Exp. Eye Res. (1992) 55, 539-550

Spatial Buffering of Extracellular Potassium by Mfiller (Glial) Cells in the Toad Retina BURKS

OAKLEY

II °be*, B R A D L E Y

J. K A T Z b'~, Z H I A N

X U c AND J I A N B I A O

ZHENG b

Department of Electrical and Computer Engineering, bNeuroscience Program, cDepartment of Biophysics, and ~Medical Scholars Program, University of Illinois at Urbana-Champaign, Urbana, IL, U.S.A. (Received Houston 26 August 1991 and accepted in revised form 8 January 1992) We examined the role of M/iller (glial) cells in buffering light-evoked changes in extracellular K+ concentration, [K+]o,in the isolated retina of the toad, Bufo marinus. We found evidence for two opposing Mtiller cell current loops that are generated by a light-evoked increase in [K+]o in the inner plexiform layer. These current loops, which are involved in the generation of the M-wave of the electroretinogram (ERG), prevent the accumulation of K+ in the inner plexiform layer by transporting K+ both to vitreous and to distal retina. In addition, under dark-adapted conditions, we found evidence for a Mfiller cell current loop that is generated by a light-evoked decrease in [K+]oin the receptor layer. This current loop, which is involved in the generation of the slow PIII component of the ERG, helps to buffer the light-evoked decrease in [K+]othroughout distal retina by transporting K+ from vitreous. The spatial buffering fluxes of K+ can be abolished by blocking Mfiller cell K÷ conductance with 200#M Ba >. The separate contributions of the M-wave and slow PIII currents to M/iller cell spatial buffering were isolated by various pharmacological treatments that were designed to enhance or suppress light-evoked activity in specific retinal neurons. Our results show that M611er cell K÷ currents not only buffer light-evoked increases in [K+Jo,but also buffer light-evoked decreases in [K+]o, and thereby diminish any deleterious effects upon neuronal function that could arise in response to large changes in [K+]o in the plexiform layers. Moreover, our results emphasize that spatial buffering currents generate many components of the electroretinogram. Key words: M/iller cell: glia; extracellular potassium; spatial buffering; barium; picrotoxin; 2-amino4-phosphonobutyric acid; kynurenic acid; electroretinogram. 1. Introduction in the vertebrate retina, light-evoked depolarization of second- and third-order neurons causes the release of K +, and thereby increases extracellular K+ concentration, [K+]o (Oakley and Green, 1976; Karwoski and Proenza, 1977). Retinal Mfiller (glial) cells are abnost exclusively permeable to K+, and Mfiller cell K + currents can transport a significant a m o u n t of K + from one retinal depth to another (Newman, 1985 a,b). The movement of K ÷ through the Miifier cells diminishes the amplitude of the light-evoked increases in [K+]o, and thereby prevents [K+]o from increasing to levels that might otherwise adversely affect neuronal function. This transport of K + by glial cells is called spatial buffering (Orkand, Nichols and Kuffler, 1967; Newman, 1986; Karwoski, Lu and Newman, 1989). Spatial buffering is especially critical in synaptic regions, where neuronal function appears to be maximally susceptible to fluctuations in [K+]o (Kuffler, Nichols and Martin, 1984). Frishman and Steinberg (1989b) investigated spatial buffering in the dark-adapted cat retina. In response to very dim illumination, neuronal release of

* For correspondence at: Department of Electrical and Computer Engineering. University of Illinois at Urbana-Champaign, 1406 West Green Street, Urbana, IL 6 1 8 0 1 - 2 9 9 1 , U.S.A.

0014-4835/92/100539 + 12 $08.00/0

K+ leads to an increase in [K+]o in the inner plexiform layer (IPL). This increase in [K+]o sets up a distally directed current loop in the Mfiller ceils, whereby K + enters the Mfiller cells in the IPL and leaves these cells through the large K+ conductance in their distal region (Newman, 1986, 198 7). This efflux of K + from the Mfiller cells in turn leads to a light-evoked increase in [K+]o in the receptor layer. It was shown that K + is transported to the receptor layer through the M/iller cells, rather than by extraceflular diffusion, since the light-evoked increase in [K+]o in the receptor layer could be blocked by the intravitrea] injection of Ba 2+ (Frishman and Steinberg, 1989b), a k n o w n blocker of Mfiller cell K + conductance (Newman, 1989). In the light-adapted mudpuppy retina, Karwoski et al. (1989) found that a light-evoked increase in [K+]o in the IPL set up a proximally directed current loop in the Mtiller cells, whereby K + entered the Mfiller cells in the IPL and left the cells through a large K+ conductance at their vitreal endfoot, the most proximal region of the cell. The efflux of K+ into the vitreous was sensed as a light-evoked increase in vitreal [K+]o. This efflux could be abolished by blocking Mfiller cell K + conductance with 50 ,aM Ba >, again indicating that K+ was being transported via a Mfiller cell pathway, rather than by extracellular diffusion. Thus, in both m a m m a l i a n and amphibian species, spatial buffering of [K+]o by Mfiller ceils transports K + away © 1992 Academic Press Limited 35-2

540

from the IPL, and thereby attenuates the amplitude of the light-evoked increase in [K+]o in the IPL. Such buffering may prevent the accumulation of high levels of [K+]o that would adversely affect synaptic transmission. The movement of K+ through Mfiller cells represents the flow of current in part of a loop, with the remainder of the loop involving the flow of current through the extracellular space. The extracellular current, which is carried by various ions (primarily Na + and C1-), has been shown to generate extracellular field potentials in the retina that are recorded as components of the electroretinogram (ERG). The current loops that arise in response to the light-evoked increases in [K+]oin the IPL generate the ERG scotopic threshold response (STR) in the mammalian retina (Sieving, Frishman and Steinberg, 1986b: Frishman and Steinberg, 1989a, b), as well as the ERG M-wave both in the amphibian retina (Karwoski and Proenza, 1977; Katz et al., 1991) and in the mammalian retina (Sieving, Frishman and Steinberg, 1986a; Frishman et al., 1991). The above studies focused primarily upon the lightevoked increase in [K+]o in the IPL. However, when the dark-adapted retina is stimulated by steps of illumination that saturate the rod voltage photoresponse, there is a large, light-evoked decrease in [K+]o in the receptor layer (Oakley and Green, 1976; Steinberg, Oakley and Niemeyer, 1980). Potassium diffuses from more proximal retinal depths down its concentration gradient towards the receptor layer, thereby decreasing [K+]o throughout distal retina. The decrease in [K+]o hyperpolarizes the Mfiller ceils throughout distal retina, and sets up a current loop whereby K+ leaves the distal regions of the Mfiller ceils and enters the Mfiiler cells at their proximal endfoot (Faber, 1969). The extracellular current that completes this loop generates an ERG component termed slow PIII (Faber, 1969 ; Witkovsky, Dudek and Ripps, 1975). Since this current loop is also a Mfiller cell spatial buffering current, the efflux of K+ from distal regions of the Mfiller cell would be expected to attenuate the amplitude of the light-evoked decrease in [K+]o throughout distal retina. In the present experiments, we sought to characterize Mfiller cell spatial buffering of [K+]o under conditions where both M-wave and slow PIII currents were generated. We used ion-selective microelectrodes (ISMs) to measure [K+]o in the isolated retina preparation of the toad, and we used different pharmacological treatments to isolate the contributions to spatial buffering of the M-wave and slow PIII currents. We were especially interested in [K+]o in the outer plexiform layer (OPL), since large changes in [K+]o could possibly affect neuronal function in this synaptic region. Moreover, there appears to be a localized increase in K+ conductance in the region of the Mfiller cell membrane within the OPL in amphibian Mfiller cells (Newman, 1985b, 1986), which could lead to

B. O A K L E Y II ET AL.

increased spatial buffering fluxes of K+ within the OPL. A preliminary report of these results was published in abstract form (Oakley et al., 1991). 2. Materials and Methods

Preparation and Solutions All experiments were performed on the isolated retina preparation of the toad, Bufo marinus, as described in detail previously (Oakley, 1983; Wen and Oakley, 1990a). Briefly, a dark-adapted toad was rapidly double-pithed, one eye enucleated, and a small piece of retina separated from the pigment epithelium. The isolated retina was pinned receptor-side up in a small chamber and was superfused with various solutions (22-24°C). The control solution was buffered with bicarbonate/CO2, and contained (in raM): 87"7 NaC1, 22-3 NaHCQ, 2"4 KC1, 0"9 CaCI 2, 1-3 MgCI~, 5.6 glucose, 0.01 EDTA, and 0"003 Phenol red. This solution was equilibrated with 98% 02/2% CO~ and had a pH of 7 . 8 0 + 0 . 0 2 . Several test solutions were made by adding one or more of the following chemicals to the control solution: ]OO/~M picrotoxin (PTX), IO0#M 2-amino-4-phosphonobntyric acid (APB), 2"0 ]/,M kynurenic acid (I(YN), or 200 #M BaC12. Test chemicals were obtained from either Sigma Chemical Co. (St Louis, MO, U.S.A.) or Mallinckrodt, Inc. (Paris, KY, U.S.A.).

K÷-Selective Microelectrodes Double-barrel, K÷-selective microelectrodes were used to measure [K+]o as described in detail recently (Wen and Oakley, 1990a,b). Briefly, these electrodes were made from thick-septum theta tubing (WPI Inc., Sarasota, FL, U.S.A.) and had beveled tips that were 0.5-1-0 # m across the beveled surface. The tip portion of the ion-selective barrel was filled with a K+-ion exchanger (#477317, Coming Medical Products, Medfield, MA, U.S.A.), and the remainder of this barrel was backfilled with 150 mM KC1. The reference barrel was filled with a solution containing 110 mM NaC1 and 2'4 mM KC1. The voltage of each barrel was measured with respect to a grounded Ag/AgCI electrode, which was connected to the bathing solution by a KCl-agar bridge. The differential amplifier used for this purpose had an input resistance of 101~ ~ (Axoprobe-1, Axon Instruments, Foster City, CA, U.S.A.) and had self- and cross-capacitance compensation adjustment (Dick and Miller, 1985). The differential voltage between the two barrels (K÷-barrel positive), which was termed V~, was a logarithmic measure of [K+]o. Electrodes were calibrated before and after an experiment (Oakley, 1983); they responded to changes in [K+]o with a logarithmic slope of 56-58 mV per decade and they had a 7 0 - 8 0 : 1 selectivity for K+ over Na +. For small changes around the resting level of retinal [K÷]o, the electrode response was essentially a linear measure of

K÷ S P A T I A L B U F F E R I N G

BY M U L L E R CELLS

541 Current sources in OPL

Distal-receptor layer (K + decrease) Outer plexiform layer (OPL)

M-wQve currents

~,~

Inner plexiforrn layer ([PL) (K + increase)

f

Slow pTIT current

~1~

Proxim¢l-Vifreous

FIG. 1. Schematic diagram of Mfiller cell currents in the toad retina. The M-wave current pathways are shown at the left of this schematic Mfiller cell, while the slow PIII current pathways are shown at the right. The M-wave has a current sink in the IPL, which is established in response to a light-evoked increase in [K+]o.The M-wave has current sources in both distal retina and at the vitreal endfoot of the Mfiller cell. Slow PIII has a current source in distal retina, which is established in response to a light-evoked decrease in [K+]o. Slow PHI has a current sink at the vitreal endfoot of the Mfiller cell. Additional details are provided in the text. the change in [K+]o, with a 1 mV change in V~ corresponding to a 0"17 mM change in [K+]o (Tucker, Wen and Oakley, 1991). There was no detectable interference with the electrode response to K÷ from any of the test chemicals at the concentrations used in these experiments.

Other Techniques All electrode voltages were recorded and processed as described recently (Tucker et al., 1991 ; Katz et al., 1991 ). Differences between individual digitized waveforms were obtained by point-by-point subtraction. To record the distal K÷ increase, which is the primary ionic source of the b-wave (Dick and Miller, 1978, 1985; Kline, Ripps and Dowling, 1978: Karwoski et al., 1985; Wen and Oakley, 1990a), the electrode was advanced into the retina in 2-#m steps while carefully observing the characteristic changes in waveform of the light-evoked change in V~. At the amplitude m a x i m u m for the distal K ÷ increase, the electrode tip was in the OPL (Wen and Oakley, 1990a). All illustrated waveforms are representative of responses observed consistently in at least three different retinas.

Light Stimulation In all experiments, the retina was stimulated repetitively (once every 8 0 - 9 0 sec) by lO-sec steps of

500 n m light, which produced 600 quanta absorbed per rod per second.

A Model of Mi~ller Cell Potassium Currents In order to interpret the experiments reported in this study, it is important to understand the relationship between [K+]o, Mfiller cell currents, and the ERG. This section provides a brief explanation of these relationships. When the dark-adapted retina is stimulated with steps of illumination that saturate the rod photoresponse, neuronal activity leads to at least three distinct changes in [K+]o: (1) a slow decrease in [K+]o in the receptor layer (Oakley and Green, 1976 ; Oakley, Flaming, and Brown, 1979); (2) a small, transient increase in [K+]o in the OPL (Dick and Miller, 1978, 1985; Kline et al., 1978; Wen and Oakley, 1990a); and (3) a larger, more prolonged increase in [K+]o in the IPL (Karwoski and Proenza, 1977; Steinberg et al., 1980). In response to these localized changes in [K+]o, K + currents flow across the Mfiller cell membrane, entering the Mfiller cells in regions of high [K+]o and leaving the Mfiller cells in regions of low [K+]o. Due to the non-uniform distribution of K+ conductance on the Mfiller cell m e m b r a n e (Newman, 1986, 1987), the flux of K + across the Mfiller cell membrane depends on the local K+ conductance, as well as the local change in [K+]o. The appearance of current sources and sinks (where K ÷ leaves and enters the Mfiller cells, respectively) on the Mfiller cell m e m b r a n e sets up

542

current pathways around the Mtiller cells, and the flow of current through the extracellular space produces extracellular voltages that can be recorded as components of the ERG (Faber, 1969; Miller and Dowling, 1970; Newman and Odette, 1984). The Mfiller cell K+ currents of importance to the present study are illustrated schematically in Fig. 1. The M-wave, which arises in response to the lightevoked increase in [K+]o in the IPL, has a current sink in the IPL and current sources located both in the inner retina (at the Mtiller cell endfoot) and in the outer retina (Frishman et al., 1991: Katz et al., 1991), as shown at the left of the schematic Mtiller cell in Fig. 1. This arrangement of current sources and sinks leads to opposing current loops in distal and proximal retina. The slow PIII component, which arises in response to the light-evoked decrease in [K+]o in the receptor layer, has a current sink at the Mtiller cell endfoot and a current source distributed over the distal regions of the cell that are exposed to the lightevoked decrease in [K+]o, as shown at the right of the schematic Mtifier cell in Fig. 1 (Faber, 1969; Newman and Odette, 1984). Since there is a light-evoked decrease in [K+]o in the OPL, there will be a slow PIII current source in the OPL. The K+ currents that flow around these Mfiller cell pathways are spatial buffering currents (Orkand et al., 1966), which are generated to maintain [K+]o homeostasis throughout the retina. The M-wave and slow PIII components of the ERG arise solely as a result of these spatial buffering currents.

3. Results

Spatial Buffering Driven by the M-Wave Currents We first chose to examine the contribution of spatial buffering to the light-evoked change in [K+]o in the OPL. As shown in Fig. 2(A), the fight-evoked change in [K+]o at this depth consisted of an initial transient increase in [K+]o, which generates the b-wave of the ERG (Miller and Dowling, 1970; Dick and Miller, 1978; Kline et al., 1978; Wen and Oakley, 1990a), followed by a slow decrease in [K+]o, which contributes to the generation of slow Pig (Witkovsky et al., 1975). Following the lO-sec period of illumination, [K+]o slowly recovered to the baseline. The light-evoked decrease in [K+]o was likely due to diffusion of K÷ from the OPL down its concentration gradient to the receptor layer, since there was a much larger decrease in [K+]o in the receptor layer (see below). In order to assess the contribution of spatial buffering to the lightevoked changes in [K+]o in the OPL, we superfused the retina with 200 FM Ba 2÷, a known blocker of Mfiller cell K+ conductance (Newman, 1989). We found that superfusion with 200 #M Ba 2÷ doubled the magnitude of the decrease in [K+]o by the end of the lO-sec period of illumination [Fig. 2(A)]. The difference between the waveforms recorded under control and Ba D÷conditions

B. O A K L E Y II ET AL.

is shown in Fig. 2(B). Following initial transients that were related to changes in the initial increase in [1(+1o, this difference waveform illustrates the Ba2+-induced change in the light-evoked decrease in [K+]o in OPL. We realized that the larger light-evoked decrease in [K+]o during Ba '~+ superfusion might have been due simply to the extracellular diffusion of K+ toward a larger light-evoked decrease in [K+]o in the receptor layer. We tested this possibility in the experiment illustrated in Fig. 3. The tip of the K+-ISM was positioned in the receptor layer, so that it recorded the maximal light-evoked decrease in [K+]o, and then the solution superfusing the retina was switched to one containing 200 FM Ba 2+. As shown in Fig. 3(A), the light-evoked decrease in [K+]o was more prolonged during superfusion with Ba 2+ than it was under control conditions. However, when the responses from Fig. 3(A) were superimposed [Fig. 3(B)], the amplitude of the light-evoked decrease in [K+]o at the end of the lO-sec period of illumination was seen to be exactly the same in Ba D+ as it was under control conditions. Superfusion with 200/ZM Ba D+was without significant effect upon the resting level of [K+]o in either the OPL or the receptor layer (data not shown). Therefore, the results illustrated in Fig. 3 rule out the possibility that the larger decrease in [K+]o in the OPL during superfusion with Ba ~+ (Fig. 2) was due simply to diffusion of K+ toward a lower level of [K+]o in the receptor layer. Since Ba 2. also would have blocked spatial buffering of [K+]o by Mfiller cells, the results from Figs 2 and 3 lead us to conclude that spatial buffering normally contributes an efflux of K+ into the OPL. This flux of K+ attenuates the large decrease in [K+]o that is observed in the absence of spatial buffering [Fig. 2(A), bottom]. We next examined the effects of Ba ~+ on the lightevoked changes in [K+]o in the IPL, as shown in Fig. 4. Under control conditions [Fig. 4(A), upper response], there were transient increases in [K+]o in the IPL, both at light onset and light offset. When 200 FM Ba 2+ was added to the superfusate, the light-evoked increase in [K÷]o became larger and more sustained, so that [K+]o continued to increase throughout the lO-sec period of illumination [Fig. 4(A), lower response]. The difference between the response waveforms recorded under control and Ba 2+ conditions is shown in Fig. 4(B). Following an initial transient, due to a change in the rate of increase in [K+]oin Ba ~+ solution, this difference waveform shows the component of the light-evoked increase in [K+]o in the IPL that normally is taken up by Mfiller cell spatial buffering. The greater light-evoked increase in [K+]o in the IPL in Ba 2÷ solution creates a larger gradient of [K+]o that would enhance the extracellular diffusion of K÷ from the IPL toward the OPL. One might reason, therefore, that the level of [K+]o in the OPL during illumination should not decrease to as low a level during superfusion with Ba 2+. Yet under this condition, where spatial buffering was greatly reduced, [K+]o in the OPL

K+ SPATIAL

BUFFERING

BY M U L L E R

CELLS

543

(A]

(B) Difference : Control - Be a+

Control

Ba2+

/

I

1.0 mV

I

I0 sec

LM _ _ [ ~ 1

Era. 2. Effects of Ba 2+ upon the light-evoked changes in [K+]oin the OPL. The tip of the K+-ISM was positioned in the OPL. In this and in all subsequent figures, the presentation of the lO-sec light stimulus is indicated by the upward deflection of the light monitor waveform (labeled LM), A, The control Vx response is shown at the top. A V~ response recorded during superfusion with 200 ffM Ba 2* solution is shown at the bottom. B, Difference waveform. The illustrated response is the difference between the two VK waveforms shown in (A) (control-Ba2+).

(A)

[B]

Control Superimposed

Ba a+

mV

I

l

1

I LM - j

[

Fro. 3. Effects of Ba 2+ upon the light-evoked changes in [K+]oin the receptor layer. The tip of the K+-ISM was positioned in the receptor layer, at the retinal depth where the fight-evoked decrease in [K+]owas of maximal amplitude. A, The control V~ response is shown at the top. A, VK response recorded during superfusion with 200 #M Ba 2+ solution is shown at the bottom. B, The two VK waveforms from (A) are shown superimposed.

actually decreased to an even lower level (Fig. 2). Thus, it appears that Mfiller cell spatial buffering currents play a greater role in transporting K + from IPL to OPL t h a n does extracellular diffusion. The results illustrated in Figs 2 - 4 are consistent with Ba '+ blocking a spatial buffering flux of K + t h r o u g h the Mfiller cells: in the IPL, K* is not able to enter the Mfiller cells, so it accumulates to a greater level; in the OPL, K* does not leave the Mfiller cells, so it falls to a lower level as K + diffuses toward the receptor layer. Thus, Mfiller cells buffer the lightevoked increase in [K*]o in the IPL by transporting K*

from IPL to OPL. At the same time, this m o v e m e n t of K + serves to buffer the light-evoked decrease in {K+]oin OPL. We next performed an additional test of our hypothesis that K + released by n e u r o n s in the IPL is transported t h r o u g h the Mfiller cells to the OPL. We predicted that if the n e u r o n a l release of K + in the IPL could be e n h a n c e d by some means, more K + should be transported to the OPL via the Mfiller cell pathway. This idea was tested in Figs 5 and 6 by using 100 #M picrotoxin (PTX) to increase the amplitude of the lightevoked increase in [K÷]o in the IPL. One effect of PTX

544

B. O A K L E Y II ET AL.

(B) Difference: Bu z÷_ Control Control

I

J

I

5 mV IO sec

LM " - [ ~

FIG. 4. Effects of Ba ~+ upon the light-evoked changes in [K+]oin the IPL. The tip of the K÷-ISM was positioned in the IPL, at the depth where the light-evoked increase in [K+]owas of maximal amplitude. A, The control VK response is shown at the top. A, V~ response recorded during superfusion with 200 #M Ba2÷ solution is shown at the bottom. B, Difference waveform. The illustrated response is the difference between the two V~ waveforms shown in (A) (control-Ba2+). (A)

(B) Differences: PTX - Control

Control

PTX

PTX 4-Bo a+

i

(PTX + Baa+) -PTX

I I

LM

--J

I

I

_J

I0 sec

I

PIG. 5. Effects of picrotoxin and Ba ~+ upon the light-evoked changes in [K']o in the IPL. The tip of the K+-ISM was positioned in the IPL, at the depth where the light-evoked increase in [K+]owas of maximal amplitude. A, The control VX response is shown at the top, a V~ response recorded during superfusion with 100 #M PTX solution is shown in the middle, and a VK response recorded during superfusion with a solution containing 100 #M PTX and 200 #M Ba ~+ is shown at the bottom. B, Difference waveforms. The upper waveform is the difference between the two V~ waveforms shown in the upper part of (A) (PTX-control), and the lower waveform is the difference between the two V~ waveforms shown in the lower part of (A) [(PTX + Ba 2+)- PTX].

is to block GABAa-mediated inhibition between amacrine cells, thereby e n h a n c i n g amacrine cell responses to diffuse illumination, and in turn, the light-evoked increase in [K+]o in the IPL (Frishman et al., 1991 ; Katz et al., 1991). As s h o w n in Fig. 5(A) (upper and middle response waveforms), the addition of PTX more t h a n doubled the amplitude of the lightevoked increase in [K+]o in the IPL. The difference between the control and PTX waveforms, s h o w n at the top of l?ig. 5(B) illustrates the a m o u n t by w h i c h PTX e n h a n c e d the light-evoked increase in [K+]o over

its control amplitude. In the presence of PTX, there was a larger gradient for the extracellular diffusion of K + from the IPL to the OPL, as well as more K + available to be transported from the IPL to the OPL via the Mfiller cell spatial buffering pathway. W h e n 200 ,aM Ba 2+ was added to the solution containing PTX, the light-evoked increase in [K+]o in the IPL became even larger, as s h o w n by the lower waveform in Fig. 5(A). The increased level of [K+]o in the IPL at the end of the lO-sec period of illumination during Ba 2+ superfusion would have further increased

K + S P A T I A L B U F F E R I N G BY M U L L E R C E L L S

545

(A)

(B) Differences: PTX- Control

Control

PTX

I PTX - (PTX + Bo 2+)

PTX + BG2+

v

LM

-J

I

FI6. 6. Effects of picrotoxin and Ba 2+upon the light-evoked changes in [K+]oin the OPL. The tip of the K+-ISMwas positioned in the OPL. A, The control V~ response is shown at the top, a VK response recorded during superfusion with 100 #M PTX solution is shown in the middle, and a V~ response recorded during superfusion with a solution containing 100 ~M PTX and 200 tiM Ba 2+ is shown at the bottom. B, Difference waveforms. The upper waveform is the difference between the two VX waveforms shown in the upper part of (A) (PTX--control), and the lower waveform is the difference between the two VK waveforms shown in the lower part of (A) [PTX-(PTX + Ba2+)]. the gradient for extracellular diffusion of K + from the IPL to the OPL. However, Ba 2+ would also have blocked any spatial buffering flux of K + through the Mfiller cells from the IPL to the OPL. We next determined the effects of the experimental treatments used in Fig. 5 while recording [K+]o in OPL, as shown in Fig. 6. The addition of 100 ~M PTX to the control solution caused the slow decrease in [K+]o to be replaced by a slow increase in [K+]o, as shown by the upper and middle waveforms in Fig. 6(A). The difference between these two waveforms, shown at the top of Fig. 6(B), indicates the extra amount of K + that was transported to the OPL as a result of superfusion with PTX. This additional flux of K + produced a lightevoked increase in [K*]o, which more than offset the light-evoked decrease in [K+]o in the OPL. The flux of K + into the OPL was most likely due to increased K+ being transported from the IPL, either by extracellular diffusion or by Mfiller cell spatial buffering. The addition of 200 #M Ba 2+ to the PTX solution caused a significant change in the light-evoked change in [K+]o in the OPL, as shown by the lower response waveform in Fig. 6(A). In the PTX/Ba 2+ solution, there was a large, light-evoked decrease in [K+]o, which was even larger in magnitude than the one observed under control conditions. The difference between the responses in picrotoxin and PTX/Ba 2+ is shown as the lower response waveform in Fig. 6(B), and this difference waveform was a measure of the amount of K + that was no longer being added to the OPL in the presence of Ba 2+. Since Ba 2+ actually increased the gradient for extracellular diffusion of K + from the IPL to the OPL,

the observation that less K+ appeared in the OPL during superfusion with Ba ~+ (Fig. 6) virtually eliminates any significant role for extracellular diffusion in the observed changes in [K+]o. However, this result is consistent with the hypothesis that K + was transported from the IPL to the OPL via a Mfiller cell pathway. By blocking this transport with Ba 2+, [K+]o accumulated to a greater level in the IPL, but fell to a lower level in the OPL. Thus, in the presence of PTX alone, the greatly enhanced level of [K+]o in the IPL during a lO-sec period of illumination increased the flux of K+ through the Mfiller cells, and produced a net increase in [K+]o in the OPL [Fig. 6(A), middle waveform].

Spatial Buffering in OPL due to Slow Pill Currents In the dark-adapted retina, the slow PIII currents that flow around the Mfiller cells (Fig. 1, right) presumably also are spatial buffering currents. Since the area of the Mfiller cell m e m b r a n e within the OPL is normally exposed to a light-evoked decrease in [K+]o [Fig. 2(A), top], this m e m b r a n e area should contribute a slow PIII current source (an efflux of K +) at this retinal depth. To test this idea, we performed the experiment illustrated in Fig. 7. After recording the light-evoked change in [K+]o in the OPL under control conditions [Fig. 7(A), upper response waveform], the solution superfusing the retina was switched to one that contained 100 ,ttM 2-amino-4-phosphonobutyric acid (APB). This treatment blocks the light-evoked activity of retinal neurons in the ON-pathway proxi-

546

B.OAKLEYII ETAL (B)

(A)

Control~ APB

Diferences: Control-APB

~,,~,~~ l

I

+ BQ2+

I I APB- (ABP + BG z+)

APB

I

I P mV

I

I LM _J'

I0sec

I

j - -

Fla. 7. Effects of APB and Ba"+ upon the light-evoked changes in [K+]oin the OPL. The tip of the K+-ISM was positioned in the OPL. A, The control V• response is shown at the top, a V~ response recorded during superfusion with 100 #M APB solution is shown in the middle, and a Vx response recorded during superfusion with a solution containing 100 #M APB and 200 #M Ba2+ is shown at the bottom. B, Difference waveforms. The upper waveform is the difference between the two VK waveforms shown in the upper part of (A) (control- APB), and the lower waveform is the difference between the two V~ waveforms shown in the lower part of (A) [APB-(APB + Ba2+)].

real to the receptors (Slaughter and Miller, 1981, 1985). Under this condition, the light-evoked change in [K+]o in the OPL consisted solely of a decrease in [K+]o during the lO-sec period of illumination [Fig. 7(A), middle response waveform]. This decrease in [K+]o was likely due to diffusion of K+ from the OPL toward the lower level of [K+]o in the receptor layer. Treatment with APB would have eliminated the lightevoked release of K + by ON-neurons in the IPL, as well as in the OPL, and thereby would have blocked the efflux of K+ from Mtiller cells in the OPL due to M-wave spatial buffering currents. However, the slow PIII currents still were flowing around the Mfifler ceils under this condition, and the light-evoked decrease in [K+]o in the OPL presumably still was being modified by a spatial buffering efilux of K+ from the M/iller cells (the slow PIII current source) at this depth. To assess the contribution of spatial buffering due to slow PIII, the solution superfusing the retina was then switched to one that contained 2 0 0 # M Ba 2+ in addition to APB. This solution caused the light-evoked decrease in [K+]o in the OPL to become larger in amplitude, as shown by the lower response waveform in Fig. 7(A). The difference between the responses in APB and APB/Ba 2+ is shown as the lower waveform in Fig. 7(B); this difference waveform was a measure of the K+ flux into the OPL that was blocked by the addition of Ba 2+, The effect of Ba ~÷ was not due to a change in the light-evoked decrease in [K+]o in the receptor layer, since Ba 2+ was without effect upon [K+]o in the receptor layer in the presence of APB (data not shown). Since APB already had blocked the neuronal release of K+ into the IPL during the lO-sec

period of illumination (Stockton and Slaughter, 1989; Xu et al., 1991), the change in the response amplitude in the OPL could only have been due to elimination of the spatial buffering efflux of K + into the OPL from the Miiller cells. Thus, it appears that the amplitude of the light-evoked decrease in [K+]o in the OPL also is attenuated by the spatial buffering efflux of K+ from the slow PIII current source on the Mfiller cell membrane at that retinal depth.

Spatial Buffering at the Vitreal Border of the Retina We were also interested in the light-evoked changes in [K÷]o at the vitreal border of the retina, since both the M-wave and slow PIlI current pathways involve current loops that include the proximal endfoot of the Mtiller cell (Fig. 1). The proximal loop of the M-wave involves a current source at the Mtiller cell endfoot, while the slow PIII current loop involves a current sink. The K+ efflux associated with this current source increases [K+]o in the extracellular space surrounding the Mtiller cell endfoot in the light-adapted mudpuppy retina, and this efflux can be blocked by Ba 2+ (Karwoski et al., 1989). In addition, the K + uptake associated with the slow P I g current sink at the Mfiller cell endfoot should decrease [K+]o. We examined the light-evoked changes in [K+]o in the layer of solution next to the vitreal border of the retina, as shown in Fig. 8(A). In this experiment, we also measured the transretinal (vitreal) ERG through the reference barrel of the K+-ISM, as shown in Fig. 8(B). The light-evoked change in [K+]o under control conditions consisted of a small, transient increase in

K+ SPATIAL

BUFFERING

BY M O L L E R

CELLS

547

(A) l/K at vitreol border

(B) ERG

Control

APB + KYN

z'-'---

0.5(3 mV VK Difference Control(APB + KYN) APB + KYN + Bo 2+

-J

[

LM

-"k.

J

I

FIG. 8. Effects of APB, KYN, and Ba 2+upon the light-evoked changes in [K+]oat the vitreal border of the retina and upon the ERG. The tip of the K+-ISMwas positioned at the vitreal border of the retina. A, The control V~ response is shown at the top, and just below the control response is a V~ response recorded during superfusion with a solution containing 100 FM APB and 2 mM KYN. The difference between these two V~ waveforms [control- (APB + KYN)] is shown as the third waveform from the top. The last V~ waveform was recorded during superfusion with a solution containing 100/~M APB, 2 mM KYN, and 200 FM Ba~÷. B, This part of the figure shows the ERG waveforms recorded simultaneously with the V~ waveforms in (A). The ERG recorded in this manner is equivalent to the vitreal ERG.

[K+]o, which recovered to baseline by the end of the lO-sec period of illumination [Fig. 8(A)]. The retina then was superfused with a solution containing IO0#M APB and 2 mM KYN to abolish all lightevoked activity in both the ON- and OFF-pathways proximal to the receptors (Slaughter and Miller, 1981 ; Coleman, Massey and Miller, 1986). This treatment had a pronounced effect on the light-evoked change in [K+]o, such that the response in APB/KYN consisted solely of a large decrease in [K+]o. The difference between the waveforms under control and APB/KYN conditions is shown in Fig. 8(A) (waveform labeled 'Difference'). This [K+]o difference waveform shows that APB/KYN eliminated a large K+ efflux at the M/iller cell endfoot. Thus, the light-evoked change in [K+]o at the vitreal border under control conditions is seen to be comprised of two opposing processes: an increase in [K+]o (which is eliminated by APB/KYN); and a somewhat slower decrease in [K+]o (which is spared by APB/KYN). This light-evoked increase in [K+]o presumably is due to Mfiller cell spatial buffering currents driven by the increase in [K+]o in the IPL (Karwoski et al., 1989), and it therefore represents the M-wave current source. This light-evoked decrease in [K+]o at the vitreal border of the retina that was observed during superfusion with the APB/KYN solution was due to a flux of K+ away from the recording site, and this flux likely was driven by the light-evoked decrease in [K+]o in the receptor layer. This flux could have been due to extracellular diffusion a n d / o r to M/iller cell spatial buffering. To differentiate between these two possi-

bilities, we subsequently added 200 #M Ba ~+ to the APB/KYN solution and we found that Ba ~+ eliminated essentially all of the light-evoked decrease in [K+]o at the vitreal border [Fig. 8(A), bottom]. This result indicates that the uptake of K+ is driven by M/iller cell spatial buffering currents, and not by extracellular diffusion. This uptake of K+ presumably represents the slow PIII current sink at the M/iller cell endfoot. The vitreal ERG is shown in Fig. 8(B). The control ERG was dominated by the positive b-wave and the negative slow PIII components [Fig. 8(B), top]. As expected, APB/KYN eliminated all the ERG components that originate post-synoptic to the receptors (Katz et al., 1991, 1992 ; Xu et al., 1991), leaving only the extracellular receptor potential and slow PIII [Fig. 8(B), second waveform from top]. The difference between the control and APB/KYN waveforms [Fig. 8(B), difference waveform] illustrates the ERG components that were eliminated by APB/KYN: the bwave, the DC component, and the M-wave. The subsequent superfusion with Ba 2+ eliminated slow Pill, leaving only the receptor potential [Fig. 8(B), bottom]. 4. Discussion Large shifts in [K÷]o in the plexiform layers, if not buffered, might lead to changes in neuronal function (Newman, 1985a; Karwoski et al., 1989). In the present experiments, we have shown that K÷ is transported from the IPL not only to the vitreous, as shown previously by Karwoski et al. (1989), but also

548

to the OPL, where the flux of K ÷ out of the M/iller cell buffers the decrease in [K+]o. A similar transport of K+ from the IPL to the distal retina has been observed in cat retina by Frishman and Steinberg (1989b) and by Frishman et al. (1991). In cat, the large K + conductance in the most distal portion of the Mtiller cells (Newman, 1987) causes a large efflux of K + to be directed toward the subretinal space, rather than toward the OPL, as in toad. In addition, we have shown that Mfiller cell spatial buffering attenuates the amplitude of the light, evoked decrease in [K+]o in the OPL by transporting K ÷ from vitreous to OPL. Thus, spatial buffering not only prevents the accumulation of [K+]o in the IPL (Karwoski et al., 1989; Frishman and Steinberg, 1989b), but it also prevents the fall of [K+]o in the OPL. Our measurements have also shown that two opposing processes affect the amplitude of the changes in [K+]o at the vitreal border of the dark-adapted toad retina. At the vitreal border of the retina, the slow PIII current sink offsets the M-wave current source, so that the net change in [K+]o at the vitreal border is m u c h smaller t h a n it would be if only one of these two current loops was active. Instead of having the vitreous act as a large source or sink for K + (Karwoski et al., 1989), K+ appears to remain largely within the retina, moving between the inner and outer retina.

Effectiveness of Spatial Buffering The flux of K + due to spatial buffering presumably has less effect upon [K÷]o in the receptor layer, where the extracellular volume is relatively large, than in the plexiform layers, where the extracellular volume is relatively small (Newman and Odette, 1984; Karwoski, Frambach and Proenza, 1985). Moreover, in our isolated retina preparation, where the retinal pigment epithelium (RPE) is removed and the extracellular volume in the receptor layer is artificially large (Oakley, 1983), any spatial buffering flux of K ÷ into the receptor layer will have even less effect upon [K+]o. The larger light-evoked decrease in [K+]o in the receptor layer that is observed in vivo will generate larger slow PIII currents, which in turn will transport more K+ to the OPL via spatial buffering. The subsequent diffusion of this K÷ from the OPL to the receptor layer also will help to buffer the light-evoked decrease in [K+]o in the receptor layer. Since the fluxes of K÷ will be larger in vivo, we expect that the contribution of the slow PIII currents to spatial buffering of [K÷]o in the OPL and the receptor layer will be even larger in vivo that we have observed in vitro.

Comparison to Spatial Buffering in Other Neuronal Tissues The spatial buffering of [K+]o by glial ceils has been characterized in a wide variety of nervous tissues, including mudpuppy optic nerve (Orkand et al., 1967),

B. O A K L E Y II ET AL.

rat neocortex and cerebellar cortex (Gardner-Medwin and Nicholson, 1983), cat cortex (Dietzel et al., 1980), and the retina of the honeybee drone (GardnerMedwin, Coles and Tsacopoulos, 1981; Coles and Orkand, 1983). In all of these tissues, the flux of K÷ through the glial cells has been shown to be significantly greater than the flux of K÷ due to extraceflular diffusion. The effects of spatial buffering upon stimulus-evoked increases in [K÷]o have been characterized in these previous studies. Due to the unique hyperpolarizing responses of photoreceptors in the vertebrate retina, we believe that our measurements in toad retina are the first to illustrate that glial cell spatial buffering can affect stimulus-evoked decreases in [K+]o.

Effects of Extracellular Barium on Glial Potassium Fluxes Our present experiments were possible only because Ba 2÷ blocks the K÷ fluxes across the Mtiller cell membranes, without significantly affecting the lightevoked K+ fluxes across the neuronal membranes (see below). Low concentrations of Ba 2÷ ( 2 0 - 2 0 0 #M) block K÷ conductance in Mtiller cells (Newman, 1989), but apparently these concentrations of Ba 2÷ do not block neuronal K÷ conductances (grishman and Steinberg, 1989b; Karwoski et al., 1989; Katz et al., 1991). This difference might be due to differential sensitivity to Ba 2÷ of the K+ conductances in neurons and M/iller cells. Another mechanism that could account for this difference is as follows: in all retinal neurons, the net light-evoked flux of K÷ across the cell membrane is in the outward direction, and this efflux must be balanced over time by uptake of K÷ by the neurons' Na+/K ÷ pumps. Note that even the light-evoked change in the flux of K÷ across the rod photoreceptor membranes is a reduction in the efflux of K + (Oakley et al., 1979). In the Mfiller cells, however, the K+ fluxes across the cell membranes must be directed in both the outward and inward directions, since the spatial buffering current sources and sinks on the Mtiller cell membranes are established passively. Thus, an obvious difference between neuronal and glial K ÷ fluxes is that K÷ must passively enter the glial cells, but not necessarily the neurons. If extracellular Ba 2+ blocks the passive influx of K+ through the membrane channels, but does not affect the efflux of K+ through these same channels, then this mechanism alone would cause the glialmediated spatial buffering of K ÷ to cease in the presence of Ba 2÷. Barium would not affect neuronal activity, since K÷ still would be able to leave through these channels. However, since Ba 2. would block the flux of K ÷ into Mtiller cells, the Mtiller cells would no longer be able to take up K+ in areas of relatively high [K÷]o, and spatial buffering would cease. Thus, we propose that Ba 2÷ blocks the passive influx of K÷ through m e m b r a n e channels. We believe that this mechanism could be responsible for the differential

K ÷ S P A T I A L BUFFERING BY M U L L E R CELLS

effects of Ba 2+ upon the neuronal and glial K+ fluxes, although additional work clearly will be needed to establish the role of this mechanism with certainty. Effects of Barium on Rod Mediated Potassium Fluxes Superfusion of the toad retina with 200 ,aM Ba 2+ prolongs the time over which [K+]o in the receptor layer continues to decrease (Fig. 3). Barium is known to block a voltage-sensitive K÷ conductance in rods, which has the effect of increasing the amplitude and duration of rod photovoltages (Brown and Flaming, 1978; Fain et al., 1980). Since the rod photoresponse lasts longer, [K+]o decreases for a longer period of time (Oakley et al., 1979). However, an increase in the amplitude of the rod photoresponse also would be expected to increase the amplitude of the light-evoked decrease in [K+]o (Oakley et al., 1979). Since the lightevoked decrease i n [K+]o is essentially unaffected by Ba 2÷ during a lO-sec period of illumination (Fig. 3), it appears that the effect of the increased voltage response is offset by the decreased K÷ conductance in the rod membrane. Alternatively, if Ba 2. blocks a spatial buffering flux of K+ from the distal end of the Mfiller cell into the receptor layer, then the observation that the amplitude of the light-evoked decrease in [K+]o is unaffacted by Ba 2+ (Fig. 3) may mean that the component of this response mediated by the rods actually becomes slightly smaller in Ba 2+. However, we are unable to distinguish between changes in these two mechanisms using our present techniques. Implications for Measurement of Retinal Extracellular Potassium The light-evoked change in [K+]othat is measured at any retinal depth with a K+-ISM is now seen to be a complex response, consisting of both neuronal and glial components. In the unstirred layer of fluid next to the vitreal border of the toad retina, almost all of the light-evoked changes in [K+]o appear to be due to K+ fluxes through the Mfiller cells, as observed previously in flog (Karwoski et al., 1989). In the [PL, the fightevoked increase in [K+]o is due to the neuronal release of K+, but this response is modified to a large extent by the M-wave current sink on the Mfiller cell membrane, which removes extracellular K+ and prevents [K+]o from increasing to extreme levels (Figs 4 and 5; see also Karwoski et al., 1989; Frishman and Steinberg, 1989b). In the OPL, the fight-evoked changes in [K+]o appear to be the most complex, since K+ is released by the ON-bipolar cells (Dick and Miller, 1978) and K+ diffuses out of the OPL toward the large, fight-evoked decrease in [K+]o in the receptor layer. In addition, both the M-wave and slow PIll current sources on the Mfiller cell membrane add K+ to the OPL, and these spatial buffering components appear comparable in magnitude to the neuronal components. Under some conditions, such as superfusion with picrotoxin, the

549

spatial buffering flux of K + can dominate the lightevoked change in [K+]o in the OPL (Fig. 6). Overall, it would be very misleading to ascribe the measured •changes in retinal [K+]o solely to neuronal release and to extracellular diffusion. Future experiments involving the measurement of retinal [K+]o must carefully determine the relative neuronal and glial contributions to each measurement.

Mi~ller Cell Spatial Buffering of K ÷ and the Electroretinogram The extracellular concentration of H ÷ is buffered by the bicarbonate/CQ system, and the extracellular concentration of Ca 2+ can be buffered by protein chelation. There are no analogous mechanisms to regulate [K~]o; it must be buffered by the net movement of K+ from regions where [K+]o is high to regions where [K+]o is low. Regulation of [K+]o is especially critical in synaptic regions, where neurons have a greater proportion of K+ conductance and therefore are more susceptible to changes in [K+]o (Kuffler et al., 1984). In retina, the redistribution of K+ is accomplished by Mfifler cell spatial buffering. As a direct result of this flux of K+, transretinal potentials are generated, which comprise much of the ERG (bwave, DC component, M-wave, and slow PHI). Thus, spatial buffering of K+ and the ERG are intimately related: the ERG results from the retina's homeostatic mechanism for regulation of [K+]o. It is interesting to note that the retinal pigment epithelium, which generates the ERG c-wave, also is thought to have a K+ spatial buffering function (Immel and Steinberg, 1986).

Acknowledgements We thank Drs Laura J. Frishman and Chester J. Karwoski for helpful discussions. This work was supported by NIH grants EY04364, GM07143, Rtl 7030, and by a grant from the Warren and Clara Cole Foundation. References Brown, K. T. and Flaming, D. G. (1978). Opposing effects of calcium and barium in vertebrate rod photoreceptors. Proc. Natl. Acad. Sci. U.S.A. 75, 1587-90. Coleman, P.A., Massey, S.C. and Miller, R.F. (1986). Kynurenic acid distinguishes kainate and quisqualate receptors in the vertebrate retina. Brain Res. 381, 172-5. Coles, J.A. and Orkand, R.K. (1983). Modification of potassium movement through the retina of the drone (Apis mellifera) by glial uptake. ]. Physiol. 340, 157-74. Dick, E. and Miller, R.F. (1978). Light-evoked potassium activity in mudpuppy retina: its relationship to the bwave of the electroretinogram. Brain Res. 154, 388-94. Dick, E. and Miller, R.F. (1985). Extracellular K÷ activity changes related to electroretinogram components. J. Cen. Physiol. 85, 885-909, Dietzel, I., Heinemann, U., Hofmeier, G. and Lux, H.D. (1980). Transient changes in the size of the extra-

550

cellular space in the sensorimotor cortex of cats in relation to stimulus-induced changes in potassium concentration. Exp. Brain Res. 40, 432-9. Faber, D. S. (1969). Analysis of slow transretinal potentials in response to light. Ph.D. dissertation. State University of New York, Buffalo, New York. Fain, G. L., Gerschenfeld, H. M. and Quandt, F. N. (1980). Calcium spikes in rods. J. Physiol. 3 0 3 , 4 9 5 - 5 1 3 . Frishman, L.J. and Steinberg, R.H. (1989a). Intraretinal analysis of the threshold dark-adapted ERG of cat retina. ]. Neurophysiol. 61, 1221-32. Frishman, L.J. and Steinberg, R. H. (1989b). Light-evoked increases in [K+]o in proximal portion of the darkadapted cat retina. I. Neurophysiol. 61, 1233-43. Frishman, L.J., Yamamoto, F., Bogucka, J. and Steinberg, R. H. (1991). Origins of the M-wave in light-adapted cat retina. Invest. Ophthalmol. Vis. Sci. 32 (Suppl.), 927. Gardner-Medwin, A.R., Coles, J.A. and Tsacopoulos, M. (1981). Clearance of extracellular potassium:evidence for spatial buffering by glial cells in the retina of the drone. Brain Res. 2 0 9 , 4 5 2 - 7 . Gardner-Medwin, A. R. and Nicholson, C. (1983). Changes of extraceflular potassium activity induced by electric current through brain tissue in the rat. ]. Physiol. (London) 335, 375-92. Immel, J. and Steinberg, R. H. (1986). Spatial buffering ofK + by the retinal pigment epithelium in frog. J. Neurosci. 6, 3197-204. Karwoski, C. J., Frambach, D. A. and Proenza, L. M. (1985). Laminar profile of resistivity in frog retina. ]. Neurophysiol. 54, 1607-19. Karwoski, C. J., Lu, H.-K. and Newman, E. A. (1989). Spatial buffering of light-evoked potassium increases by retinal Mfiller (glial) cells. Science 244, 578-80. Karwoski, C. ]., Newman, E. A., Shimazaki, H. and Proenza. L. M. (1985). Light-evoked increases in extracellular K+ in the plexiform layers of amphibian retinas. J. Gen. Physiol. 86, 189-213. Karwoski, C.J. and Proenza, L.M. (1977). Relationship between MfiUer cell responses, a local transretinal potential, and potassium flux. J. NeurophysioL 40, 244-59. Katz, B.J., Wen, R., Zheng, J., Xu, Z. and Oakley, B. II. (1991). M-wave of the toad electroretinogram. J. Neurophysiol. 66, 1927-40. Katz, B.J., Zheng, J., Xu, Z. and Oakley, B. II. (1992). PII component of the toad electroretinogram. J. Neurophysiol 68, 333-41. Kline, g. P., Ripps, H. and Dowling, J. E. (1978). Generation of b-wave currents in the skate retina. Proe. Natl. Acad. Sci. U.S.A. 75, 5727-31. Kuffler, S. W., Nichols, J. G. and Martin, R. A. (1984). From Neuron to Brain 2nd edn. Pp. 358-9. Sinauer Associates, Inc. : Sunderland, MA, U.S.A. MiIler, R.F. and Dowling, J,E. (1970). IntraceIlular responses of the Mfiller (glial) cells of mudpuppy retina: their relation to b-wave of the electroretinogram. ]. Neurophysiol. 33, 323-41. Newman, E.A. (1985a). Regulation of extracellular potassium by glial cells in the retina. Trends NeuroscL 8, 156-9. Newman, E.A. (1985b). Membrane physiology of retinal glial (Mfiller) cells. ]. Neurosci. 5, 2225-39. Newman, E.A. (1986). Regional specialization of the

B. O A K L E Y II ET AL.

membrane of retinal glial cells and its importance to K+ spatial buffering. Ann. N.Y. Acad, Sci. 4 8 1 , 2 7 3 - 8 6 . Newman, E.A. (1987). Distribution of potassium conductance in mammalian Miiller (glial) cells: a comparative study. J. Neurosci. 7, 2423-32. Newman, E.A. (1989). Potassium conductance block by barium in amphibian Mfiller cells. Brain Res. 498, 4326-30. Newman, E. A. and Odette, L. L. (1984). Model of electroretinogram b-wave generation: a test of the K+ hypothesis. I. Neurophysiol. 51, 164-82. Oakley, B. II. (1983). Effects of maintained illumination upon [K+]oin the subretinal space of the isolated retina of the toad. Vision Res. 23, 1325-37. Oakley, B. II, Flaming, D. G. and Brown, K. T. (1979). Effects of the rod receptor potential upon retinal extracellular potassium concentration. ]. Gen. Physiol. 74, 713-37. Oakley, B. II and Green, D. G. (1976). Correlation of lightinduced changes in retinal extracellular potassium concentration with c-wave of the electroretinogram. ]. Neurophysiol. 39, 1117-33. Oakley, B. II, Katz, B. J., Xu, Z. and Zheng, J. (1991). Mfiller cell spatial buffering currents transport potassium to distal retina. Invest. Ophthalmol. Vis. Sci. 32 (Suppl.), 25. Orkand, R. K., Nicholls, J. G. and Kuffler, S. W. (1966). Effect of nerve impulses on the membrane potential of glial cells in the central nervous system of amphibia. ]. Neurophysiol. 29, 788-806. Sieving, P. A., Frishman, L. J. and Steinberg, R. H. (1986a). M-wave of proximal retina in cat. ]. Neurophysiol. 56, 1039-48. Sieving, P. A., Frishman, L. J. and Steinberg, R. H. (1986b). Scotopic threshold response of proximal retina in cat. ]. Neurophysiol. 56, 1049-61. Slaughter, M.M. and Miller, R.F. (1981). 2-amino-4phosphonobutyric acid: a new pharmacological tool for retinal research. Science 211. 182-5. Slaughter, M. M. and Miller, R. F. (1985). Characterization of an extended glutamate receptor of the ON bipolar neuron in the vertebrate retina. ]. Neurosci. 5,224-33. Steinberg, R.H., Oakley, B. II and Niemeyer, G. (1980). Light-evoked changes in [K+]oin retina of intact cat eye. ]. Neurophysiol. 44, 897-921. Stockton, R. A. and Slaughter, M. M. (1989). B-wave of the electroretinogram: a reflection of ON bipolar cell activity. J. Gen. Physiol. 93, 101-22. Tucker, J., Wen, R. and Oakley, B. II. (1991). A deconvolution technique for improved estimation of rapid changes in ion concentration recorded with ion-selective microelectrodes. IEEE Trans. Biomed. Eng. 38, 156-60. Wen, R. and Oakley, B. II. (1990a). K+-evoked Mfiller cell depolarization generates b-wave of electroretinogram in toad retina. Proc. Natl. Acad. Sci. U.S.A. 87, 2117-21. Wen, R. and Oakley, B. II. (1990b). Ion-selective microelectrodes suitable for recording rapid changes in extracellular ion concentration. ]. Neurosci. Methods 31, 207-13. Witkovsky, P., Dudek, F. E. and Ripps, H. (1975). Slow PIII component of the carp electroretinogram. ]. Gen. Physiol. 65, 119-34. Xn, X., Xu, J., Huang, B., Livsey, C. and Karwoski, C.J. (1991). Comparison of pharmacological agents (aspartate vs. aminophosphonobutyric plus kynurenie acids) to block synaptic transmission from retinal photoreceptors in frog. Exp. Eye Res. 52, 691-8.