reaccumulation process most likely involves the response of the Na'/K'
pump to some event other than just the decrease in [K + ] o (such as a decrease in
[Na''] ;). Since the reaccumulation of K+ is not affected by blocking synaptic
transmission to second-order neurons with aspartate (Oakley, 1983), it seems
unlikely that events are occurring in second-order neurons that would affect
their Na'/K' pumps in the necessary manner. In addition, it seems unlikely
that the Na+/K+ pumps on Miiller cells (Stirling and Lee, 1980) would be
affected in the necessary manner, since these pumps presumably respond only
to the light-evoked decrease in [K +]o.
A Possible Mechanism for the Reaccumulation Process
Matsuura, Miller, and Tomita (1978) developed and tested a model (termed
here the MMT model) of the light-evoked decrease in [K+]o, and this model
was supported in a subsequent study by Oakley et al . (1979) . According to this
model, the light-evoked hyperpolarization of the rod membrane, caused by a
decrease in Na conductance, gNa, reduced the driving force on K+ across the
rod membrane and thus reduced the passive efflux of K+ out of the rods. The
activity of the Na+/K+ pump in the rod membrane was assumed to be regulated
by [K+ ]o, in that the active influx of K+ into the rod was assumed to vary linearly
with [K + ]o. The difference between the active influx and the passive efflux of
K+ was the net uptake of K+ from the extracellular space, and it was this net
uptake that was responsible for decreasing [K +]o. The MMT model yielded a
linear, first-order differential equation describing the kinetics of the lightevoked decrease in [K+]o. In response to brief light stimuli, which caused rod
membrane responses having durations of
