Cells of the Rat Retina: Effects of K+, Veratridine, and Ethylenediamine’

Home , Latrotoxin

0270.6474/83/0312-2494$02,00/O The Journal of Neuroscience Copyright 0 Society for Neuroscience Vol. 3, No. 12, pp. 2494-2503 Printed in U.S.A. December 1983

RELEASE OF [3H]y-AMINOBUTYRIC ACID FROM GLIAL (MijLLER) CELLS OF THE RAT RETINA: EFFECTS OF K+, VERATRIDINE, AND ETHYLENEDIAMINE’

P. VIJAY SARTHY

Department of Ophthalmology, University of Washington School of Medicine, Seattle, Washington 98195

Received September 28, 1982; Revised July 5, 1983; Accepted July 6, 1983

Abstract In several neural systems, glial cells appear to take up and release y-aminobutyric acid (GABA) upon depolarization. We have studied the release of [3H]GABA from Miiller (glial) cells in the rat retina by a double isotope-labeling technique in which Miiller cells are preloaded with “H-GABA while a population of neurons is prelabeled with [14C]glycine. By autoradiography, we have confirmed that [3H]GABA is taken up by the radially oriented Miiller cells, whereas [“Hlglycine is accumulated by a subset of amacrine cells (neurons). Using the double-labeling procedure, we have examined the effects of two depolarizing agents, high KC and veratridine, and the GABA mimetic, ethylenediamine, on transmitter release from glial cells and neurons simultaneously. We found the following. (1) Depolarization with 56 mM K+ released both [“HIGABA and [14C]glycine. About 70 to 80% of this release was blocked in Ca2+-free medium. (2) Veratridine (10 pM) also released both of the transmitters. This release was strongly inhibited by 100 nM tetrodotoxin or 1 mM procaine. Under Ca’+-free conditions, < 20% isotope release was observed. (3) Ethylenediamine released [“HIGABA readily, whereas little [14C]glycine release was observed. Removal of Ca*’ had no significant effect on transmitter release. Furthermore, in Na+-free medium ethylenediamine failed to induce [3H] GABA or [ 14C]glycine release. These results suggest that high K’ and veratridine release [“HI GABA from Miiller cells by a Ca’+-dependent process. Ethylenediamine, on the other hand, appears to induce [3H]GABA release by a Ca’+- independent, carrier-mediated exchange mechanism.

Like neurons, glial cells appear to take up and release striatal slices by a Ca2+-independent mechanism (Lloyd y-aminobutyric acid (GABA) upon depolarization (Va- et al., 1982). The effect of this chemical on GABA release ron and Somjen, 1979). This observation has led to the from retina has not been reported. question as to whether glial cells and neurons release the We have examined release of GABA from glial cells in transmitter by analogous mechanisms. In several neural the vertebrate retina. In the rat retina, [3H]GABA is systems, glia have been reported to release preloaded taken up by a sodium-dependent, high affinity uptake GABA upon stimulation with high K+ or veratridine system. Autoradiographic localization shows that the (Bowery and Brown, 1972; Voaden and Starr, 1972; sites of uptake are the glial (Miiller) cells (Neal and Minchin and Iversen, 1974; Roberts, 1974; Schrier and Iverson, 1972). In an earlier study, [“HIGABA release Thompson, 1974; Minchin, 1975; Hertz, 1977; Sellstrom was not detected upon exposure of preloaded rat retina and Hamberger, 1977; Bowery et al., 1979; Neal and to high K+ or veratridine (Neal and Bowery, 1979). Bowery, 1979; Hertz and Schousboe, 1980; Jaffe and During a reinvestigation of this question, we observed Cuello, 1981; Pearce et al., 1981). However, there has that preloaded [3H]GABA could be released by depolar- been disagreement as to whether this release is Ca*+- ization of the retina. However, the amount of release was dependent or not. Recently, a GABA mimetic, ethyle- small and appeared to depend on the physiological con- nediamine, has been shown to release GABA from rat dition of the tissue. In order to pursue this observation more rigorously, we used a double isotope-labeling technique in which 1 This study was supported in part by National Institutes of Health Research Grants EY-03523, EY-03664, and EY-01730 from the Na- glial cells were preloaded with [“HIGABA while a popu- tional Eye Institute, and in part by an unrestricted grant from Research lation of neurons was prelabeled with [14C]glycine. Au- to Prevent Blindness, Inc. I thank Bill Catterall for suggesting the use toradiographic studies have shown that, unlike GABA, of cu-latrotoxin, Brad Clifton and Jim Foltz for photographic help, and which is taken up by glial cells, glycine is accumulated Julie Seng for secretarial assistance. by a subset of amacrine cells (neurons) located in the 2494 The Journal of Neuroscience [3H]GABA Release from Retinal Glia 2495 inner nuclear layer (Bruun and Ehinger, 1974; Marshall (DMP-30). After a 3-hr treatment with resin and DMP- and Voaden, 1974). This double-labeling procedure pro- 30 mixture, the tissue was transferred to fresh Epon in vided us with an opportunity to examine glial and neu- flat embedding molds and carefully oriented for radial ronal release concurrently. sectioning. Polymerization was carried out at 65°C for Using this approach, we have demonstrated that [“HI 48 hr. GABA can be released from glial cells by depolarization About l-pm-thick sections were cut on a Reichardt with high K’ or veratridine. Furthermore, this release microtome and placed on acid-cleaned glass slides. Slides was suppressed by removal of external Ca’+. We have were dipped in the dark, in 50% Kodak NTB2 nuclear also shown that [“HIGABA can be selectively released tract emulsion at 40°C. After exposure for 1 to 4 weeks from glial cells by exposure to ethylenediamine and that at 4°C and 0% relative humidity, slides were developed this release occurs by a Ca’+-dependent mechanism. in 50% Dektol for 2 min at 15°C. After a brief rinse in water, the slides were fixed in Kodak fixer for 15 min at 15°C. After three washes in water, the sections were Materials and Methods stained with diluted Richardson’s stain. Autoradiograms Sprague-Dawley rats (-200 to 250 gm body weight) were examined under transmitted and darkfield illumi- were anesthetized with ether and decapitated. After en- nation to assess the distribution of silver grains. ucleation, the lens was cut off and the retina was gently Release experiments. After preincubation, retinas were isolated in mouse bicarbonate Ringer (MBR). The com- transferred to 0.2 ml of MBR containing 100 pM AOAA, 5 mM APS mixture, and 1 &i/ml each of [“HIGABA position of this Ringer is (in mM): NaCl, 120; KCl; 4.8; NaHC08, 25; CaC12, 1.9; MgC12, 0.49; dextrose, 5.0; and and [14C]glycine (112 mCi/mmol). Following 20 min in- HEPES, 2.8. The medium was adjusted to pH 7.55 and cubation, retinas were rinsed and then washed three bubbled with a 95% 0, and 5% CO, mixture to yield a times, for 5 min each, in 20 ml of MBR containing AOAA final pH of 7.45. (100 PM). Next, a piece of retina was introduced into a Autoradiography. After preincubation for 5 min at 15-ml conical tube containing 2 ml of MBR. At 2-min 37°C retinas were incubated in 0.2 ml of MBR contain- intervals, 1.9-ml aliquots were withdrawn and replaced ing either 100 &i/ml of [“HIGABA (28.2 Ci/mmol) or by the same amount of fresh medium. At 10 min, MBR 50 &X/ml of [“Hlglycine (12.3 Ci/mmol). The incubation contained 56 mM K+ or the test chemical, such as vera- medium also contained 100 pM amino-oxyacetic acid tridine, was added. After recovering three 2-min samples, (AOAA) and a mixture of alanine, proline, and serine, the medium was changed back to MBR, and isotope each at 5 mM concentrations (APS mixture). These release was followed for an additional 12 min. At the end amino acids have been reported to block the leucine- of the experiment, the tissue was homogenized in 1 ml preferring (L) neutral amino acid uptake system by of 10 InM HCl. Aliquots (1 ml) of the different fractions which glycine enters cells, without significantly affecting and the homogenate were mixed with 10 ml of Aquasol the high affinity system present in glycinergic neurons II (New England Nuclear Corp., Boston, MA), and the (Oxender and Christensen, 1963; Sepulveda and Smith, radioactivity was determined in a Beckman scintillation 1978; Chin and Lam, 1980). Incubations with [“HIGABA counter. AOAA was present throughout the course of the were carried out in the presence of 4 pM glycine, and experiments. [“Hlglycine incubations were carried out in the presence Radioactive counts were calculated on the basis of of 3.5 pM GABA. Following a 20-min incubation at 37”C, single or double-isotope internal standards. “H activity the retinas were washed for 5 min in MBR and fixed in was counted at an efficiency of 12%, whereas 14C effi- 3% glutaraldehyde in 0.1 M sodium phosphate buffer, pH ciency was about 40%. The counting windows were set 7.4, for 4 hr at room temperature, and overnight at 4°C. such that about 15% of the 14C counts spilled into the In order to examine the effect of the presynaptic toxin, “H channel. Release is defined as: latrotoxin, retinas preincubated with [“HIGABA or [“HI (cpm under peak of release - basal level efflux) glycine were rinsed and further incubated for 20 min in (cpm in all fractions + homogenate) 0.2 ml of bicarbonate Ringer containing 20 ~1 of the toxin solution and AOAA and APS. Subsequently, retinas were where basal level efflux = (average cpm per prestimula- washed twice in 5 ml of Ringer for 5 min and fixed in tion fraction x number of peak fractions). In several glutaraldehyde. Spider venom toxin was prepared by control experiments, the basal level efflux values were homogenizing 10 black widow spider venom glands found to be similar to those of the nonstimulated con- (Sigma Chemical Co., St. Louis, MO) in 0.25 ml of cold trols. Furthermore, the nonstimulated controls never 0.05 M Tris-HCl buffer, pH 8.2. The homogenate was showed transmitter release, either in the presence or in spun at 15,000 x g, and the supernatant was used (Fron- the absence of Ca’+. This method of estimating the tali et al., 1976). percentage of release is limited by the subjective nature The fixed tissue was washed twice in phosphate buffer of the basal level and peak cpm estimates. However, (5 min each), and treated with 1% 0~0~ for 2 hr. After since the same criteria are used throughout the study, three 5min washes in the buffer, the tissue was dehy- results from different experiments can be compared. drated in an alcohol series consisting of 25, 75, 90, and Veratridine was used at a concentration of 10 PM in 100% (15 min each). Following dehydration, the tissue all experiments. In order to test for the specificity of was treated with propylene oxide two times, for 15 min veratridine action, Na+ channel blockers, tetrodotoxin each, and left overnight in a 1:l mixture of propylene or procaine were included along with veratridine, at oxide and 2, 4, 6-tri(dimethylaminomethyl)phenol concentrations of 100 nM and 1 mM, respectively. In 2496 Sarthy Vol. 3, No. 12, Dec. 1983 addition, retinas were pretreated with the inhibitors for the 14C uptake in the different retinas. 2 min before exposure to veratridine. Product identification. The identity of 3H and 14C Uptake experiments. Three-millimeter discs of retina compounds accumulated in the tissue was determined as were incubated in 0.5 ml of MBR, containing 5 pM [3H] follows. Retinas were incubated with [3H]GABA and GABA, 1 pM [‘“Clglycine, 100 pM AOAA, and 5 InM APS [14C]glycine as described earlier and homogenized in 250 mixture for 15 min at 37°C. In addition, 1 mM nipecotic ~1 of 0.1 N HCl containing GABA and glycine (2 mg/ml) acid or 5 mM ethylenediamine (EDA) were included when each. Fifty microliters of this mixture were spotted on required. After incubation, retinas were washed for 5 min Whatman 3 MM paper and electrophoresed at 6000 V to get rid of extracellular radioactivity, homogenized in for 1 hr at pH 1.9. The paper was air dried and later 10 mM HCl, and counted for radioactivity. 14C counts stained with ninhydrin. The spots corresponding to accumulated in the different conditions were close but GABA and glycine were cut out and eluted with 1 ml of not identical, and the 3H radioactivity was normalized to 10 mM HCl, and their radioactivity was determined.

Figure 1. Autoradiograms of retinas incubated with [3H]GABA (A) and [14C]glycine (D). Retinas were incubated with 100 pCi/ml of [3H]GABA or [3H]glycine in the presence of AOAA (0.1 mM) and 5 mM each of alanine, proline, and serine (APS). Glycine was included at a concentration of 4 pM in [3H]GABA incubations, and 3.5 pM GABA was added to [“‘C]glycine incubations. After 20 min at 37”C, the retinas were rinsed, fixed in 3% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), and processed for autoradiography. Alternatively, preloaded retinas were washed five times for 5 min in Ringer containing AOAA (C) and stimulated with veratridine for 6 min before fixation (D). In some experiments, retinas prelabeled with [3H]GABA (E) or [3H]glycine (F) were incubated with spider toxin for 20 min at 37°C in the presence of AOAA and APS. Subsequently, the retinas were washed three times for 5 min, fixed, and processed. on, outer nuclear layer; in, inner nuclear layer; ip, inner plexiform layer. Scale bar = 60 pm. The Journal of Neuroscience [“HIGABA Release from Retinal Glia 2497

More than 99% of 3H and -95% 14C activity in the tissue distribution of silver grains was essentialy the same as co-migrated with GABA and glycine spots, respectively. that in unwashed controls (Fig. lA), which suggests that These results established that little GABA and glycine little redistribution of “H activity occurred during postin- metabolism occurred under our experimental conditions. cubation of the retina. Similarly, the distribution of [“HI Solutions. High potassium (56 mM) Ringer was pre- glycine label was not changed in veratridine-treated re- pared by replacing an equivalent amount of Na+ from tinas (Fig. 1E). MBR with K’. Na+-free medium was made up by replac- K+-euoked release. Because glial cells are considered to ing NaCl and NaHC03 with choline chloride and adding be highly sensitive to external K+ (Kuffler and Nicholls, 10 mM Tris-HCl to the medium. 1966), we examined the effect of elevated K+ on the Ca*+-free medium was prepared by eliminating CaClz release of [“HIGABA and [14C]glycine. Retinas were in- and including 10 mM MgCl, or 5 mM CoC12 in the cubated with the labeled transmitters, washed to elimi- medium. nate extracellular radioactivity, and exposed to a 6-min Materials. All chemicals used were AnalR grade. pulse of 56 mM K+-Ringer. When Ca2+ was eliminated GABA and AOAA were purchased from Sigma Chemical from the high K+-Ringer, 5 mM Co2+ or 10 mM Mg2+ was Co., St. Louis, MO. The radiochemicals used were pur- included. The effect of high K+ stimulation on neuro- chased from New England Nuclear Corp., Boston, MA. transmitter release is presented in Figure 2. Upon expo- Tetrodotoxin and procaine were kindly given to us by sure to high K+, a marked increase in the release of both Dr. Bertil Hille, and Dr. William Catterall provided us “H and 14C isotope was observed. This high release rate with veratridine. was reduced, upon returning the tissue to normal Ringer. When Ca2+ was eliminated from the high K+ medium, Results only a small increase in isotope efflux was observed. Localization of uptake sites. Before we began experi- Exposure to high K’ released about 2.1% of t3H]GABA ments on transmitter release, we examined the pattern and 10.5% [14C]glycine from the retina. In the absence of GABA and glycine uptake under our experimental of Ca*+, the amount was reduced to 0.63% and 2.2% for conditions. Retinas were incubated with [3H]GABA or GABA and glycine, respectively (Table I). [“Hlglycine for 20 min at 37”C, rinsed briefly, fixed in glutaraldehyde, and processed for autoradiography. Re- sults of these experiments presented in Figure 1 show . GA13A,+Co2+ that, with [“HIGABA, silver grains were located radially 0 GABA,- Ca2+ throughout the retina with intensely labeled cell bodies A Gly ,+ Ca2+ distributed in the inner nuclear layer (INL). Silver grains were also dispersed at a high density along the external limiting membrane and throughout the inner retina (Fig. 1A). The disposition of the silver grains suggests that most of the labeled cells were the radially oriented glial (Miiller) cells which have cell bodies in the INL and send out processes into both the outer and inner retina (Ra- mon y Cajal, 1982). In the case of [3H]glycine, labeling was confined essentially to the inner retina. Labeled cell bodies were seen in the INL, and the inner plexiform layer (IPL) was also heavily labeled, presumably due to labeling of glycinergic processes (Fig. 1D). We infer that these labeled cells are a subset of amacrine cells which are glycinergic neurons. Our observations on GABA labeling of glial cells and glycine uptake by amacrine cells are in agreement with previous reports (Neal and Iversen, 1972; Brunn and Ehinger, 1974; Marshall and Voaden, 1974; Voaden, 1976). It has been reported that a rapid redistribution of 3H activity occurs with prolonged washing of rabbit retinas preloaded with [“HIGABA (Ehinger, 1977). Since some of the newly labeled cells appear to be a population of amacrine cells, it has been suggested that postincubation 2 4 washing may lead to transfer of 3H label from glial cells Fraction number to neurons. To examine this possibility, we incubated rat retinas with [“HIGABA for 15 min at 37°C in the pres- Figure 2. Effect of high K’ on [3H]GABA and [‘*C]glycine release. Retinas were incubated with [3H]GABA and [‘“Cl ence of AOAA and APS mixture. After washing five glycine for 15 min at 37°C. After a series of washes, they were times for 5 min at 37°C in normal Ringer, the retinas exposed to bicarbonate Ringer containing 56 mM K’ for 6 min. were fixed in glutaraldehyde and processed for autora- Radioactivity in samples collected during the experiment was diography. Alternatively, washed retinas were exposed to determined by standard double-isotope scintillation counting 10 pM veratridine for 6 min and processed, Results of methods. The Ca’+-free Ringer was devoid of Ca2+ and con- the experiment presented in Figure 1B show that the tained 5 mM Co*+. 2498 Sarthy Vol. 3, No. 12, Dec. 1983

TABLE I Amount of transmitter released under various conditions Transmitter release is expressed as percentage of input cpm which ranged from 10 to 12 X lo4 cpm for [3H]GABA and 18 to 20 X lo3 for [‘“Cl glycine. All values are mean -t SEM. K+ Veratridine EDA GABA

+Procaine NH,’ (3) +Ca*+ (6) -Ca*+ (6) +Ca*+ (6) +TTX (3) -ca*+ (4) +c!a*+ (3) -ca2+ (3) +ca2+ (4) -c2+ (4) (3) [3H] GAB A 2.1 0.63 0.93 CO.06 co.17 0.26 1.8 1.3 11.0 6.5 0.18 k1.0 kO.2 kO.03 fO.l f0.2 kO.3 k1.6 k2.1 fO.l [‘%]Glycine 10.5 2.2 16.4 1.1 0.90 3.2 CO.1 CO.1 1.1 0.4 1.6 f2.4 kO.8 k1.5 kO.6 kO.2 kO.7 kO.5 f0.3 kO.4 ’ Numbers in parentheses, number of separate experiments.

Release from spider toxin-treated retina. Although our 0 GABA, - ST autoradiographic data show that [3H]GABA is accumu- o GABA, + ST lated predominantly by the radially oriented Muller cells, A Gly ,-ST the data cannot rule out labeling of some GABAergic A Gly , + ST neurons additionally. Furthermore, the K+-evoked, Ca*+- dependent release of [“HIGABA could be adduced as evidence for the presence of [3H]GABA-labeled neurons in the retina. In order to examine this possibility, we have studied transmitter release from retinas treated with the presynaptic toxin, oc-latrotoxin. This compound, which is present in the venom glands of the black widow spider (Latrodectus mactans tredecimguttatus), is a protein of M, = 130,000, which has been demonstrated to cause massive release of acetylcholine, noradrenaline, dopa- mine, and GABA from cortical slices (Tzeng and Siek- evitz, 1978, 1979; Tzeng et al., 1978) and synaptosomes (Grass0 and Senni, 1979; Baba and Cooper, 1980; Mel- dolesi, 1982; Nicholls et al., 1982). Retinas were prelabeled with [3H]GABA and [‘“Cl glycine, rinsed, and incubated with the toxin for 20 min at 37°C. After three 5min washes in 5 ml of Ringer containing AOAA, Kc-evoked transmitter release was examined. Results of the experiment are presented in Figure 3. Treatment with high K+ resulted in a large K+ increase in [“HIGABA release from both toxin-treated I I r I I I 2 4 6 8 IO 12 14 I6 and control retinas. In contrast, only a small increase in Fraction number [14C]glycine release occurred from toxin-treated retinas, although substantial glycine release was observed in un- Figure 3. High K+-evoked release of [“H]GABA and [‘“C] treated control retinas. In two separate experiments, the glycine from spider toxin-treated retina. Retinas prelabeled average GABA release was 2.3 and 2.1% for the control with [“HJGABA and [“Clglycine were rinsed and incubated in and the toxin-treated retinas, respectively. With glycine, 0.2 ml of Ringer containing 20 ~1 of spider venom gland extract (ST) and AOAA and APS. After 20 min at 37”C, the retinas release amounted to 9.5% for the control and 1.5% for were washed three times for 5 min and exposed to high K+- the toxin-treated retinas. Since toxin treatment leads to Ringer. Radioactivity in different fractions was determined by a loss of >80% of [14C]glycine release, it may be con- scintillation counting. cluded that most of the [“HIGABA release occurred from Muller cells. To further explore the effect of spider toxin, we have Veratridine-induced release. The alkaloid, veratridine, examined the distribution of [3H]GABA and [“Hlglycine depolarizes neurons by selectively increasing the perme- in toxin-treated retinas using autoradiography. Results ability of resting membrane to external Na+ by activation of the study are shown in Figure 1, C and F. Whereas of the voltage-dependent Na+ channel (Ulbricht, 1969; toxin treatment did not appear to alter prelabeled [“HI Narahashi, 1974). In order to examine the presence of GABA distribution (Fig. lC), the pattern of [“Hlglycine such channels on glial cell membrane, we examined the uptake was markedly affected. In these retinas, most of effect of 10 PM veratridine on the efflux of [“HIGABA the [“Hlglycine had been depleted from cell bodies and and [14C]glycine from preloaded retinas. Results of the the IPL. However, many heavily labeled cells were still experiments presented in Figure 4 show that exposure to present after toxin treatment (Fig. 1F). These autoradi- veratridine resulted in a significant increase in the re- ogaphic data are in agreement with the lack of [‘“Cl lease of both “H and 14C radioactivity. This elevated glycine release from toxin-treated retinas. release was diminished when veratridine-containing me- The Journal of Neuroscience [3H]GABA Release from Retinal Glia 2499 dium was replaced by normal Ringer. Although less effective than high K’, veratridine still released about 0 GABA ,+ Co*+ 0.93% of [“HIGABA. [14C]Glycine release was enhanced 0 GAEA,- Co*+ to 16.4% (Table I). A Gly ,+Ca*+ The specificity of veratridine action was established by carrying out stimulation with the drug in the presence of 100 nM Tetrodotoxin (TTX), a specific inhibitor of the Na+-channel (Narahashi et al., 1964). As shown in Figure 4, veratridine-stimulated release was markedly suppressed in the presence of the toxin. Only about 0.06% of 3H and 1.1% of 14C radioactivity was released under these conditions (Table I). In analogous experiments, procaine, another inhibitor of voltage-sensitive Na+ channel, which acts by a mechanism different from that of TTX (Hille, 1977), also blocked veratridine-stimulated release of [3H]GABA and [14C]glycine (Table I). It is generally accepted that transmitter release by veratridine involves an influx of Ca*+ following activation of the Na+ channel at the nerve terminal (Blaustein, 1975). Since veratridine stimulated [“HIGABA release from glial cells, we examined the Ca*+ dependency of the release mechanism. In these experiments, preloaded retinas were exposed to Ca’+-free Ringer containing 10 mM Mg’+ and 10 pM veratridine. Results of the experiment presented in Figure 5 show that removal of Ca*+ leads to Ver 0, 8 I I 1 t 0 2 4 6 8 IO 12 14 Fraction number

l GABA, - TTX Figure 5. Influence of Ca*+ on veratridine-stimulated release 0 GABA,+ TTX of [3H]GABA and [‘Tlglycine. Retinas preloaded with [“HI A Gly ,- TTX GABA and [%]glycine were exposed to 10 pM veratridine in normal Ringer or Ca’+-free Ringer supplemented with 10 pM A Gly,+TTX Mg’+.

a reduction in the amount of GABA (0.26%) and glycine (3.2%) released. Effect of ethylenediamine. EDA has been observed to be a strong inhibitor of neuronal firing in the cerebral cortex with a potency comparable to that of GABA. Moreover, this action of EDA could be blocked by bicuculline, a GABA receptor antagonist (Perkins et al., 1981). EDA has also been reported to block uptake of [‘“CIGABA by brain slices, stimulate [3H]diazepam binding, block [“HIP-alanine uptake into cortical slices (glial uptake), and selectivity release preloaded GABA from cortical slices or synaptosomes by a Ca’+-independent mechanism (Forster et al., 1981; Davis et al., 1982; Lloyd et al., 1982). We have examined the effect of EDA on the efflux of GABA from Muller cells both in the presence and in the absence of external Ca*+. Retinas preloaded with [3H] GABA and [14C]glycine were exposed to 5 mM EDA in Ver normal Ringer or in Ca*+-free Ringer containing 10 mM , I 1 I I I 2 4 6 8 IO 12 14 16 Mg*+. Results of the experiments presented in Figure 6 Fraction number show that, upon exposure to EDA, there was a significant release of [“HIGABA, whereas little [14C]glycine release Figure 4. Effect of veratridine on [3H]GABA and [14C]glycine occurred. When Ca*+ was eliminated from the Ringer release. Retinas were incubated for 20 min at 37°C in bicarbonate Ringer containing [3H]GABA and [14C]glycine. After initial solution containing EDA, a large induction of [“HI GABA release still occurred, whereas [14C]glycine release re- washes, the retinas were exposed to 10 FM veratridine for 6 min. Radioactivity in the different fractions was determined by mained unaffected. EDA stimulated about 1.8% GABA scintillation counting. When stimulation with veratridine was and < 0.1% glycine release in normal Ringer. In the carried out in the presence of TTX, the retinas were preincu- absence of Ca*+, EDA induced about 1.3% GABA, bated with TTX in normal Ringer for 2 min. whereas no change in glycine release was noted. 2500 Sarthy Vol. 3, No. 12, Dec. 1983 In order to examine the role of Na+ in EDA-stimulated l GABA,+ Ca2+ release of [3H]GABA, experiments were performed in 0 GABA,- Ca2+ Na+-free medium with and without added Ca2+. Although A Gly,+Ca2+ the basal rates of “H and 14C isotope efflux were lo-fold A Gly , - Ca2+ higher in this medium than in normal Ringer, no meas- urable stimulation of [3H]GABA or [14C]glycine was observed upon exposure to EDA. No differences in release were noted in Ca2+-containing or Ca2+-free medium (Fig. 7). Effect of GABA. Since EDA is a GABA mimetic, a direct mechanism for the selective release of [3H]GABA would be by homoexchange of intracellular [3H]GABA with exogenous EDA. In order to examine this possibility, we studied the release of [3H]GABA in the presence of 1 mM GABA. Results of the experiment presented in Figure 8 show that, in the presence of GABA, about 11% of [3H]GABA was released from the tissue, whereas only 1.1% [‘4C]glycine was released concurrently. Further- more, in the absence of Ca2+, about 6.5% [“HIGABA was still released compared to 0.4% for [14C]glycine. Effect of NH4+. Yet another mechanism by which EDA could release [3H]GABA is by entering cells as an un- charged molecule. Once inside, it could get protonated, and thus elevate the intracellular pH, which in turn could EDA lead to release of GABA. It is well established that 1 I I 1 treatment of cells with NH4+ leads to an increase in 2 4 6 8 IO 12 14 Fraction number Figure 6. Effect of EDA on [“H]GABA and [Wlglycine release. Retinas preloaded with [“HIGABA and [“‘Clglycine 0 GABA,+Ca2+ were exposed to 5 mM EDA in normal or Ca*+-free Ringer 0 GABA,- Ca2+ containing 10 mM Mg”. A Gly ,+Ca2+ A Gly ,- Ca2+

)- l GABA,+ Ca2+ 0 GABA, - Co’+ A Glv ,+ Ca2+ A Gly I- Ca2+

I-

EDA ‘5%., . . . *. .::.. I 0 2 4 6 6 IO 12 1’4 Fraction number Figure 8. Effect of GABA on [3H]GABA and [Wlglycine Figure 7. Effect of EDA on [3H\GABA and [YJglycine release. Retinas preloaded with the transmitters were exposed release in Na+-free medium. Retinas preloaded with [“H]GABA to 1 mM GABA in normal or Ca’+-free Ringer containing 10 and [‘%]glycine were washed and exposed to 5 mM EDA in mM Mg’+. The dotted bar indicates duration of exposure to Na+-free medium either containing Ca2+ or without Ca’+. GABA. The Journal of Neuroscience [“HIGABA Release from Retinal Glia 2501

A Gly , + Ca2+ Voaden and Starr, 1972; Minchin and Iversen, 1974; 0 GABA,+ Ca2’ Roberts, 1974; Schrier and Thompson, 1974; Minchin, 1975; Hertz, 1977; Sellstrom and Hamberger, 1977; Bow- ery et al., 1979; Neal and Bowery, 1979; Hertz and Schousboe, 1980; Jaffe and Cuello, 1981; Pearce et al., 1981). In our experiments using 56 DIM K+ stimulation, about 70% of the GABA release was blocked by removal of external Ca2+ from the medium. We do not know whether the remaining (30%) [“HIGABA release is due to a Ca2+-independent component or some other factor, since [ 14C]glycine release from neurons was only reduced by about 80% under the same conditions. With veratridine, glia from central nervous tissue have been reported to release significant amounts of preloaded @ NH; GABA, although peripheral glia from sensory, sympa- 0 1 I I I 1 thetic, or trigeminal ganglion cells responded poorly 0 2 4 6 8 IO 12 14 (Minchin, 1975; Neal and Bowery, 1979; Jaffe and Cuello, Fraction number 1981). In a different system, glial cells in culture were Figure 9. Effect of NH,’ on [“HIGABA and [Ylglycine found to accumulate exogenous 22Na upon stimulation release. Retinas preloaded with the transmitters were exposed with veratridine (Munson et al., 1979; see also Catterall to 5 mM NH&l in normal Ringer. and Nirenberg, 1973; Stallcup and Cohen, 1976; Villegas et al., 1976). These observations have been put forward as evidence for the presence of voltage-sensitive Na+ channels on glial cells. Our experimental results also suggest that Miiller cells in the retina may respond to veratridine stimulation by activation of TTX-inhibited, voltage-sensitive Na+ channels on their membrane. It is intriguing that, although the K+-evoked release of both GABA and glycine was Ca2+ dependent, the effect of a-latrotoxin on release was markedly different for the two transmitters. Whereas K+-evoked [3H]GABA release was unaffected, [14C]glycine release was strongly suppressed in toxin-treated retinas. a-Latrotoxin appears to act by increasing the Na+ and Ca2+ conductance of presynaptic membranes by a mechanism which is independent of the voltage-sensitive Na+ and Ca2+ channels. This in turn results in loss of the ion electrochemical potential and subsequent release of transmitter (Nicholls et al., Figure 10. Inhibition of [“HIGABA uptake by EDA and 1982). If such a mechanism accounts for the [14C]glycine nipecotic acid (NA). Retinas were incubated with [3H]GABA efflux from toxin-treated retinas, our results with [3H] and [14C]glycine for 15 min at 37°C. EDA and NA were present GABA release suggest that a-latrotoxin-binding sites are at a concentration of 5 mM and 1 mM, respectively. After absent on the Miiller (glial) cells. Our autoradiographic incubation, the retinas were washed and homogenized in 0.5 data are in accord with this inference. ml of 10 mM HCl, and the radioactivity was determined by double-isotope scintillation counting. 3H counts were normal- When examining the effect of veratridine and EDA on ized to the 14C counts taken up. Values are mean f SEM for GABA release, we did not employ toxin-treated retinas three separate experiments. because concurrent release of the second radiolabel, [‘“Cl glycine, could not be monitored in such experiments. In such a protocol, the advantage of double-label studies intracellular pH (Thomas, 1974). Hence, we examined would be lost. the effect of 5 mM NH4+ on the efflux of [3H]GABA and A basic limitation of our studies is the assumption that [14C]glycine. We found that NH,+ ions stimulated release the Miiller cells alone take up [“HIGABA under our of both [3H]GABA (0.18%) and [14C]glycine (1.6%) from experimental conditions. Although this observation is in the retina (Fig. 9). general agreement with findings of others, we have not performed any independent tests to confirm the validity Discussion of our assumption. Indeed, even if a small number of Both K+- and veratridine-evoked release of GABA GABAergic neurons were labeled in addition to the glial have been examined in glial cells from the CNS and PNS cells, the [3H]GABA release observed may preferentially with varying results. A partial suppression of GABA occur from neurons and not glial cells. However, our release was noted in some cases under Ca’+-free condi- experiments with the presynaptic toxin, a-latrotoxin, tions, whereas others have reported release to be com- indicate that most of the [“HIGABA released probably pletely Ca’+-independent (Bowery and Brown, 1972; did not occur from GABAergic neurons. Due to its limited 2502 Sarthy Vol. 3, No. 12, Dec. 1983 availability and the difficult purification procedure, we Catterall, W. A., and M. Nirenberg (1973) Sodium uptake have used an aqueous extract of the venom glands instead associated with activation of action potential ionophores of of the toxin in our studies. It has been demonstrated cultured neuroblastoma and muscle cells. Proc. Natl. Acad. (Frontali et al., 1976) that the physiological effect of the Sci. U. S. A. 70: 3759-3763. extract on vertebrate neurons is mediated primarily by Chin, C. A., and D. M. K. Lam (1980) The uptake and release the presynaptic toxin, a-latrotoxin, present in this prep- of [3H]glycine in the goldfish retina. J. Physiol. (Lond.) 308: 185-195. aration. Davies, L. P., J. W. Hambley, and G. A. R. Johnston (1982) Neuronal labeling may also occur by redistribution of Ethylenediamine as a GABA agonist: Enhancement of diaze- [“HIGABA during the postincubation washing proce- pam binding and interaction with GABA receptors and up- dures, as reported for the rabbit retina (Ehinger, 1977). take sites. Neurosci. Lett. 29: 57-61. However, our autoradiographic data from retinas fixed Ehinger, B. (1977) Glial and neuronal uptake of GABA, glu- after postincubation wash and veratridine stimulation tamic acid, glutamine and glutathione in the rabbit retina. show that little redistribution of [“HIGABA occurred Exp. Eye Res. 25: 221-234. under these conditions. The presence of 100 pM AOAA Forster, P., H. G. E. Lloyd, P. F. Morgan, M. Parker, M. N. in all of our media might have retarded such a redistri- Perkins, and T. W. Stone (1981) Ethylenediamine acts at GABA receptor and uptake sites. Br. J. Pharmacol. 74: 274P. bution process (Ehinger, 1977). Frontali, N., B. Ceccarelli, A. Gorio, A. Mauro, P. Siekevitz, Yet another mechanism for [‘H]GABA release is that M. C. Tzeng, and W. P. Hurlburt (1976) Purification from endogenous retinal transmitters such as GABA and glu- black widow spider venom of a protein factor causing the tamate, released by veratridine stimulation of neurons, depletion of synaptic vesicles at neuromuscular junctions. J. may act on Muller cells directly or indirectly to release Cell Biol. 68: 462-479. preloaded GABA (Hosli et al., 1980). For example, en- Grasso, A., and M. I. Senni (1979) A toxin purified from the dogenous GABA released into the synaptic cleft may venom of black widow spider venom affects uptake and release preloaded [3H]GABA by a homoexchange pro- release of radioactive y-aminobutyrate and N-epinephrine cess; alternatively, glutamate or aspartate released from from rat brain synaptosomes. Eur. J. Biochem. 102: 337-344. neurons may depolarize Muller cells to release preloaded Hertz, L. (1977) Biochemistry of glial cells. In Cells, Tissue and GABA. Since veratridine-stimulated release of [3H] Organ Cultures in Neurobiology, S. Federoff and L. Hertz, eds., pp. 39-71, Academic Press, Inc., New York. GABA was Ca2+ dependent, we could not distinguish Hertz, L., and A. Schousboe (1980) Interactions between neu- between these possibilities. rons and astrocytes in the turnover of GABA. Brain Res. Unlike K+-and veratridine-stimulated release of Bull. 5: 389-396. GABA, which were suppressed by Ca2+ removal from the Hille, B. (1977) Local anesthetics: Hydrophilic and hydropho- medium, EDA-stimulated release was largely Ca2+ inde- bic pathways for the drug receptor interaction. J. Gen. Phys- pendent. Moreover, little glycine release was observed iol. 69: 497-515. under these conditions. These results, taken together Hosli, L., P. F. Anders, and E. Hosli (1980) Neurone-glia with other studies on neurons, show that EDA can re- interaction: Indirect effects of amino acid transmitters on lease preloaded GABA from both neurons and glial cells cultured glial cells. In Tissue Culture in Neurobiology, E. by a Ca2+-independent process. An obvious mechanism Giacobini, A. Veruadakis, and A. Shahan, eds., pp. 373-383, Raven Press, New York. for this phenomenon would be the homoexchange of Jaffe, E. H. and A. C. Cue110 (1981) Neuronal and glial release intracellular [“HIGABA with exogenous EDA. The ob- of [“HIGABA from the rat olfactory bulb. J. Neurochem. 37: served homoexchange of [“HIGABA with exogenous 1457-1466. GABA (Fig. 8) and the reduction of [“HIGABA uptake Kuffler, S. W., and J. G. Nicholls (1966) The physiology of by EDA (Fig. 10) are compatible with this interpretation. neuroglial cells. Ergeb. Physiol. 57: l-90. A lack of Ca2’ dependency of EDA-stimulated GABA Lloyd, H. G. E., M. N. Perkins, and T. W. Stone (1982) efflux also supports a carrier-mediate exchange rather Ethylenediamine as a specific releasing agent of y-amino- than a vesicular release mechanism. butyric acid in rat striatal slices. J. Neurochem. 38: 1168- 1169. Marshall, J., and M. Voaden (1974) An investigation of the References cells incorporating [“HIGABA and [“Hlglycine in the isolated Baba, A., and J. R. Cooper (1980) The action of black widow retina of the rat. Exp. Eye Res. 18: 367-370. spider venom on cholinergic mechanisms in synaptosomes. Meldolesi, J. (1982) Studies on a-latrotoxin receptors in rat J. Neurochem. 34: 1369-1379. brain synaptosomes: Correlation between toxin-binding and Blaustein, M. P. (1975) Effects of potassium, veratridine and stimulation of transmitter release. J. Neurochem. 38: 1559- scorpion venom on calcium accumulation and transmitter 1569. release by nerve terminals in vitro J. Physiol (Lond.) 247: Minchin, M. C. W. (1979) Factors influencing the efflux of [“HI 617-655. gamma-aminobutyric acid from satellite glial cells in rat Bowery, N. G., D. A. Brown (1972) y-Aminobutyric acid uptake sensory ganglia. J. Neurochem. 24: 571-577. by sympathetic ganglia. Nature New Bio1238: 89-91. Minchin, M. C. W., and L. L. Iversen (1974) Release of [“HI Bowery, N. G., D. A. Brown, and S. Marsh (1979) y-Amino- gamma-aminobutyric acid from glial cells in rat dorsal root butyric acid efflux from sympathetic glial cells. Effect of ganglia. J. Neurochem. 23: 533-540. depolarizing agents. J. Physiol. (Lond.) 293: 75-101. Munson, R., Jr., B. Westermark, B. and L. Glaser (1979) Bruun, A., and B. Ehinger (1974) Uptake of certain possible Tetrodotoxin-sensitive sodium channels in normal human neurotransmitters into retinal neurones of some mammals. fibroblasts and normal human glia-like cells. Proc. Natl. Exp. Eye Res. 19: 435-447. .4cad. Sci. U. S. A. 76: 6425-6429. The Journal of Neuroscience [3H]GABA Release from Retinal Glia 2503

Narahashi, T. (1974) Chemicals as tools in the study of excit- Res. 119: 189-198. able membranes. Physiol. Rev. 54: 813-889. Sepulveda, F. V., and M. W. Smith (1978) Discrimination Narahashi, T., J. W. Moore, and W. R. Scott (1964) Tetrodo- between different entry mechanisms for neutral amino acids toxin blockage of sodium conductance increase in lobster in the rabbit ileal mucosa. J. Physiol. (Lond.) 282: 73-90. giant axons. J. Gen. Physiol. 47: 965-974. Stallcup, W. B., and M. Cohen (1976) Correlation of surface Neal, M. J., and N. G. Bowery (1979) Differential effects of antigens and cell type in cloned cell lines from the rat central veratridine and potassium depolarization on neuronal and nervous system. Exp. Cell Res. 98: 285-294. glial GABA release. Brain Res. 167: 337-343. Thomas, R. C. (1974) Intracellular pH of snail neurons meas- Neal, M. J., and L. L. Iversen (1972) Autoradiographic locali- ured with a new pH-sensitive glass microelectrode. J. Physiol. zation of [“HIGABA in rat retina. Nature New Biol. 235: (Lond.) 238: 159-180. 217-218. Tzeng, M. C., and P. Siekevitz (1978) The effect of the purified Nicholls, D. G., M. Rugolo, I. G. Scott, and J. Meldolesi (1982) major protein factor (oc-latrotoxin) of black widow spider oc-Latrotoxin of black widow spider venom depolarizes the venom on the release of acetylcholine and norepinephrine plasma membrane, induces massive calcium influx, and stim- from mouse cerebral cortex slices. Brain Res. 139: 190-196. ulates transmitter release in guinea pig brain synaptosome. Tzeng, M. C. and P. Siekevitz (1979) The binding interaction Proc. Natl. Acad. Sci. U. S. A. 79: 7924-7928. between cu-latrotoxin from black widow spider venom and a Oxender, D. L., and H. N. Christensen (1963) Distinct mediat- dry cerebral cortex synaptosomal membrane preparation. J. ing systems for the transport of neutral amino acids by the Neurochem. 33: 363-274. Ehrlich cell. J. Biol. Chem. 238: 3686-3699. Tzeng, M. C., R. S. Cohen, and P. Siekevitz (1978) Release of Pearce, B. R., D. N. Currie, R. Beale, and G. R. Dutton (1981) neurotransmitters and depletion of synaptic vesicles in ce- Potassium-stimulated, calcium-dependent release of [“HI rebral cortex slices by a-latrotoxin from black widow spider GABA from neuron and glia-enriched cultures of cells dis- venom. Proc. Natl. Acad. Sci. U. S. A. 75: 4016-4020. sociated from rat cerebellum. Brain Res. 206: 485-489. Ulbricht, W. (1969) The effect of veratridine on excitable Perkins, M. N., N. G. Bowery, D. R. Hill, and T. W. Stone membranes of nerve and muscle. Ergeb. Physiol. 61: 18-71. (1981) Neuronal responses to ethylenediamine: Preferential Varon, S. S., and G. G. Somjen (1979) Neuron-glia interactions. blockade by bicuculline. Neurosci. Lett. 23: 325-327. Neurosci. Res. Program Bull. 17: l-239. Ramon y Cajal, S. (1892) The Structure of the Retina, S. A. Villegas, J., C. Sevcik, F. V. Barnola, and R. Villegas (1976) Thorpe and M. Glickstein, transl. (1972), Charles C Thomas, Grayanotoxin, veratridine an Tetratoxin-sensitive sodium Springfield, IL. pathways in the Schwann cell membrane of squid nerve Roberts, P. J. (1974) Amino acid release from isolated rat dorsal fibers. J. Gen. Physiol. 67: 369-380. root ganglia. Brain Res. 74: 327-332. Voaden, M. J. (1976) Gamma-aminobutyric acid and glycine as Schrier, B. K., and E. J. Thompson (1974) On the role of glial retinal neurotransmitters. In Transmitters in the Visual Pro- cells in the mammalian nervous system. Uptake, excretion, gress, S. L. Bonting, Ed., pp. 107-125, Pergamon Press, New and metabolism of putative neurotransmitters by cultured York. glial tumor cells. J. Biol. Chem. 249: 1769-1780. Voaden, M. J., and M. S. Starr (1972) The efflux of radioactive Sellstrom, A., and A. Hamberger (1977) Potassium-stimulated GABA from rat retina in uitro. Vision Res. 12: 559-566. y-aminobutyric acid release from neurons and glia. Brain