MICROSCOPY RESEARCH AND TECHNIQUE 36:43–56 (1997) Light Adaptation Affects Synaptic Vesicle Density But Not the Distribution of GABAA Receptors in Goldfish Photoreceptor Terminals STEPHEN YAZULLA* AND KEITH M. STUDHOLME Department of Neurobiology and Behavior, State University of New York, Stony Brook, New York 11794 KEY WORDS GABAA receptors; immunocytochemistry; retina; photoreceptors; light adaptation; monoclonal antibodies ABSTRACT GABA is a likely feedback transmitter from H1 horizontal cells to cone photoreceptors in fish retinas. Spinules arise from H1 cell dendrites in light-adapted retinas, are correlated with responses attributed to feedback, and have been proposed to be the GABA release sites. We used mAb 62-3G1, an antibody against the b2/b3 subunits of the GABAA receptor complex, to visualize GABAA receptor immunoreactivity (GABAr-IR) in photoreceptors as a function of light and dark adaptation at the electron microscopical level. Regardless of adaptation, GABAr-IR was restricted to the synaptic terminals of all cones and most rods; synaptic vesicular membrane and plasma membrane exhibited GABAr-IR. Contrary to expectations, the density of GABAr-IR was least on the plasma membrane within the invagination, regardless of the presence or absence of spinules. Dense GABAr-IR was observed on the lateral surface of cone pedicles, on cone processes proximal to the invagination, and on presumed telodendria from nearby cones. There was no difference in GABAr-IR of rod plasma membranes within or outside of the invagination or with adaptation. The only novel effect of adaptation was in regards to the density of synaptic vesicles. Cones showed a 29% increase in vesicle density with dark adaptation, whereas rods showed a 17% decrease. We conclude that all goldfish photoreceptors will be GABA-sensitive and that the sensitivity is distributed over the surface of the synaptic terminal rather than localized to within the invagination. The role of spinules in GABA release remains to be determined, but we conclude that spinules are not related to the GABA sensitivity of goldfish photoreceptors. Microsc Res Tech 36:43–56, 1997. r 1997 Wiley-Liss, Inc. INTRODUCTION There is overwhelming evidence supporting the hypothesis that H1 horizontal cells of the fish retina are GABAergic, providing a feedback inhibition onto cone photoreceptors (for review see Yazulla, 1986). This inhibition appears to be mediated by a bicucullinesensitive, GABA-activated chloride conductance (Djamgoz and Ruddock, 1979; Murakami et al., 1982a,b; Toyoda and Fujimoto, 1983), indicative of GABAA receptors. The actual sites of GABA feedback from horizontal cells to cones have not been identified, and the function of this feedback still generates considerable controversy (for review see Burkhardt, 1993; Wu, 1992). Horizontal cell dendrites that invaginate cone pedicles do not contain synaptic vesicles or other specializations typical of chemical synapses (Stell, 1967; Witkovsky and Dowling, 1969), although small conventional synapses have been observed in the horizontal cell body (Marc and Liu, 1985; Witkovsky and Dowling, 1969). There is strong evidence indicating that H1 horizontal cells release GABA by a sodium-dependent transport carrier (Ayoub and Lam, 1984; Schwartz, 1982, 1987; Yazulla and Kleinschmidt, 1982, 1983), a process that would eliminate the need for synaptic vesicles. H1 horizontal cell dendrites contain electron-dense spinelike protrusions that invaginate the cone pedicle (Stell, r 1997 WILEY-LISS, INC. 1967). These spinules display a striking degree of plasticity with changes in ambient illumination. Numerous spinules protrude from the distal ends of horizontal cell dendrites in the light-adapted state but retract and eventually disappear with dark adaptation (Raynauld et al., 1979; Wagner, 1980). The presence and disappearance of the spinules is directly correlated with light adaptive electrophysiological events (i.e., chromaticity responses in C-type horizontal cells), leading to the suggestion that the spinules may be the sites of feedback interaction (Djamgoz et al., 1988; Weiler and Wagner, 1984). Cones of all spectral classes are innervated by the GABAergic H1 horizontal cells, (Stell and Lightfoot, 1975; Stell et al., 1975), indicating that all cones should be GABA-sensitive. GABAA receptors have been localized to the fish outer plexiform layer (OPL) using 3H-muscimol binding (Lin and Yazulla, 1994) and immunocytochemical detection of monoclonal antibodies (62-3G1) (Vitorica et al., 1988) against the b2/b3 subunit of the GABAA receptor/ benzodiazepine receptor/Cl2 (GABAA/BZDr/Cl2) chan- *Correspondence to: Stephen Yazulla, Dept. Neurobiology and Behavior, SUNY, Stony Brook, NY 11794-5230. Received 27 June 1994; Accepted in revised form 12 October 1994. 44 S. YAZULLA AND K.M. STUDHOLME nel complex (Yazulla et al., 1989). GABAA receptor immunoreactivity (GABAr-IR) was found on the plasma membrane and intracellularly on synaptic vesicle membrane of all cones and to a lesser extent in rods. Curiously, there was no relative increase in GABAr-IR on cone membranes in the region of horizontal cell dendrites, the area of presumed GABA feedback. There are several possible explanations. First, the retinas used in that study were obtained from dark-adapted animals and dissected in a mesopic state. As a result of this procedure, there were very few horizontal cell spinules in the cone pedicles. Since spinules have been implicated in GABAergic feedback, there could have been a reduction in the level of GABAA receptors on the cone plasma membrane in the absence of the spinules. Second, the large amount of presumed intracellular GABAr-IR on synaptic vesicles could have been due to diffusion of the diaminobenzidine (DAB) reaction product from the plasma membrane. Third, since mAb 62-3G1 had not been characterized in any retinal tissue, it was possible that 62-3G1 did not recognize GABAA receptors in goldfish retina. However, this appears not to be the case, because Lin and Yazulla (1994) recently showed that mAb 62-3G1 immunoprecipitated 3H-muscimol binding activity from detergentsolubilized goldfish retinal membranes and, on immunoblots, reacted with 55–57.5 kDa Mr polypeptides, similar to bovine brain (Park and De Blas, 1991). Therefore, we have reinvestigated the distribution of mAb 62-3G1– derived immunoreactivity in goldfish photoreceptors using pre- and postembedding immunocytochemical techniques but as a function of light and dark adaptation to determine if there is a relationship between the distribution of GABAr-IR in the outer plexiform layer and the state of horizontal cell spinule formation. MATERIALS AND METHODS Subjects Goldfish (Carassius auratus), approximately 10–13 cm standard body length, were obtained from Mt. Parnell Fisheries (Mercersburg, PA), maintained in aerated tanks at 22°C filled with tap water circulating through a polyester fiber/charcoal filter system, and fed crushed Purina trout chow. The major light source was a 60 W tungsten lamp about 1 m over the fish tanks. A 12/12 h light/dark cycle was controlled by a timer which turned the light on at 7 AM and off at 7 PM. Fish were either light-adapted under room light fluorescence or dark-adapted for 3 h, after which dissections took place in the early afternoon. Goldfish were decapitated (in compliance with procedures for the sacrificing of small animals). The eyes were enucleated and hemisected. Dissections of dark-adapted fish occurred under indirect red illumination (25 W, BCJ red safety light). Immunoreagents The production and characterization of mAb 62-3G1 has been reported in detail elsewhere (Vitorica et al., 1988). This monoclonal antibody was raised against the GABAAr/BZDr/Cl2 complex from bovine brain that purified by affinity chromatography on the immobilized benzodiazepine R07-1986/1. In goldfish retina, mAb 62-3G1 reacts with 55–57.5 kDa Mr polypeptides on immunoblots and immunoprecipitates 3H-muscimol binding activity but not 3H-flunitrazepam binding activity (Lin and Yazulla, 1994). Goat anti-mouse IgG and mouse peroxidase-antiperoxidase (PAP) were purchased from Dako (Carpinteria, CA); goat anti-mouse IgG conjugated to fluorescein isothiocyanate (FITC) was obtained from Boehringer-Mannheim (Indianapolis, IN). Immunocytochemistry Preembedding Procedures. Following enucleation, half-eyecups were placed vitreous-side down on a type HA Millipore filter (0.45 µm) on a Swinnex filter holder. Light suction was applied to remove the vitreous humor and to provide a stable mechanical support for slicing and subsequent handling of the retina. Procedures for preembedding immunocytochemistry were as described in Eldred et al. (1983) and Yazulla et al. (1989). In brief, retinas were fixed for 1 h in 4% paraformaldehyde and 0.15% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4) at 20°C. After 15 additional minutes in the same fixative without glutaraldehyde, retinas were fixed overnight in 4% paraformaldehyde in 0.1M sodium bicarbonate buffer (pH 10.4) at 4°C. Retinas were then washed briefly, postfixed in 2% OS O4 at 4° for 1 hr in 0.1 M sodium phosphate buffer (pH 7.4) with 4% sucrose and 0.15 M CaCl2 (rinse buffer). The tissue was washed for 30 min in rinse buffer, incubated for 30 min in 1% sodium borohydride in rinse buffer, and washed in several changes of rinse buffer for at least 1 h (until bubble formation ceased) in order to restore much of the immunoreactivity which would otherwise have been masked following glutaraldehyde fixation. The retina was cryoprotected (30 min in rinse buffer with 5% glycerin 1 15% sucrose, 1 h in rinse buffer with 10% glycerin 1 15% sucrose, 12 h in rinse buffer with 10% glycerin 1 20% sucrose), frozen on dry ice, and thawed to enhance the penetration of immunological reagents. Tissue samples were then incubated for 48 h at 4°C with hybridoma culture media containing the mAb 62-3G1 at a 1:50 dilution. Visualization of the antibody labeling was accomplished with a standard peroxidaseantiperoxidase technique (PAP). Conditioned medium of the parental myeloma P3X63Ag18.104.22.168 was used as a control. Controls showed no immunocytochemical staining. Previous studies showed displacement of mAb 62-3G1 immunoreactivity after prior incubation of the antibody with 9 µg of affinity-purified GABAr/BZDr complex (De Blas et al., 1988). Retinal slices were dehydrated and embedded in Durcupan A.C.M. resin. Ultrathin sections were collected on formvar-coated slot grids and, unless stated otherwise, were viewed, photographed, and presented in this paper without heavy metal counterstaining so that the immunoreaction product would not be confused with electron density imparted to the membranes by heavy metal staining. Postembedding Procedures. Serial 1 µm sections were collected on Fisher Superfrost slides. The resin was etched in saturated sodium ethanolate solution for 30–45 min at room temperature (RT). Slides were rinsed in 100% ethanol (3 3 5 min) followed by distilled H2O rinse (3 3 5 min) and then incubated in 1% sodium periodate for 7 min at RT to remove OsO4. Sections were rinsed in distilled H2O and incubated in 1% sodium borohydride for 30 sec to reduce autofluores- GABA RECEPTORS IN PHOTORECEPTORS cence of the tissue due to aldehyde fixation. Sections were incubated in mouse monoclonal mAb 62-3G1 antibodies at concentrations of 1:10 to 1:100 for 48 h at 4°C. Tissues were incubated in goat-anti-mouse IgGs conjugated to FITC for 30 min at 37°C. RESULTS General Observations Consistent with our previous findings (Yazulla et al., 1989), GABAr-IR was concentrated in the distal portion of the axon and synaptic terminals of rods and cones; this was the case regardless of whether the retinas were light- or dark-adapted (Fig. 1). The state of adaptation at the time of fixation could be verified at the light microscopical level in that cone myoids are fully contracted and rods fully extended in the lightadapted state (Fig. 1A); the reverse situation is observed for dark adaptation (Fig. 1B) (Ali, 1971; Burnside and Nagel, 1983). In well-oriented sections, GABAr-IR could be followed from the synaptic terminals along the axon into the layer of rod nuclei (Fig. 1, arrowheads). It was clear from Figure 1 and from Yazulla et al. (1989) that GABAr-IR is contained within the photoreceptor terminals rather than confined to the plasma membrane. It was possible that the intracellular GABAr-IR was due to diffusion of DAB reaction product during processing. However, visualization of GABAr-IR with a postembedding technique on 1 µm thick resin sections (Fig. 2) clearly shows that the photoreceptor terminals are labeled uniformly with GABAr-IR rather than outlined as would be expected if receptors were confined to the plasma membrane. Notice that GABAr-IR can be followed along the cone axon for short distances into the outer nuclear layer (ONL) with both preembedding and postembedding procedures (compare Figs. 1 and 2). Thus, the intracellular distribution of GABAr-IR following prembedding procedures is not an artifact of DAB processing but represents an accurate view of GABAr-IR in photoreceptor terminals. At the ultrastructural level, GABAr-IR was found intracellularly and on the plasma membrane of all cones and most rods in both the light-adapted and dark-adapted states (Fig. 3). Rod spherules, although more variable in their staining for GABAr-IR, tended to stain more intensely than cone pedicles in the lightadapted state (Fig. 3A) but not in the dark-adapted state (Fig. 3B). Most of the intracellular GABAr-IR was associated with the membrane of synaptic vesicles, with occasional clumps of GABAr-IR scattered throughout the terminal. For the most part, synaptic vesicles were not filled with GABAr-IR but appeared hollow with GABAr-IR concentrated around the cytoplasmic surface of the synaptic vesicle membrane (e.g., see Figs. 6B and 8A). GABAr-IR synaptic vesicles were distributed throughout the receptor terminal, extending into the photoreceptor axon where their numbers and consequently GABAr-IR decreased rather sharply (Fig. 3, arrowheads). It was our impression that the differential GABAr-IR staining in rod and cone terminals could be due to differences in the density of synaptic vesicles. 45 Adaptation Affects Synaptic Vesicle Density To test this idea, retinas from light-adapted fish and those dark-adapted for 3 h were processed in parallel with those reacted for GABAr-IR, but they were not subjected to immunocytochemical processing (Fig. 4). This was done to increase the visibility of synaptic vesicles that would be obscured by the DAB reaction product. It was obvious from even a casual inspection of the micrographs that the density of synaptic vesicles was less in light-adapted cone pedicles than rod spherules (Fig. 4A), whereas the densities appeared more comparable in the dark-adapted state (Fig. 4B). A grid was used to count the number of synaptic vesicles in four separate regions of each synaptic terminal. The four samples were pooled to produce the vesicle density for that terminal. At least ten synaptic terminals each of rods and cones were tabulated for the light- and dark-adapted conditions. Three pair of retinas were analyzed in this fashion. Synaptic vesicle densities were quantified, and the results are presented in Figure 5. The ratio of synaptic vesicle density for rods:cones was 1.86:1 (P Ò 0.001) for the light-adapted state and 1.2:1 (P , 0.05) for the dark-adapted state. The reduction in this ratio was accounted for by two factors: a 29% increase (P , 0.001) in cone synaptic vesicle density and a 17% decrease (P , 0.001) in rod synaptic vesicle density. Cone Pedicles It should be pointed out again that the micrographs to be described were obtained from tissue sections that were not counterstained with heavy metals. The electron density of plasma membranes and synaptic vesicle membranes was due largely to GABAr-IR. However, the electron density of synaptic ribbons and the intracellular surface of horizontal cell spinules and processes is observed with osmication but without counterstaining and thus is not due to GABAr-IR. Our original rationale was based on the hypothesis that horizontal cell spinules were the sites of GABA release (Djamgoz et al., 1988; Weiler and Wagner, 1984), and thus GABAr-IR should be enriched on cone membrane facing the spinules. However, despite an extensive survey of numerous cone pedicles, obtained from several retinas, we observed no obvious increase in GABAr-IR on cone membranes opposing the spinules (Fig. 6). GABAr-IR appeared on the plasma membrane of the entire synaptic terminal, with dense patches of GABAr-IR on the perimeter of the pedicle (Fig. 7B, small arrows). Although cone membrane opposite horizontal cell spinules was weakly GABAr-IR (Fig. 6B), the density of GABAr-IR opposite the spinules was no greater and often appeared less than GABAr-IR on other regions of the cone pedicle membrane (Fig. 7). Similarly with dark-adapted cones, GABAr-IR did not appear more dense on plasma membrane within the invagination than in other regions (Fig. 8). Dendrites of horizontal cells contain very few spinules in the darkadapted state. Instead, the dendrites are smooth and mostly associated with synaptic ribbons of photoreceptors. It often appeared as if GABAr-IR increased in density as one proceeded along the cone membrane away from the synaptic ribbon to the more lateral 46 S. YAZULLA AND K.M. STUDHOLME Fig. 1. Light micrographs of GABAr-IR in light-adapted (A) and dark-adapted (B) goldfish retinas; 1 µm sections of retinas embedded in Durcupan resin after preembed immunocytochemistry. GABAr-IR was found in both rod and cone synaptic terminals regardless of adaptive state. Arrowheads indicate GABAr-IR in the connecting axon of cones that are extending into the rod nuclear layer. Cones (C) are fully contracted in the light-adapted state, and rods (R) are fully contracted in the dark-adapted state. Calibration bar 5 20 µm. GABA RECEPTORS IN PHOTORECEPTORS 47 Rod Spherules Regardless of level of adaptation, GABAr-IR of rod spherules was far more variable than that observed with cone pedicles, ranging from very weak to intense. Intensely labeled rod spherules are illustrated in Figure 9. Unlike with cone pedicles, the plasma membrane of rod spherules was stained uniformly, both within and outside the invagination. There seemed to be no differential distribution of GABAr-IR along the plasma membrane of rod spherules nor were any differences in GABAr-IR vis-à-vis horizontal cell dendrites noted. Also, except for a decrease in the density of synaptic vesicles, there were no apparent differences in GABAr-IR of rod spherules with light or dark adaptation. Fig. 2. Light micrograph of 1 µm resin section of light-adapted goldfish retina processed for GABAr-IR by postembedding immunofluorescence (1:25 dilution of mAb 62-3G1). GABAr-IR was most intense in the synaptic terminals of rods and cones in the outer plexiform layer. Arrowheads indicate GABAr-IR in cone axon terminals as they enter the region of rod nuclei, the same as observed in Fig. 1. Calibration bar 5 20 µm. aspect of the horizontal cell dendrites (Fig. 8A, arrowheads). Regardless of the state of adaptation, membranes with the most intense GABAr-IR were found on processes that were just proximal to the invagination of the cone pedicle. In some cases, these were continuous with the overlying pedicle and appeared to encapsulate a horizontal cell process (Fig. 8b, arrowheads). However, in most examples, these processes with GABAr-IR were observed in isolation, but they had cytological features that were identical to the cone pedicles (Figs. 6A, 7, large arrows), indicating that they were part of cone pedicles. In some cases, these processes with GABAr-IR were within, though not deeply, the invagination (Figs. 6A, 7B, arrows); whereas in other cases they were proximal to the pedicle (Fig. 7A, arrows). Other intense processes with GABAr-IR were smaller in diameter and occasionally were observed projecting from a pedicle (Fig. 8C). Curiously, when cut longitudinally, these processes did not exhibit GABAr-IR uniformly along their length. Identifiable bipolar cell dendrites within the cone pedicle did not show GABAr-IR (Fig. 8C, B arrowheads), and it seems likely that these all of these intense GABAr-IR processes were cone telodendria. Note the similarity in appearance between the intense GABAr-IR processes continuous with or projecting from a pedicle (Fig. 8C,T, arrowheads) with the disconnected GABAr-IR process proximal to the pedicle. We did not determine the origin of the cone telodendria—that is, whether the telodendria derived from the overlying or nearby cone pedicles. There was no obvious presynaptic structure adjacent to these GABAr-IR processes. DISCUSSION We have corroborated our earlier finding that GABAr-IR is found on the plasma membrane of all cone pedicles and many rod spherules and is associated with the membrane of synaptic vesicles (Yazulla et al., 1989). In addition, we found that despite the light-dependent plasticity in the formation of horizontal cell spinules, there was no relative increase in GABAr-IR on cone plasma membrane opposing the spinules, nor was there any corresponding change in the density of GABAr-IR on the cone plasma membrane opposing the horizontal cell dendrites. On the contrary, the density of GABAr-IR was weakest on cone membrane within the invagination whether or not spinules were present, as opposed to patches of high GABAr-IR density on the perimeter of the pedicle and on cone processes proximal to the invagination. The only novel effect of light adaptation was on synaptic vesicle density: an increase in rod spherules and a decrease in cone pedicles. Our findings raise several issues: 1) how the distribution of GABArIR, determined by mAb 62-3G1, relates to GABAA receptors in fish retina, 2) what the locations are of GABA receptive elements in the fish OPL, and 3) how to explain the effect of light and dark adaptation on synaptic vesicle density in photoreceptor terminals. mAb 62-3G1 Localizes GABAA Receptors in Fish OPL The validity of mAb 62-3G1 as an indicator of GABAA receptors in the fish retina has been treated in detail by Lin and Yazulla (1994). However, this paper specifically relates to the outer plexiform layer. The presence of intracellular GABAr-IR in photoreceptor terminals appears valid and not an artifact of the HRP/DAB reaction process because it persists with postembedding ICC procedures (see Fig. 2) that are not subject to diffusion of reaction products. Lin and Yazulla (1994) showed that 3H-muscimol binding sites were immunoprecipitated by mAb 62-3G1 and in addition were localized to the OPL in goldfish retina by dry autoradiography, indicating that at least some of the GABAr-IR, as determined by mAb 62-3G1, is representative of GABAA receptors. It is possible that there are GABAA receptor sites in the OPL that are not recognized by mAb 62-3G1. For example, the synaptic terminals of the large mixed rod/cone bipolar cells receive massive GABAergic input of the GABAA type (Heidelberger and Matthews, 1991; Marc et al., 1978; Tachibana and 48 S. YAZULLA AND K.M. STUDHOLME Fig. 3. Electron micrographs of GABAr-IR in photoreceptor terminals of light-adapted (A) and dark-adapted (B) retinas. Cone pedicles are indicated (C); other GABAr-IR profiles are rod spherules. Arrowheads indicate the connecting axon of a cone in A and rod in B, illustrating the decrease in GABAr-IR as the axon enters the outer nuclear layer. Note that cone terminals show less GABAr-IR than rod terminals with light adaptation (A) but appear comparably stained with dark adaptation (B). Calibration bar 5 1 µm. Note that sections were not counterstained with heavy metals. GABA RECEPTORS IN PHOTORECEPTORS Fig. 4. Electron micrographs of goldfish photoreceptor terminals after light adaptation (A) and dark adaptation (B). The density of synaptic vesicles appears higher in rods (R) than cones (C) with light adaptation but not with dark adaptation. Also, note the presence of 49 horizontal cell spinules(S) invaginated into the cone pedicle that was light-adapted (A); dark-adapted cone pedicles (B) do not contain spinules. Sections were lightly counterstained with lead citrate and uranyl acetate. sr, synaptic ribbons. Calibration bar 5 0.5 µm. 50 S. YAZULLA AND K.M. STUDHOLME Fig. 5. Histobars illustrating the density of synaptic vesicles of rod and cone terminals as a function of light and dark adaptation (X 6 s.e.m.). All differences were statistically significant: light adapta- tion—rod/cone: t 5 19.7, P Ò 0.0001; dark adaptation—rod/cone: t 5 2.5, P , 0.05; light/dark adaptation—cones: t 5 4.2, P , 0.001; light/dark adaptation—rods: t 5 4.0, P , 0.001. Kaneko, 1987; Yazulla et al., 1987), yet mAb 62-3G1 stains very few of the numerous GABAergic amacrine cell synaptic contacts onto the bipolar cell synaptic terminal (Yazulla et al., 1989). We conclude that mAb 62-3G1 localizes GABAA receptors on goldfish photoreceptor terminals, but the possibility remains that additional epitopes exist that are not recognized by this antibody. Another monoclonal antibody (bd-17; reportedly against the same b2/b3 subunits of the GABAA receptor complex (Ewert et al., 1992) does not label the goldfish OPL (Lin and Yazulla, 1994), a further indication of potential epitope differences. within this environment. Additionally, there were smalldiameter processes within the pedicle that showed intense GABAr-IR. We believe these to be cone telodendria rather than bipolar cell dendrites for two reasons. First, they had similar cytology, and, on occasion, we could identify the initial portion of a GABAr-IR telodendrite emanating from a pedicle (see Fig. 8C). Secondly, no GABA sensitivity has been observed on the dendrites of fish bipolar cells (Tachibana and Kaneko, 1987). Telodendria are a common feature of vertebrate cones (Ramon y Cajal, 1933), and, in teleost fish, they extend about 15 µm from each cone where they make gap junctions and shallow invaginating contacts with nearby cones of similar and different spectral classes (i.e., Kraft and Burkhardt, 1986; Scholes, 1975; Stell, 1980; Witkovsky et al., 1974). Thus, H1 horizontal cells could provide GABAergic feedback to cones at encapsulating basal processes of the overlying cone and to nearby cones via telodendria. Although this feedback may be involved in spatiotemporal and color coding in the outer retina, its role is still the subject of much debate (for critical reviews of this topic see Burkhardt, 1993; Wu, 1992). Another conclusion is that rods should be sensitive to GABA. Such a situation would require diffusion of GABA from horizontal cells to the rods because there is no evidence for contact between GABAergic H1 horizontal cells and rods in the goldfish retina (Marc et al., 1978; Stell and Lightfoot, 1975). This so-called action at a distance is feasible because the high affinity uptake of GABA is not 100% efficient, considering that it is GABAA Receptor Localization in the OPL What are the locations of GABA receptive elements in the fish OPL? With respect to cone pedicles, we suggest that GABA sensitivity is distributed all over the cone pedicle plasma membrane. Intense GABAr-IR was found on pedicle processes that were proximal to the invagination rather than deeper within the invagination near the synaptic ribbons. This is not to say that cone membrane opposing spinules or the lateral face of horizontal cell dendrites is devoid of GABAr-IR. Figures 6B and 8A clearly showed GABAr-IR on cone membrane within the invagination, indicative of GABA sensitivity. The point is that GABA sensitivity may be broadly distributed and even higher at other areas of the pedicle. The cone pedicle comprises a highly enclosed space of interdigitating neuropil, analogous to a glomerulus. It is highly likely that GABA released into this space would affect membrane receptors anywhere GABA RECEPTORS IN PHOTORECEPTORS 51 Fig. 6. GABAr-IR in cone pedicles from light-adapted retinas. GABAr-IR was found on the membrane of synaptic vesicles and on the cone plasma membrane and opposing horizontal cell dendritic spinules (S, arrowheads). There was no relative increase of GABAr-IR on the cone membrane within the invagination opposing the spinules. B is a higher magnification view (rotated 90° counterclockwise) of a portion of the upper horizontal cell dendrite and spinules in A. The arrow in A indicates intense GABAr-IR of a cone pedicle. Sections were counterstained for 10 sec in lead citrate. H, horizontal cell dendrites; S, spinules. Calibration bar 5 0.5 µm (A) or 0.25 µm (B). possible to detect a stimulated release of 3H-GABA as low as 0.03% and as high as 20% of the total 3H-GABA content of the fish retina (Yazulla, 1983, 1985). Although the small size of goldfish rods has precluded any test of whether rods are GABA-sensitive, we suggest that future studies will find GABA sensitivity of rods. If so, goldfish would differ from turtle, in which isolated red and green cones showed a high sensitivity to GABA, whereas blue cones and rods showed a low sensitivity (Tachibana and Kaneko, 1984). that GABAr-IR occurs on the cytoplasmic surface of both the plasma membrane and synaptic vesicles, which would be expected for vesicles formed by endocytosis. Also, the distribution of GABAr-IR synaptic vesicles throughout the synaptic terminals is similar to that reported for HRP-containing synaptic vesicles, formed by endocytosis during darkness, in frog, skate, and turtle photoreceptor terminals (Ripps et al., 1976; Schacher et al., 1974, 1976; Schaeffer and Raviola, 1978). Perhaps the rapid turnover of plasma membrane, due to vesicular fusion, within receptor invaginations precludes the concentration of GABA receptors directly opposing the horizontal cell dendrites. In this way GABAA receptors could be located on plasma membranes that were less active endocytotically in order to present a more stable environment for GABA reception by photoreceptors. Unfortunately, this explanation provides no hint as to the site GABA release from horizontal cells. GABAr-IR and Synaptic Vesicle Membrane Consistent with our previous report (Yazulla et al., 1989), we found that GABAr-IR was associated with the membrane of synaptic vesicles. Since the synaptic vesicles appeared clear and not filled with reaction product, we conclude that GABAr-IR was associated with the cytoplasmic surface of the synaptic vesicle membrane. The plasma membrane of photoreceptor terminals undergoes considerable endocytosis, probably to retrieve membrane added by vesicular exocytosis (Ripps and Chappell, 1991; Ripps et al., 1976; Schacher et al., 1976; Schaeffer and Raviola, 1975). This endocytotic activity could explain the presence of GABAr-IR synaptic vesicles in view of the observation Adaptation and Synaptic Vesicle Density We found that light and dark adaptation affected the density of synaptic vesicles of both rods and cones in contrast to previous studies on teleost fish retinas that have not reported such effects (e.g., Kohler et al., 1990; 52 S. YAZULLA AND K.M. STUDHOLME Fig. 7. GABAr-IR in cone pedicles from light-adapted retinas. The most intense GABAr-IR was in patches on the perimeter of the cone plasma membrane (B, small arrows), on cone processes proximal to the pedicle (large arrows), and on presumed cone telodendria (T, arrows). There was no relative increase of GABAr-IR on the cone membrane within the invagination opposing the spinules (arrowheads). Sections were not counterstained. Calibration bars 5 0.5 µm. Wagner, 1980; Weiler and Wagner, 1984). This discrepancy is puzzling, considering that we have corroborated their observations on horizontal cell spinule formation and convolutions in cone terminals. One possible explanation is that these previous studies employed 2.5% glutaraldehyde in the fixative, whereas we used 0.15% glutaraldehyde prior to overnight fixation in 4% paraformaldehyde at pH 10.4. We have found that our procedure maximizes tissue preservation that is compatible with preembedding immunocytochemistry for a wide variety of antigens, including enzymes as well as transmitter/glutaraldehyde conjugates (Eldred et al., 1983; Yazulla and Studholme, 1991). Since our goal was to investigate the source of the apparent changes in intracellular GABAr-IR with adaptation, all retinas were fixed identically in low glutaraldehyde whether or not they were processed for GABAr-IR. The answer to the discrepancy may lie in a parametric study on the effects of fixation on synaptic vesicle density. This, however, is beyond the scope of the present investigation that was directed towards GABAr-IR and horizontal cell spinules. Given these qualifications, some comments as to possible implications of these data are warranted. Goldfish cones showed a dramatic 29% increase in synaptic vesicle density with 3 h of dark adaptation. The plasma membrane of goldfish cone terminals would be expected to show an increase in surface area with light adaptation due to invagination by horizontal cell spinules (Raynauld et al., 1979; Wagner, 1980). With dark adaptation and retraction of the spinules, there may be an increase in the recovery of cone plasma membrane (Wagner, 1980) and a consequent increase in synaptic vesicle density. Also, cone neurotransmission in teleost fish is suppressed markedly during prolonged dark adaptation (e.g., Raynauld et al., 1979; Yang et al., 1988), perhaps resulting in an excess accumulation of synaptic vesicles. Reported effects of lighting conditions on synaptic vesicles have been inconsistent and appear to be speciesdependent. An early report of a dark-induced decrease in synaptic vesicle size in rat retina (De Robertis and Franchi, 1958) was not corroborated (Mountford, 1963; Cragg, 1972), and more recent studies in rat did not address specifically effects on synaptic vesicle density (Brandon and Lam, 1983; Case and Plummer, 1993). Studies in amphibian retinae also have yielded varying results. Ball and Dickson (1983) reported, in newt GABA RECEPTORS IN PHOTORECEPTORS Fig. 8. GABAr-IR in cone pedicles from dark-adapted retinas. GABAr-IR is found on the membrane of synaptic vesicles and on the cone plasma membrane. H, horizontal cell dendrites. A: GABAr-IR is very light on cone membrane in the region of the synaptic ribbon but appears to get denser with increasing distance from the ribbon (arrowheads). Calibration bar 5 0.25 µm. B: Dense GABAr-IR was found on cone membrane away from the synaptic ribbon but adjacent 53 to horizontal cell dendrites (arrowheads). Notice that this portion of cone pedicle appears to encapsulate the invaginating processes. Calibration bar 5 0.5 µm. C: Dense GABAr-IR also was observed on presumed cone telodendria (T, arrowheads). Bipolar cell dendrites (B) make basal contacts with cone terminals and are not GABAr-IR. Calibration bar 5 0.5 µm. 54 S. YAZULLA AND K.M. STUDHOLME Fig. 9. GABAr-IR in rod spherules from light-adapted (A) and dark-adapted (B) retinas. GABAr-IR appears on the membrane of synaptic vesicles and on the plasma membrane. There does not appear to be any differential staining with GABAr-IR inside or outside the invagination or with states of adaptation. H, horizontal cell dendrites. Calibration bars 5 0.25 µm. 55 GABA RECEPTORS IN PHOTORECEPTORS retina, that synaptic vesicle density increased during the day cycle, but this increase could be accounted for by a decrease in synaptic terminal volume because the total number of synaptic vesicles per terminal did not change. Schacher et al. (1976) found no effects of light adaptation on synaptic vesicle density in frog photoreceptors. However, dense-core vesicles appeared and increased in frequency in Xenopus photoreceptors following prolonged light (Monaghan and Osborne, 1975) and in newt photoreceptors during the day cycle (Ball and Dickson, 1983); this finding was interpreted to represent ‘‘supercharging’’ of vesicles with a neurotransmitter during a period of low transmitter release. Rod spherules showed a modest (17%) reduction in synaptic vesicle density with dark adaptation. Perhaps this reduction was due to increased transmitter release by rods in the dark. Rod spherules are not invaginated by horizontal cell spinules, unlike cone pedicles. Rather, the reverse situation occurs in that processes of rod spherules protrude into horizontal cell processes during dark adaptation (Brandon and Lam, 1983). Thus, reduced synaptic vesicle density in goldfish cones and rods is correlated with increased membrane convolutions. This is the opposite of that reported for the newt (Ball and Dickson, 1983), in which increased synaptic vesicle density was an epiphenomenon of decreased volume. Light-dependent plasticity of fish photoreceptor terminals includes not only invagination by horizontal cell spinules, changes in synaptic vesicle density, and changes in surface contour but also effects on synaptic ribbons. The length and number of synaptic ribbons per cone terminal are reduced by dark adaptation and during the night phase (Wagner, 1973, 1975; Wagner and Ali, 1977). We did not quantify effects on synaptic ribbons. However, it is evident from the representative micrographs in Figure 3 that synaptic ribbons in lightadapted cones (Fig. 3A) were longer than in darkadapted cones (Fig. 3B). This corroborative finding means that during periods of inactivity (prolonged darkness) (Raynauld et al., 1979; Yang et al., 1988), there is a concomitant decrease in synaptic ribbon length and an increase in synaptic vesicle density in goldfish cones. Summary and Conclusions We have shown that GABAr-IR on the plasma membrane of photoreceptor terminals appears independent of the state of light adaptation and is not concentrated within the invagination, the site of hypothesized GABA release. We conclude that GABA sensitivity of photoreceptors is broadly distributed on the synaptic terminal and is not affected by adaptation. We suggest that GABA released into the invagination of a pedicle can affect not only the overlying cone but also nearby cones via action on telodendria. The proposed widespread distribution of GABA sensitivity raises questions as to the role of horizontal cell spinules in GABA release. Horizontal cell spinules have been studied intensively over the last few years, yet their function remains controversial (i.e., Burkhardt, 1993; Downing and Djamgoz, 1989; Wagner and Djamgoz, 1993). Perhaps the spinules are involved in the efficacy of signal transmission from horizontal cells to cones, without being the actual site of GABA release. An analogous situation may be found in the ampulla of Lorenzini of elasmobranch fishes, in which transmission from firstto second-order neurons is correlated with the degree of invagination between the processes (Fields and Ellisman, 1985). The invaginating spinules may serve a similar role for horizontal cell to photoreceptor transmission, although the mechanism for this action remains to be determined. ACKNOWLEDGMENTS We thank Dr. A.L. De Blas, who generously supplied us with mAb 62-3G1 against the GABA receptor. This study was supported by NIH grant EYO1682 to S.Y. REFERENCES Ali, M.A. (1971) Les réponses rétinomotrices: caractères et mécanismes. Vision Res., 11:1225–1288. Ayoub, G.S., and Lam, D.M.K. (1984) The release of g-aminobutyric acid from horizontal cells of the goldfish (Carassius auratus) retina. J. Physiol. (Lond.), 355:191–214. Ball, A.K., and Dickson, D.H. (1983) Diurnal variations in the photoreceptor synaptic terminals of the newt retina. Amer. J. Anat. 168:305– 320. Brandon, C., and Lam, D.M. (1983) The ultrastructure of rat rod synaptic terminals: Effects of dark-adaptation. J. Comp. Neurol., 217:167–175. Burkhardt, D.A. (1993) Synaptic feedback, depolarization, and color opponency in cone photoreceptors. Vis. Neurosci., 10:981–989. Burnside, B., and Nagle, B. (1983) Retinomotor movements of photoreceptors and retinal pigment epithelium: Mechanisms and regulation. Prog. Retinal Res., 27:67–109. Case, C.P., and Plummer, C.J. (1993) Changing the light intensity of the visual environment results in large differences in numbers of synapses and in photoreceptor size in the retina of the young adult rat. Neuroscience, 55:653–666. Cragg, B.G. (1972) Plasticity of synapses. In: The Structure and Function of Nervous Tissue, Vol. 4. G.H. Bourne, ed. Academic Press, New York, pp. 1–60. De Blas, A.L., Vitorica, J., and Friedrich, P. (1988) Localization of the GABAA receptor in the rat brain with a monoclonal antibody to the 57,000 Mr peptide of the GABAA receptor/benzodiazepine receptor/ Cl2 channel complex. J. Neurosci., 8:602–614. De Robertis, E., and Franchi, C.M. (1958) Electron microscopic observations of synaptic vesicles in synapses of retinal rods and cones. J. Biophys. Biochem. Cytol., 3:307–317. Djamgoz, M.B.A., and Ruddock, K.H. (1979) Effects of picrotoxin and strychnine on fish retinal S potentials: Evidence for inhibitory control of depolarizing responses. Neurosci. Lett., 12:329–334. Djamgoz, M.B.A., Downing, J.E.G., Kirsch, M., Prince, D.J., and Wagner, H.-J. (1988) Plasticity of cone horizontal cell functioning in cyprinid fish retina: Effects of background illumination of moderate intensity. J. Neurocytol., 17:701–710. Downing, J.E.G., and Djamgoz, M.B.A. (1989) Quantitative analysis of cone photoreceptor-horizontal cell connectivity patterns in the retina of a cyprinid fish: Electron microscopy of functionally identified and HRP-labelled horizontal cells. J. Comp. Neurol., 289:537–553. Eldred, W.D., Zucker, C., Karten, H.J., and Yazulla, S. (1983) Comparison of fixation and penetration enhancement techniques for use in ultrastructural immunocytochemistry. J. Histochem. Cytochem. 31:285–292. Ewert, M., De Blas, A.L., Möhler, H., and Seeburg, P.H. (1992) A prominent epitope on GABAA receptors is recognized by two different monoclonal antibodies. Brain Res., 569:57–62. Fields, R.D., and Ellisman, M.H. (1985) Synaptic morphology and differences in sensitivity. Science, 228:197–199. Heidelberger, R., and Matthews, G. (1991) Inhibition of calcium influx and calcium current by gamma-aminobutyric acid in single synaptic terminals. Proc. Natl. Acad. Sci. U. S. A. 88:7135–7139. Kohler, K., Kolbinger, W., Kurz-Isler, G., and Weiler, R. (1990) Endogenous dopamine and cyclic events in the fish retina, II: Correlation of retinomotor movement, spinule formation, and connexon density of gap junctions with dopamine activity during light/dark cycles. Vis. Neurosci., 5:417–428. Kraft, T.W., and Burkhardt, D.A. (1986) Telodendrites of cone photoreceptors: Structure and probable function. J. Comp. Neurol., 249:13– 27. 56 S. YAZULLA AND K.M. STUDHOLME Lin, Z.S., and Yazulla, S. (1994) Heterogeneity of GABAA receptors in goldfish retina. J. Comp. Neurol., 345:429–439. Marc, R.E., and Liu, W.L.S. (1984) Horizontal cell synapses onto glycine-accumulating interplexiform cells. Nature, 312:266–268. Marc, R.E., Stell, W.K., Bok, D., and Lam, D.M.K. (1978) GABAergic pathways in the goldfish retina. J. Comp. Neurol., 182:221–246. Monaghan, P., and Osborne, M.P. (1975) Light-induced formation of dense-core vesicles in rod photoreceptors in retina of Xenopus laevis. Nature, 256:586–587. Mountford, S. (1963) Effects of light and dark adaptation on the vesicle population of receptor-bipolar synapses. J. Ultrastruct. Res., 9:403– 418. Murakami, M., Shimoda, Y., Nakatani, K., Miyachi, E., and Watanabe, S. (1982a) GABA-mediated negative feedback and color opponency in carp retina. Jpn. J. Physiol., 32:927–935. Murakami, M., Shimoda, Y., Nakatani, K., Miyachi, E., and Watanabe, S. (1982b) GABA-mediated negative feedback from horizontal cells to cones in carp retina. Jpn. J. Physiol., 32:911–926. Park, D., and De Blas, A.L. (1991) Peptide subunits of g-aminobutyric acidA/benzodiazepine receptors from bovine cerebral cortex. J. Neurochem., 56:1971–1979. Ramon y Cajal, S. (1933) The Structure of the Retina. S.A. Thorpe and M. Glickstein, compilers and translators, 1972. Charles C. Thomas Co., Springfield, IL. Raynauld, J.-P., Laviolette, J.R., and Wagner, H.-J. (1979) Goldfish retina: A correlate between cone activity and morphology of the horizontal cell in cone pedicles. Science, 204:1436–1438. Ripps, H., and Chappell, R.L. (1991) Ultrastructural and electrophysiological changes associated with K1-evoked release of neurotransmitter at the synaptic terminals of skate photoreceptors. Vis. Neurosci., 7:597–610. Ripps, H., Shakib, H.M., and MacDonald, E.D. (1976) Peroxidase uptake by photoreceptor terminals of the skate retina. J. Cell Biol., 70:86–96. Schacher, S.M., Holtzman, E., and Hood, D.C. (1974) Uptake of horseradish peroxidase by frog photoreceptor synapses in the dark and the light. Nature, 249:261–263. Schacher, S., Holtzman, E., and Hood, D.C. (1976) Synaptic activity of frog retinal photoreceptors: A peroxidase uptake study. J. Cell Biol., 70:178–192. Schaeffer, S.F., and Raviola, E. (1975) Ultrastructural analysis of functional changes in the synaptic endings of turtle cone cells. In: Cold Spring Harbor Symposia on Quantitative Biology, Vol. 40. The Synapse. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, pp. 521–528. Schaeffer, S.F., and Raviola, E. (1978) Membrane recycling in the cone cell endings of the turtle retina. J. Cell Biol., 79:802–825. Scholes, J.H. (1975) Colour receptors, and their synaptic connexions in the retina of a cyprinid fish. Philos. Trans. R. Soc. Lond. [Biol.], 270:61–118. Schwartz, E.A. (1982) Calcium-independent release of GABA from isolated horizontal cells of the toad retina. J. Physiol. (Lond.), 323:211–227. Schwartz, E.A. (1987) Depolarization without calcium can release g-aminobutyric acid from a retinal neuron. Science, 238:350–355. Stell, W.K. (1967) The structure and relationships of horizontal cells and photoreceptor-bipolar synaptic complexes in goldfish retina. Am. J. Anat., 120:401–424. Stell, W.K. (1980) Photoreceptor-specific synaptic pathways in goldfish retina: A world of colour, a wealth of connections. In: Colour Vision Deficiencies, Vol. 5. G. Verriest, ed. Adam Hilger, Bristol, pp. 1–14. Stell, W.K., and Lightfoot, D.O. (1975) Color-specific interconnections of cones and horizontal cells in the retina of the goldfish. J. Comp. Neurol., 159:473–502. Stell, W.K., Lightfoot, D.O., Wheeler, T.G., and Leeper, H.F. (1975) Goldfish retina: Functional polarization of cone horizontal cell dendrites and synapses. Science, 190:989–990. Tachibana, M., and Kaneko, A. (1984) g-aminobutyric acid acts at axon terminals of turtle photoreceptors: Difference in sensitivity among cell types. Proc. Natl. Acad. Sci. U. S. A. 81:7961–7964. Tachibana, M., and Kaneko, A. (1987) g-aminobutyric acid exerts a local inhibitory action on the axon terminal of bipolar cells: Evidence for negative feedback from amacrine cells. Proc. Natl. Acad. Sci. U. S. A. 84:3501–3505. Toyoda, J., and Fujimoto, M. (1983) Analyses of neural mechanisms mediating the effect of horizontal cell polarization. Vision Res., 23:1143–1150. Vitorica, J., Park, D., Chin, G., and De Blas, A.L. (1988) Monoclonal antibodies and conventional antisera to the GABAA receptor/ benzodiazepine receptor/Cl2 channel complex. J. Neurosci., 8:615– 622. Wagner, H.-J. (1973) Darkness-induced reduction of the number of synaptic ribbons in fish retina. Nature New Biology, 246:53–55. Wagner, H.-J. (1975) Quantitative changes of synaptic ribbons in the cone pedicles of Nannacara: Light dependent or governed by a circadian rhythm? In: Vision in Fishes: New Approaches in Research. M.A. Ali, ed. Plenum Press, New York, pp. 679–686. Wagner, H.-J. (1980) Light-dependent plasticity of the morphology of horizontal cell terminals in cone pedicles of fish retinas. J. Neurocytol., 9:573–590. Wagner, H.-J., and Ali, M.A. (1977) Cone synaptic ribbons and retinomotor changes in the brook trout, Salvelinus fontinalis (Salmonidae, Teleostei), under various experimental conditions. Can. J. Zool., 55:1684–1691. Wagner, H.-J., and Djamgoz, M.B.A. (1993) Spinules: A case for retinal synaptic plasticity. Trends Neurosci., 16:201–206. Weiler, R., and Wagner, H.-J. (1984) Light-dependent change of cone–horizontal cell interactions in carp retina. Brain Res., 298:1–9. Witkovsky, P., and Dowling, J.E. (1969) Synaptic relationships in the plexiform layers of carp retina. Z. Zellforsh., 100:60–82. Witkovsky, P., Shakib, M., and Ripps, H. (1974) Interreceptoral junctions in the teleost retina. Invest. Ophthalmol., 13:996–1009. Wu, S.M. (1992) Feedback connections and operation of the outer plexiform layer of the retina. Curr. Opin. Neurobiol., 2:462–468. Yang, X.-L., Tornqvist, K., and Dowling, J.E. (1988) Modulation of cone horizontal cell activity in the teleost fish retina. I. Effects of prolonged darkness and background illumination on light responsiveness. J. Neurosci., 8:2259–2268. Yazulla, S. (1983) Stimulation of GABA release from retinal horizontal cells by potassium and acidic amino acid agonists. Brain Res., 275:61–74. Yazulla, S. (1985) Evoked efflux of 3H-GABA from goldfish retina in the dark. Brain Res., 325:171–180. Yazulla, S. (1986) GABAergic mechanisms in the retina. Prog. Retinal Res., 5:1–52. Yazulla, S., and Kleinschmidt, J. (1982) Dopamine blocks carriermediated release of GABA from retinal horizontal cells. Brain Res., 233:211–215. Yazulla, S., and Kleinschmidt, J. (1983) Carrier-mediated release of GABA from retinal horizontal cells. Brain Res., 263:63–75. Yazulla, S., and Studholme, K.M. (1991) Glycine-receptor immunoreactivity in retinal bipolar cells is postsynaptic to glycinergic and GABAergic amacrine cell synapses. J. Comp. Neurol., 310:11–20. Yazulla, S., Studholme, K.M., and Wu, J. (1987) GABAergic input to the synaptic terminals of mb1 bipolar cells in the goldfish retina. Brain Res., 411:400–405. Yazulla, S., Studholme, K.M., Vitorica, J., and De Blas, A.L. (1989) Immunocytochemical localization of GABAA receptors in goldfish and chicken retinas. J. Comp. Neurol., 280:15–26.