THE JOURNAL OF COMPARATIVE NEUROLOGY 296~173-178(1990)
GABA and Tyrosine Hydroxylase Immunocytochemistry Reveal Different Patterns of Colocalization in Retinal Neurons of Various Vertebrates I. FWLLE AND H.-J. WAGNER Institut fur Anatomie und Zellbiologie, Philipps Universitat Marburg, D-3550 Marburg, West Germany
ABSTRACT Colocalization of GABA- and tyrosine hydroxylase-like immunoreactivity was studied in the retinae of various vertebrate species in order to ascertain whether the presumed coexistence of GABA and dopamine, reported earlier for mammals (Kosaka et al.: Exp. Brain Res. 66:191-210, '87; Wassle and Chun: J. Neurosci. 8:3383-3394, '88) is a common phenomenon. GABA-immunopositive cells constituted a separate population from tyrosine hydroxylasepositive cells in fish and amphibians, whilst in higher-i.e., amniote-vertebrates, such as reptiles, birds, and mammals, all dopaminergic cells contained GABA-like immunoreactivity. No clear correlation was found between the type of dopaminergic cell (amacrine/interplexiform) and the presence or absence of colocalization. Key words: neurotransmitters, dopamine, amacrine cells, interplexiform cells
Colocalization of classical neurotransmitters and neuropeptides is a widely accepted feature of neurons in many parts of the PNS and the CNS. More uncommon is the colocalization of two classical neurotransmitters. In the retina, the simultaneous presence of GABA and acetylcholine in rabbit amacrine cells has recently been demonstrated (Brecha et al., '88; Vaney and Young, '88). Furthermore, colocalization of GABA and dopamine has been shown in cat retina (Wassle and Chun, '88) and several brain regions (Kosaka et al., '87) of the rat. GABAergic neurons are present in all retinae studied so far in high density, making up 40-4576 of the neurons in the amacrine cell layer. It may therefore be speculated that they do not represent a morphologically and functionally homogeneous group of neurons. Colocalization with other transmitters or neuropeptides may therefore constitute a valuable tool further to characterize the large population of GABAergic cells. As shown by Caldwell et al. ('78) and Caldwell and Daw ('78), GABAergic neurons contribute to the coding of complex receptive field properties of ganglion cells. Dopaminergic neurons possess large somata located in the inner nuclear layer (INL) at the border of the inner plexiform layer (IPL). They are present in low density and are characterized by large dendritic fields. Dopamine is capable of suppressing the spontaneous activity and light response in all ganglion cells encountered and of shifting their
o 1990 WILEY-LISS, INC.
centersurround balance in favour of the center (Thier and Alder '84). On the basis of their observations in the rat brain and retina, Kosaka et al. ('87) proposed that the colocalization of GABA and dopamine is a common phenomenon in the central nervous system, especially in intrinsic neurons that form short and intermediate-length local circuits. This suggestion is supported by the findings of W a d e and Chun ('88) for the cat retina. In the present study we concentrated on the retina in order to ascertain whether colocalization of GABA and dopamine is a phenomenon common to all vertebrates. We used postembedding immunocytochemistry on consecutive semithin sections of fish, amphibians, reptiles, birds, and mammals and compared the respective patterns of labelling. Our results suggest that colocalization of GABA and dopamine is restricted to amniote vertebrates, such as reptiles, birds, and mammals.
MATERIALS AND METHODS A t least two specimens of each of the following species were used: roach (Rutilus rutilus), glass catfish (Kryptopterus bicirrhis), clawed frog (Xenopus Zaeuis), turtle
Accepted December 19,1989.
I. WULLE AND H.-J. WAGNER
174
Figures 1-10
PATTERNS OF GABA AND TH COLOCALIZATION IN VERTEBRATE RETINAE (Pseudemys scripta elegans), chicken, and mouse. Fish, frogs, and turtles were decapitated after deep anesthesia in MS 222; chick and mice were sacrificed by an overdose of nembutal. After enucleation of the eyes, the retinae were isolated and fixed for 2 hours in 4% paraformaldehyde, 0.2% glutaraldehyde, 15% saturated picric acid in 0.06 M phosphate buffer (pH 7.4) for 2 hours at room temperature (RT). They were subsequently washed in 0.06 M phosphate buffer (pH 7.4) for 3 hours before the tissue was embedded in Epon 812. Semithin radial or tangential sections (0.5 pm) were dried on gelatinized coverslips (diameter 8 mm) for 2 hours at 45OC. Consecutive sections were then incubated either with a GABA-antiserum or a tyrosine hydroxylase (TH) antiserum according to the method of Sternberger ('79). In the first step, the Epon was removed from the tissue by etching the sections for 15 minutes in a saturated solution of sodium hydroxide in ethanol. Subsequently, the preparations were washed in 100% ethanol and 0.2 M Tris-HC1 buffer (pH 7.4) before preincubation with 0.2 M Tris-HC1, 10% normal goat serum (NGS) and 0.5% Triton X-100. Alternating successive sections were then incubated with the antisera in the following way: Antisera were diluted in 0.2 M Tris, 3% NGS, and 0.5% Triton-X-100; after each incubation step the sections were washed with Tris-HC1. First antiserum: Rabbit-anti-GABA; 1:1,000-1:4,000 (a detailed description of the characterization of this antiserum is in preparation [Wulle and Schnitzer, in prep.]; results obtained with the same antiserum have previously been published by Wassle and Chun "881 and Wassle et al. "891); alternatively, rabbit-anti-TH (1:120; Eugene Tech Corp., Allendale) at 4°C overnight. Second antiserum: Goat-anti-rabbit (1:20, Sigma) 1hour at RT. Third antiserum: PAP (1:50; Sigma) 1 hour at RT. The reaction product was visualized by using 0.05 % diaminobenzidine, 0.01% H,O,. After further washing, the sections were counterstained with toluidine blue, then transferred to and fixed on microscopic slides for evaluation. Controls were performed by omitting the primary antisera and including all subsequent steps as outlined above; in these preparations, the cone ellipsoids displayed some unspecific label, but no staining of the neural components of the retina was observed.
Figs. 1-10. Light micrographs of semithin (0.5 pm) sections of the retinae of some anamniote vertebrates, arranged in pairs of consecutive sections after postembedding immunostaining with antisera to tyrosine hydroxylase (1, 3, 5, 7, 9) or GABA (2, 4, 6, 8, 10). Arrows in the anti-GABA-stained figures indicate the position of the TH-like immunopositive somata. In these species no colocalization is observed. x 550. Scale bar valid for all figures: 20 pm. Figs. 1, 2. Tangential sections of the junction of the inner nuclear and plexiform layers of the roach retina. Figs. 3,4. Radial sections of the roach retina. Figs. 5,6. Radial sections of the glass catfish retina. Figs. 7,8.
Radial sections of the Xenopus retina.
Figs. 9, 10. Tangential sections of the Xenopus retina at the junction of the inner nuclear and plexiform layers.
175
RESULTS In this study, we examined two teleost species routinely studied in our laboratory, and representatives of amphibians (Xenopus),widely used for electrophysiological studies, reptiles (turtle), birds (chick), and mammals (mouse) in order to investigate the colocalization of GABA-like and TH-like immunoreactivity in retinal neurons. T H is the rate-limiting enzyme of catecholamine synthesis, and the antiserum against it was used as a marker for dopaminergic neurones, since dopamine is the only catecholamine present in detectable amounts in vertebrate retinae. For the demonstration of GABAergic neurons, we used an antiserum directed to GABA that was conjugated to bovine serum albumin (Wulle and Schnitzer, in prep.). In each retina, five pairs of 0.5 pm thick, consecutive radial sections of about 2.5-3 mm total length were recorded photographically and by line drawings of perikarya. The patterns of labelling were matched by superposition of a line drawing and a light micrograph. In this way, perikarya of all labelled cells could easily be identified in both sections. Each radial section contained between three and five cells bodies with positive T H immunoreactivity. Since, in the mouse retina, the density of TH-positive cells was substantially lower in radial sections, only pairs of tangential sections comprising INL and IPL were prepared and analysed in this species. In all retinae studied, GABA immunoreactivity was restricted to perikarya localized predominantly in the INL, at the border of the inner and outer plexiform layers. Cells lining the IPL could be related to and identified as amacrine and/or interplexiform cells, whilst perikarya next to the outer plexiform layer were readily recognized as horizontal cells. In addition, GABA-reactive perikarya of displaced amacrine cells were encountered in the ganglion cell layer. Processes of GABA-positive cells formed a very dense network that almost completely filled the IPL with an all but homogeneous pattern of immunoreactive stain. The TH-positive retinal neurons in all vertebrate species investigated were characterized by large somata located in the inner INL, at the border of the IPL. In whole-mount preparations (not shown) we found these neurons to be distributed evenly and in low densities all over the retinae. Processes of TH-immunoreactive cells were much sparser than for GABA-positive cells and, in most species, could be identified as continuous fibers or rows of individual varicosities. Most often, stained fibers were found in sublayer 1and some deeper layers of the IPL which were, however, not further characterized. In the outer plexiform layer (OPL) the most prominent plexus of TH-immunoreactive fibers was observed in the roach retina (Fig. 3). In the catfish, a less conspicuous plexus of such fibers was present in the OPL (Fig. 5). Therefore, in these two teleost species THimmunopositive cells were identified as interplexiform cells. In the other vertebrate species studied in whole-mount preparations (not shown) we found several isolated immunostained fibers in the OPL; however, we cannot be certain whether TH-immunoreactive neurons in these cases are interplexiform cells or rather, as thought traditionally, amacrine cells. The assessment of colocalization of GABA- and TH-like immunoreactivity was performed by superimposing transparent drawings of anti-TH stained preparations onto light micrographs of anti-GABA labelled sections. For greater
176
I. WULLE AND H.-J. WAGNER
Figures 11-16
PATTERNS OF GABA AND TH COLOCALIZATION I N VERTEBRATE RETINAE
177
and a primate (Macaca nemestrina, Nguyen-Legros and Simon, '88). By contrast, no colocalization was found in neurons of fish and Xenopus retinae. This indicates that colocalization of GABA and T H is not a general phenomenon among vertebrates; rather, it is present in retinal neurons of higher, i.e., amniote vertebrates and is lacking in retinal neurons of lower, anamniote vertebrates. The respective individual functions of dopamine and GABA in retinal signal processing have been extensively studied in various vertebrate species. As reviewed by Kamp ('85) and Besharse et al. ('88), dopamine plays an essential role in controlling the adaptational processes preparing the retina to function under photopic as well as scotopic conditions. In the outer retina of fish, dopamine acts on horizontal cells to modulate intercellular coupling (Teranishi et al., '83) and on photoreceptors to induce light adaptational retinomotor movements (Dearry and Burnside, '86), while in the inner retina, dopaminergic neurons are involved in local circuits with bipolar and amacrine cells (Yazulla and Zucker, '88). In turtle, dopamine similarly decreased the permeability of horizontal cell gap junctions (Piccolino et al., '84, '87). In mammals, on the other hand, effects of dopamine on ganglion cell responses were mostly studied; receptive field properties were made more sustained and spatial summation characteristics more linear; interestingly, off-centre cells appeared to be more strongly affected by dopamine depletion than on-centre cells (Maguire and Smith, '85; Jensen and Daw, '86). DISCUSSION As for the function of GABA, it appears at the same time We used an antiserum directed to T H and indirect to be far more complex, yet less well defined than that of immunocytochemistry to visualize the dopaminergic neu- dopamine: In a general way, GABA is suggested to contribrons in the retinae of various vertebrates. Our results are in ute to the organization of complex receptive fields of agreement with previous observations based on formalde- ganglion cells (Caldwell and Daw, '78; Caldwell et al., '78; hyde-induced fluorescence (Ehinger '66, '76; Dowling and Morgan, '85). In the IPL, this is thought to be accomplished Ehinger, '75, '78) and with biochemical analyses showing through the extensive plexus of GABA-like immunoreactive that dopamine is the only neuronal catecholamine present amacrine cell processes surrounding the bipolar terminals; in the retina (for review see Kamp, '85). As for the visualiza- in the outer retina, by GABAergic L-type horizontal cells tion of GABAergic neurons, the antiserum was raised against feeding back onto photoreceptors, thus also acting on the a glutaraldehyde-based conjugate of GABA and bovine size and chromatic organization of receptive fields of bipolar serum albumine. Specificity tests as well as an evaluation of and ganglion cells. There are intricate interrelationships between dopamine the staining pattern in view of previous findings using autoradiography and glutamate decarboxylase (GAD)immu- and GABA in some vertebrate retinae. In fish, GABAergic nocytochemistry are outlined by Wassle and Chun ('88) and amacrine cells inhibit the activity of dopaminergic interplexiform cells (Ishita et al., '88); conversely, dopamine is Wassle et al. ('89). Our observations show that GABA-like and TH-like capable of inhibiting the dark induced release of GABA immunoreactivity are colocalized in the retinal neurons of from H1 horizontal cells (Yazulla '85). While this release turtle, chick, and mouse. As far as mammals are concerned, appears to be Ca++-independent,the Ca++-dependentliberthis finding is supported by previous immunocytochemical ation of GABA is enhanced by dopamine (O'Brien and findings suggesting the colocalization of GABA and dopa- Dowling, '85). By contrast, no effect of dopamine on GABA mine in retinal neurons of the cat (Wassle and Chun, '88) release has been found in chick (Nistico et al., '83). Finally, GABA released from amacrine cells in rat has been observed to inhibit both synthesis and release of dopamine (Pro11and Morgan, '83). In summary, however, no difference in these Figs. 11-16. Light micrographs of semithin (0.5 pm) sections of the pharmacological or physiological effects are apparent which retinae of some amniote vertebrates, arranged in pairs of consecutive might be related to the different patterns of colocalization sections after postembedding immunostaining with antisera to tyrosine observed here. hydroxylase (11,13,15) or GABA (12,14,16).Arrows in the anti-GABAThis situation is reminiscent of the interactions between stained figures indicate the position of the TH-like immunopositive GABA and acetylcholine which also colocalize in a specific somata. In these species colocalization of the two markers is observed. type of amacrine cell (Brecha et al., '88; Vaney and Young, x 550. Scale bar valid for all figures: 20 pm. '88). Both GABA and acetylcholine are co-released from Figs. 11,12. Radial sections of the turtle retina. these neurons (Masland et al., '89). Acetylcholine release was optimally stimulated by flashing light or moving stimFigs. 13,14. Radial sections of the chick retina. uli, while GABA release was not affected by this paradigm. GABA release was Ca++-independentwhereas acetylcholine Figs. 15, 16. Tangential section through the inner nuclear layer of was liberated in a Ca++-dependent manner. They also the mouse retina.
clarity we concentrated on the perikarya which could be unequivocally identified in each pair of neighbouring sections. In the figures, the observations are presented as pairs of light micrographs where cell bodies exhibiting colocalization of TH- and GABA-like reaction product (or the lack of it) are marked with arrows in the anti-GABA-stained preparations. Figures 1-10 show the situation in anamniote vertebrates, i.e., fish and amphibians; in all three cases, it is obvious that the neurons containing TH-like immunoreactivity (Figs. 1, 3 , 5 , 7 , 9 ) are not labelled by GABA-antibodies (Figs. 2 , 4 , 6 , 8,lO). On the other hand, horizontal and amacrine cells are clearly GABA-positive in both species of teleost as well as in Xenopus. This lack of colocalization in fish and amphibia exemplified by these micrographs was also observed in all other pairs of sections not documented here. In higher, amniote vertebrates, by contrast, the comparison of the two labelling patterns shows that perikarya with TH-like immunostaining are also stained with GABA-like immunoreactivity; this applies for the radial sections of turtle and chick retinae (Figs. 11-14) as well as for the tangential views of the mouse IPL (Figs. 15, 16). With regard to the other material analysed it can be stated that we found not a single TH-positive cell (n = 13-19, depending on the species) in these three species that did not also contain GABA positive reaction product.
I. WULLE AND H.-J. WAGNER
178 reported that although both substances were released from the same cells, only acetylcholine was stored in vesicles and liberated by vesicle exocytosis; by contrast, GABA was probably released by a carrier mechanism. It was speculated that the release of two transmitters from the same cell could serve to specifically refine and shape the pre- and/or postsynaptic response. On a more general level, cholinergic starburst amacrine cells were shown to be tonically inhibited by endogenous GABA (Agardh, '86). It is unknown whether two neuroactive substances contained within the same neuron are directed to different targets or whether they may act on the same postsynaptic site. A possible mechanism for the action of two conventional neurotransmitters on a single postsynaptic cell was proposed recently by Foster and Kemp ('89). For glutamate and glycine, they suggested an agonistic action leading to the enhancement of the N-methyl-D-aspartate (NMDA)mediated response to glutamate. In the case of a colocalization of a neuropeptide (neuropeptide Y) and epinephrine it has been suggested that the neuropeptide might modulate the action of the small transmitter molecules at the level of the smooth muscle targets (Cooper et al., '86). As for the functional role of colocalization of GABA and dopamine for retinal information processing, there is not even a basis for speculation. Our finding that the coexistence of these two transmitters is not a common phenomenon among vertebrates would indicate that it is not essential to retinal functioning. It is obviously not related to the cell type expressing dopamine since it is observed in both amacrine and interplexiform cells. Rather, the lack of colocalization in anamniote, i.e., aquatic vertebrates and the simultaneous expression of both these transmitters in amniote, i.e., essentially terrestial species may indicate either an evolutionary trend or may be linked to visual or adaptational problems associated with the transition from water to land.
ACKNOWLEDGMENTS We are grateful to M. Schneider for skillful technical help and Dr. M. Kirsch for critically reading and commenting on the manuscript. The financial support by the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
LITERATURE CITED Agardh, E. (1986) GABA antagonists increase the release of 3H acetylcholine from the chick retina. Acta Physiol. Scand. 126r33-38. Besharse, J.C., P.M. Iuvone, and M.E. Pierce (1988) Regulation of rhythmic photoreceptor metabolism: A role for post-receptoral neurons. Progress in Retinal Research 7t21-61. Brecha, N., D. Johnson, L. Peichl, and H. Wassle (1988) Cholinergic amacrine cells of the rabbit retina contain glutamate decarboxylase and gammaaminobutyrate immunoreactivity. Proc. Natl. Acad. Sci. USA 85:61876191. Caldwell, J.H., N.W. Daw, and H.J. Wyatt (1978) Effects of picrotoxin and strychnine on rabbit retinal ganglion cells: Lateral interactions for cells with more complex receptive fields. J. Physiol. 276277-298. Caldwell, J.H., and N.W. Daw (1978) Effects of picrotoxin and strychnine on rabbit retinal ganglion cells: Changes in centre surround receptive fields. J. Physiol. 276t299-310. Cooper, J.R., F.E. Bloom, and R.H. Roth (1986) Neuroactive peptides. The biochemical basis of neuropharmacology, Fifth edition, Oxford University Press 352-393. Dearry, A., and B. Burnside (1986)Dopaminergic regulation of cone retinomotor movement in isolated teleost retinas. I.: Induction of cone contraction is mediated by D2 receptors. J. Neurochem. 46r1006-1021.
Dowling, J.E., and B. Ehinger (1975) Synaptic organization of the amine containing interplexiform cells of the goldfish and cehus monkey retinas. Science, Wash. 188,270-273. Dowling, J.E., and B. Ehinger (1978) The interplexiform cell system. I. Synapses of the dopaminergic neurons of the goldfish retina. Proc. R. SOC. Lond. [Biol.] 201t7-26. Ehinger, B. (1966) Adrenergic retinal neurons. Z. Zellforsch. mikrosk. Anatomie 71:146-152. Ehinger, B. (1976) Weak neuronal accumulation of octopamin in dopaminergic neurons of rabbit retina. Experientia 32690. Foster, A.C., and J.A. Kemp (1989) Glycine maintains excitement. Nature 338:337-338. Ishita, S., K. Negishi, T. Teranishi, Y. Shimada, and S. Kato (1988) GABAergic inhibition of dopaminergic cells of the fish retina: A ('H)dopamine release study with isolated fractions. J. Neurochem. 50:1-6. Jensen, R.J., and N.W. Daw (1986) Effects of dopamine and its agonists and antagonists on the receptive field properties of ganglion cells in the rabbit retina. Neuroscience 17t837-855. Kamp, C.W. (1985) The dopamine system of the retina. In: W.W. Morgan (ed): Retinal Transmitters and Modulators. Models for the Brain Vol. 11. Boca Raton: CRC Press, pp. 1-31. Kosaka, T., K. Kosaka, Y. Hataguchi, I. Nagatsu, J.-Y. Wu, O.P. Ottersen, J. Storm-Mathisen, and K. Hama (1987) Catecholaminergic neurons containing GABA-like and/or glutamic acid decarboxylase-like immunoreactivities in various brain regions of the rat. Exp. Brain Res. 66:191-210. Maguire, G.W., and E.L. Smith, 111 (1985) Cat retinal ganglion cell receptive field alterations after 6-hydroxydopamine induced dopaminergic amacrine lesions. J. Neurophysiol. 53t1431-43. Masland, R., C. Cassidy, and D. O'Malley (1989) The release of acetylcholine and GABA from amacrine cells. In: R. Weiler and N.N. Osborne, eds.: Neurobiology of the inner retina. Berlin, Heidelberg: Springer, p. 15-26. Morgan, W.W. (1985) GABA. A potential neurotransmitter in retina. In: W.W. Morgan (ed) Retinal Transmitters and Modulators: Models for the Brain. Vol. I. Boca Raton: CRC Press, pp. 63-96. Nistico, G., A. De Sarro, G. Urna, A. Pujia, and R. Jentile (1983) Lack of relation between dopaminergic and GAGAergic mechanisms in chick retina. Res. Commun. Chem. Pathol. Pharmacol. 40:29-39. Nguyen-Legros, J., and A. Simon (1988) Dopamine-GABA relationship in the retina of Maeaca nernestrina. Abstr. of the 11th Ann. Meeting of the Eur. Neurosci. Association 425. OBrien, D.R., and J.E. Dowling (1985) Dopaminergic regulation of GABA release from the intact goldfish retina. Brain Res. 360:41-50. Piccolino, M., J. Neyton, and H.M. Gerschenfeld (1984) Decrease of gap junction permeability induced by dopamine and cyclic adenosine 35'monophosphate in horizontal cells of turtle retina. J. Neurosci. 424772477. Piccolino, M., P. Witkovsky, and C. Trimarchi (1987) Dopaminergic mechanisms underlying the reduction of electrical coupling between horizontal cells of the turtle retina induced by d-amphetamine, bicuculline and veratridine. J. Neurosci. 7r2273-2283. Proll, M.A., and W.W. Morgan (1983) Use of butyric acid (GABA) transaminase inhibitors and GABA uptake inhibitors to investigate the influence of GABA neurons on dopamine-containing amacrine cells of the rat retina. J. Pharmacol. Exp. Ther. 227t627-632. Sternberger, L.A. (1979) Immunocytochemistry, 2nd Ed. New York John Wiley and Sons, Inc. Teranishi, T., K. Negishi, and S. Kato (1983) Dopamine modulates S potential amplitude and dye coupling between external horizontal cells in carp retina. Nature (London) 301:243-246. Thier, P., and V. Alder (1984) Action of iontophorectically applied dopamine on cat retinal ganglion cells. Brain Res. 292109-121. Vaney, D.I., and H.M. Young (1988) GABA-like immunoreactivity in cholinergic amacrine cells of the rabbit retina. Brain Res. 438t369-373. Wassle, H., and M.H. Chun (1988) Dopaminergic and indoleamine-accumulating amacrine cells express GABA-like immunoreactivity in the cat retina. J. Neurosci. 8:3383-3394. Wassle, H., U. Grunert, J. Rohrenbeck, and B.B. Boycott (1989) Cortical magnification factor and the ganglion cell density of the primate retina. Nature 341r643-646. Yazulla, S. (1985)Evoked efflux of 3H-GABA from goldfish retina in the dark. Brain Res. 325:171-180. Yazulla, S., and C.L. Zucker (1988) Synaptic organization of dopaminergic interplexiform cells in the goldfish retina. Visual Neuroscience Ir13-29.
GABA and Tyrosine Hydroxylase Immunocytochemistry Reveal Different Patterns of Colocalization in Retinal Neurons of Various Vertebrates I. FWLLE AND H.-J. WAGNER Institut fur Anatomie und Zellbiologie, Philipps Universitat Marburg, D-3550 Marburg, West Germany
ABSTRACT Colocalization of GABA- and tyrosine hydroxylase-like immunoreactivity was studied in the retinae of various vertebrate species in order to ascertain whether the presumed coexistence of GABA and dopamine, reported earlier for mammals (Kosaka et al.: Exp. Brain Res. 66:191-210, '87; Wassle and Chun: J. Neurosci. 8:3383-3394, '88) is a common phenomenon. GABA-immunopositive cells constituted a separate population from tyrosine hydroxylasepositive cells in fish and amphibians, whilst in higher-i.e., amniote-vertebrates, such as reptiles, birds, and mammals, all dopaminergic cells contained GABA-like immunoreactivity. No clear correlation was found between the type of dopaminergic cell (amacrine/interplexiform) and the presence or absence of colocalization. Key words: neurotransmitters, dopamine, amacrine cells, interplexiform cells
Colocalization of classical neurotransmitters and neuropeptides is a widely accepted feature of neurons in many parts of the PNS and the CNS. More uncommon is the colocalization of two classical neurotransmitters. In the retina, the simultaneous presence of GABA and acetylcholine in rabbit amacrine cells has recently been demonstrated (Brecha et al., '88; Vaney and Young, '88). Furthermore, colocalization of GABA and dopamine has been shown in cat retina (Wassle and Chun, '88) and several brain regions (Kosaka et al., '87) of the rat. GABAergic neurons are present in all retinae studied so far in high density, making up 40-4576 of the neurons in the amacrine cell layer. It may therefore be speculated that they do not represent a morphologically and functionally homogeneous group of neurons. Colocalization with other transmitters or neuropeptides may therefore constitute a valuable tool further to characterize the large population of GABAergic cells. As shown by Caldwell et al. ('78) and Caldwell and Daw ('78), GABAergic neurons contribute to the coding of complex receptive field properties of ganglion cells. Dopaminergic neurons possess large somata located in the inner nuclear layer (INL) at the border of the inner plexiform layer (IPL). They are present in low density and are characterized by large dendritic fields. Dopamine is capable of suppressing the spontaneous activity and light response in all ganglion cells encountered and of shifting their
o 1990 WILEY-LISS, INC.
centersurround balance in favour of the center (Thier and Alder '84). On the basis of their observations in the rat brain and retina, Kosaka et al. ('87) proposed that the colocalization of GABA and dopamine is a common phenomenon in the central nervous system, especially in intrinsic neurons that form short and intermediate-length local circuits. This suggestion is supported by the findings of W a d e and Chun ('88) for the cat retina. In the present study we concentrated on the retina in order to ascertain whether colocalization of GABA and dopamine is a phenomenon common to all vertebrates. We used postembedding immunocytochemistry on consecutive semithin sections of fish, amphibians, reptiles, birds, and mammals and compared the respective patterns of labelling. Our results suggest that colocalization of GABA and dopamine is restricted to amniote vertebrates, such as reptiles, birds, and mammals.
MATERIALS AND METHODS A t least two specimens of each of the following species were used: roach (Rutilus rutilus), glass catfish (Kryptopterus bicirrhis), clawed frog (Xenopus Zaeuis), turtle
Accepted December 19,1989.
I. WULLE AND H.-J. WAGNER
174
Figures 1-10
PATTERNS OF GABA AND TH COLOCALIZATION IN VERTEBRATE RETINAE (Pseudemys scripta elegans), chicken, and mouse. Fish, frogs, and turtles were decapitated after deep anesthesia in MS 222; chick and mice were sacrificed by an overdose of nembutal. After enucleation of the eyes, the retinae were isolated and fixed for 2 hours in 4% paraformaldehyde, 0.2% glutaraldehyde, 15% saturated picric acid in 0.06 M phosphate buffer (pH 7.4) for 2 hours at room temperature (RT). They were subsequently washed in 0.06 M phosphate buffer (pH 7.4) for 3 hours before the tissue was embedded in Epon 812. Semithin radial or tangential sections (0.5 pm) were dried on gelatinized coverslips (diameter 8 mm) for 2 hours at 45OC. Consecutive sections were then incubated either with a GABA-antiserum or a tyrosine hydroxylase (TH) antiserum according to the method of Sternberger ('79). In the first step, the Epon was removed from the tissue by etching the sections for 15 minutes in a saturated solution of sodium hydroxide in ethanol. Subsequently, the preparations were washed in 100% ethanol and 0.2 M Tris-HC1 buffer (pH 7.4) before preincubation with 0.2 M Tris-HC1, 10% normal goat serum (NGS) and 0.5% Triton X-100. Alternating successive sections were then incubated with the antisera in the following way: Antisera were diluted in 0.2 M Tris, 3% NGS, and 0.5% Triton-X-100; after each incubation step the sections were washed with Tris-HC1. First antiserum: Rabbit-anti-GABA; 1:1,000-1:4,000 (a detailed description of the characterization of this antiserum is in preparation [Wulle and Schnitzer, in prep.]; results obtained with the same antiserum have previously been published by Wassle and Chun "881 and Wassle et al. "891); alternatively, rabbit-anti-TH (1:120; Eugene Tech Corp., Allendale) at 4°C overnight. Second antiserum: Goat-anti-rabbit (1:20, Sigma) 1hour at RT. Third antiserum: PAP (1:50; Sigma) 1 hour at RT. The reaction product was visualized by using 0.05 % diaminobenzidine, 0.01% H,O,. After further washing, the sections were counterstained with toluidine blue, then transferred to and fixed on microscopic slides for evaluation. Controls were performed by omitting the primary antisera and including all subsequent steps as outlined above; in these preparations, the cone ellipsoids displayed some unspecific label, but no staining of the neural components of the retina was observed.
Figs. 1-10. Light micrographs of semithin (0.5 pm) sections of the retinae of some anamniote vertebrates, arranged in pairs of consecutive sections after postembedding immunostaining with antisera to tyrosine hydroxylase (1, 3, 5, 7, 9) or GABA (2, 4, 6, 8, 10). Arrows in the anti-GABA-stained figures indicate the position of the TH-like immunopositive somata. In these species no colocalization is observed. x 550. Scale bar valid for all figures: 20 pm. Figs. 1, 2. Tangential sections of the junction of the inner nuclear and plexiform layers of the roach retina. Figs. 3,4. Radial sections of the roach retina. Figs. 5,6. Radial sections of the glass catfish retina. Figs. 7,8.
Radial sections of the Xenopus retina.
Figs. 9, 10. Tangential sections of the Xenopus retina at the junction of the inner nuclear and plexiform layers.
175
RESULTS In this study, we examined two teleost species routinely studied in our laboratory, and representatives of amphibians (Xenopus),widely used for electrophysiological studies, reptiles (turtle), birds (chick), and mammals (mouse) in order to investigate the colocalization of GABA-like and TH-like immunoreactivity in retinal neurons. T H is the rate-limiting enzyme of catecholamine synthesis, and the antiserum against it was used as a marker for dopaminergic neurones, since dopamine is the only catecholamine present in detectable amounts in vertebrate retinae. For the demonstration of GABAergic neurons, we used an antiserum directed to GABA that was conjugated to bovine serum albumin (Wulle and Schnitzer, in prep.). In each retina, five pairs of 0.5 pm thick, consecutive radial sections of about 2.5-3 mm total length were recorded photographically and by line drawings of perikarya. The patterns of labelling were matched by superposition of a line drawing and a light micrograph. In this way, perikarya of all labelled cells could easily be identified in both sections. Each radial section contained between three and five cells bodies with positive T H immunoreactivity. Since, in the mouse retina, the density of TH-positive cells was substantially lower in radial sections, only pairs of tangential sections comprising INL and IPL were prepared and analysed in this species. In all retinae studied, GABA immunoreactivity was restricted to perikarya localized predominantly in the INL, at the border of the inner and outer plexiform layers. Cells lining the IPL could be related to and identified as amacrine and/or interplexiform cells, whilst perikarya next to the outer plexiform layer were readily recognized as horizontal cells. In addition, GABA-reactive perikarya of displaced amacrine cells were encountered in the ganglion cell layer. Processes of GABA-positive cells formed a very dense network that almost completely filled the IPL with an all but homogeneous pattern of immunoreactive stain. The TH-positive retinal neurons in all vertebrate species investigated were characterized by large somata located in the inner INL, at the border of the IPL. In whole-mount preparations (not shown) we found these neurons to be distributed evenly and in low densities all over the retinae. Processes of TH-immunoreactive cells were much sparser than for GABA-positive cells and, in most species, could be identified as continuous fibers or rows of individual varicosities. Most often, stained fibers were found in sublayer 1and some deeper layers of the IPL which were, however, not further characterized. In the outer plexiform layer (OPL) the most prominent plexus of TH-immunoreactive fibers was observed in the roach retina (Fig. 3). In the catfish, a less conspicuous plexus of such fibers was present in the OPL (Fig. 5). Therefore, in these two teleost species THimmunopositive cells were identified as interplexiform cells. In the other vertebrate species studied in whole-mount preparations (not shown) we found several isolated immunostained fibers in the OPL; however, we cannot be certain whether TH-immunoreactive neurons in these cases are interplexiform cells or rather, as thought traditionally, amacrine cells. The assessment of colocalization of GABA- and TH-like immunoreactivity was performed by superimposing transparent drawings of anti-TH stained preparations onto light micrographs of anti-GABA labelled sections. For greater
176
I. WULLE AND H.-J. WAGNER
Figures 11-16
PATTERNS OF GABA AND TH COLOCALIZATION I N VERTEBRATE RETINAE
177
and a primate (Macaca nemestrina, Nguyen-Legros and Simon, '88). By contrast, no colocalization was found in neurons of fish and Xenopus retinae. This indicates that colocalization of GABA and T H is not a general phenomenon among vertebrates; rather, it is present in retinal neurons of higher, i.e., amniote vertebrates and is lacking in retinal neurons of lower, anamniote vertebrates. The respective individual functions of dopamine and GABA in retinal signal processing have been extensively studied in various vertebrate species. As reviewed by Kamp ('85) and Besharse et al. ('88), dopamine plays an essential role in controlling the adaptational processes preparing the retina to function under photopic as well as scotopic conditions. In the outer retina of fish, dopamine acts on horizontal cells to modulate intercellular coupling (Teranishi et al., '83) and on photoreceptors to induce light adaptational retinomotor movements (Dearry and Burnside, '86), while in the inner retina, dopaminergic neurons are involved in local circuits with bipolar and amacrine cells (Yazulla and Zucker, '88). In turtle, dopamine similarly decreased the permeability of horizontal cell gap junctions (Piccolino et al., '84, '87). In mammals, on the other hand, effects of dopamine on ganglion cell responses were mostly studied; receptive field properties were made more sustained and spatial summation characteristics more linear; interestingly, off-centre cells appeared to be more strongly affected by dopamine depletion than on-centre cells (Maguire and Smith, '85; Jensen and Daw, '86). DISCUSSION As for the function of GABA, it appears at the same time We used an antiserum directed to T H and indirect to be far more complex, yet less well defined than that of immunocytochemistry to visualize the dopaminergic neu- dopamine: In a general way, GABA is suggested to contribrons in the retinae of various vertebrates. Our results are in ute to the organization of complex receptive fields of agreement with previous observations based on formalde- ganglion cells (Caldwell and Daw, '78; Caldwell et al., '78; hyde-induced fluorescence (Ehinger '66, '76; Dowling and Morgan, '85). In the IPL, this is thought to be accomplished Ehinger, '75, '78) and with biochemical analyses showing through the extensive plexus of GABA-like immunoreactive that dopamine is the only neuronal catecholamine present amacrine cell processes surrounding the bipolar terminals; in the retina (for review see Kamp, '85). As for the visualiza- in the outer retina, by GABAergic L-type horizontal cells tion of GABAergic neurons, the antiserum was raised against feeding back onto photoreceptors, thus also acting on the a glutaraldehyde-based conjugate of GABA and bovine size and chromatic organization of receptive fields of bipolar serum albumine. Specificity tests as well as an evaluation of and ganglion cells. There are intricate interrelationships between dopamine the staining pattern in view of previous findings using autoradiography and glutamate decarboxylase (GAD)immu- and GABA in some vertebrate retinae. In fish, GABAergic nocytochemistry are outlined by Wassle and Chun ('88) and amacrine cells inhibit the activity of dopaminergic interplexiform cells (Ishita et al., '88); conversely, dopamine is Wassle et al. ('89). Our observations show that GABA-like and TH-like capable of inhibiting the dark induced release of GABA immunoreactivity are colocalized in the retinal neurons of from H1 horizontal cells (Yazulla '85). While this release turtle, chick, and mouse. As far as mammals are concerned, appears to be Ca++-independent,the Ca++-dependentliberthis finding is supported by previous immunocytochemical ation of GABA is enhanced by dopamine (O'Brien and findings suggesting the colocalization of GABA and dopa- Dowling, '85). By contrast, no effect of dopamine on GABA mine in retinal neurons of the cat (Wassle and Chun, '88) release has been found in chick (Nistico et al., '83). Finally, GABA released from amacrine cells in rat has been observed to inhibit both synthesis and release of dopamine (Pro11and Morgan, '83). In summary, however, no difference in these Figs. 11-16. Light micrographs of semithin (0.5 pm) sections of the pharmacological or physiological effects are apparent which retinae of some amniote vertebrates, arranged in pairs of consecutive might be related to the different patterns of colocalization sections after postembedding immunostaining with antisera to tyrosine observed here. hydroxylase (11,13,15) or GABA (12,14,16).Arrows in the anti-GABAThis situation is reminiscent of the interactions between stained figures indicate the position of the TH-like immunopositive GABA and acetylcholine which also colocalize in a specific somata. In these species colocalization of the two markers is observed. type of amacrine cell (Brecha et al., '88; Vaney and Young, x 550. Scale bar valid for all figures: 20 pm. '88). Both GABA and acetylcholine are co-released from Figs. 11,12. Radial sections of the turtle retina. these neurons (Masland et al., '89). Acetylcholine release was optimally stimulated by flashing light or moving stimFigs. 13,14. Radial sections of the chick retina. uli, while GABA release was not affected by this paradigm. GABA release was Ca++-independentwhereas acetylcholine Figs. 15, 16. Tangential section through the inner nuclear layer of was liberated in a Ca++-dependent manner. They also the mouse retina.
clarity we concentrated on the perikarya which could be unequivocally identified in each pair of neighbouring sections. In the figures, the observations are presented as pairs of light micrographs where cell bodies exhibiting colocalization of TH- and GABA-like reaction product (or the lack of it) are marked with arrows in the anti-GABA-stained preparations. Figures 1-10 show the situation in anamniote vertebrates, i.e., fish and amphibians; in all three cases, it is obvious that the neurons containing TH-like immunoreactivity (Figs. 1, 3 , 5 , 7 , 9 ) are not labelled by GABA-antibodies (Figs. 2 , 4 , 6 , 8,lO). On the other hand, horizontal and amacrine cells are clearly GABA-positive in both species of teleost as well as in Xenopus. This lack of colocalization in fish and amphibia exemplified by these micrographs was also observed in all other pairs of sections not documented here. In higher, amniote vertebrates, by contrast, the comparison of the two labelling patterns shows that perikarya with TH-like immunostaining are also stained with GABA-like immunoreactivity; this applies for the radial sections of turtle and chick retinae (Figs. 11-14) as well as for the tangential views of the mouse IPL (Figs. 15, 16). With regard to the other material analysed it can be stated that we found not a single TH-positive cell (n = 13-19, depending on the species) in these three species that did not also contain GABA positive reaction product.
I. WULLE AND H.-J. WAGNER
178 reported that although both substances were released from the same cells, only acetylcholine was stored in vesicles and liberated by vesicle exocytosis; by contrast, GABA was probably released by a carrier mechanism. It was speculated that the release of two transmitters from the same cell could serve to specifically refine and shape the pre- and/or postsynaptic response. On a more general level, cholinergic starburst amacrine cells were shown to be tonically inhibited by endogenous GABA (Agardh, '86). It is unknown whether two neuroactive substances contained within the same neuron are directed to different targets or whether they may act on the same postsynaptic site. A possible mechanism for the action of two conventional neurotransmitters on a single postsynaptic cell was proposed recently by Foster and Kemp ('89). For glutamate and glycine, they suggested an agonistic action leading to the enhancement of the N-methyl-D-aspartate (NMDA)mediated response to glutamate. In the case of a colocalization of a neuropeptide (neuropeptide Y) and epinephrine it has been suggested that the neuropeptide might modulate the action of the small transmitter molecules at the level of the smooth muscle targets (Cooper et al., '86). As for the functional role of colocalization of GABA and dopamine for retinal information processing, there is not even a basis for speculation. Our finding that the coexistence of these two transmitters is not a common phenomenon among vertebrates would indicate that it is not essential to retinal functioning. It is obviously not related to the cell type expressing dopamine since it is observed in both amacrine and interplexiform cells. Rather, the lack of colocalization in anamniote, i.e., aquatic vertebrates and the simultaneous expression of both these transmitters in amniote, i.e., essentially terrestial species may indicate either an evolutionary trend or may be linked to visual or adaptational problems associated with the transition from water to land.
ACKNOWLEDGMENTS We are grateful to M. Schneider for skillful technical help and Dr. M. Kirsch for critically reading and commenting on the manuscript. The financial support by the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
LITERATURE CITED Agardh, E. (1986) GABA antagonists increase the release of 3H acetylcholine from the chick retina. Acta Physiol. Scand. 126r33-38. Besharse, J.C., P.M. Iuvone, and M.E. Pierce (1988) Regulation of rhythmic photoreceptor metabolism: A role for post-receptoral neurons. Progress in Retinal Research 7t21-61. Brecha, N., D. Johnson, L. Peichl, and H. Wassle (1988) Cholinergic amacrine cells of the rabbit retina contain glutamate decarboxylase and gammaaminobutyrate immunoreactivity. Proc. Natl. Acad. Sci. USA 85:61876191. Caldwell, J.H., N.W. Daw, and H.J. Wyatt (1978) Effects of picrotoxin and strychnine on rabbit retinal ganglion cells: Lateral interactions for cells with more complex receptive fields. J. Physiol. 276277-298. Caldwell, J.H., and N.W. Daw (1978) Effects of picrotoxin and strychnine on rabbit retinal ganglion cells: Changes in centre surround receptive fields. J. Physiol. 276t299-310. Cooper, J.R., F.E. Bloom, and R.H. Roth (1986) Neuroactive peptides. The biochemical basis of neuropharmacology, Fifth edition, Oxford University Press 352-393. Dearry, A., and B. Burnside (1986)Dopaminergic regulation of cone retinomotor movement in isolated teleost retinas. I.: Induction of cone contraction is mediated by D2 receptors. J. Neurochem. 46r1006-1021.
Dowling, J.E., and B. Ehinger (1975) Synaptic organization of the amine containing interplexiform cells of the goldfish and cehus monkey retinas. Science, Wash. 188,270-273. Dowling, J.E., and B. Ehinger (1978) The interplexiform cell system. I. Synapses of the dopaminergic neurons of the goldfish retina. Proc. R. SOC. Lond. [Biol.] 201t7-26. Ehinger, B. (1966) Adrenergic retinal neurons. Z. Zellforsch. mikrosk. Anatomie 71:146-152. Ehinger, B. (1976) Weak neuronal accumulation of octopamin in dopaminergic neurons of rabbit retina. Experientia 32690. Foster, A.C., and J.A. Kemp (1989) Glycine maintains excitement. Nature 338:337-338. Ishita, S., K. Negishi, T. Teranishi, Y. Shimada, and S. Kato (1988) GABAergic inhibition of dopaminergic cells of the fish retina: A ('H)dopamine release study with isolated fractions. J. Neurochem. 50:1-6. Jensen, R.J., and N.W. Daw (1986) Effects of dopamine and its agonists and antagonists on the receptive field properties of ganglion cells in the rabbit retina. Neuroscience 17t837-855. Kamp, C.W. (1985) The dopamine system of the retina. In: W.W. Morgan (ed): Retinal Transmitters and Modulators. Models for the Brain Vol. 11. Boca Raton: CRC Press, pp. 1-31. Kosaka, T., K. Kosaka, Y. Hataguchi, I. Nagatsu, J.-Y. Wu, O.P. Ottersen, J. Storm-Mathisen, and K. Hama (1987) Catecholaminergic neurons containing GABA-like and/or glutamic acid decarboxylase-like immunoreactivities in various brain regions of the rat. Exp. Brain Res. 66:191-210. Maguire, G.W., and E.L. Smith, 111 (1985) Cat retinal ganglion cell receptive field alterations after 6-hydroxydopamine induced dopaminergic amacrine lesions. J. Neurophysiol. 53t1431-43. Masland, R., C. Cassidy, and D. O'Malley (1989) The release of acetylcholine and GABA from amacrine cells. In: R. Weiler and N.N. Osborne, eds.: Neurobiology of the inner retina. Berlin, Heidelberg: Springer, p. 15-26. Morgan, W.W. (1985) GABA. A potential neurotransmitter in retina. In: W.W. Morgan (ed) Retinal Transmitters and Modulators: Models for the Brain. Vol. I. Boca Raton: CRC Press, pp. 63-96. Nistico, G., A. De Sarro, G. Urna, A. Pujia, and R. Jentile (1983) Lack of relation between dopaminergic and GAGAergic mechanisms in chick retina. Res. Commun. Chem. Pathol. Pharmacol. 40:29-39. Nguyen-Legros, J., and A. Simon (1988) Dopamine-GABA relationship in the retina of Maeaca nernestrina. Abstr. of the 11th Ann. Meeting of the Eur. Neurosci. Association 425. OBrien, D.R., and J.E. Dowling (1985) Dopaminergic regulation of GABA release from the intact goldfish retina. Brain Res. 360:41-50. Piccolino, M., J. Neyton, and H.M. Gerschenfeld (1984) Decrease of gap junction permeability induced by dopamine and cyclic adenosine 35'monophosphate in horizontal cells of turtle retina. J. Neurosci. 424772477. Piccolino, M., P. Witkovsky, and C. Trimarchi (1987) Dopaminergic mechanisms underlying the reduction of electrical coupling between horizontal cells of the turtle retina induced by d-amphetamine, bicuculline and veratridine. J. Neurosci. 7r2273-2283. Proll, M.A., and W.W. Morgan (1983) Use of butyric acid (GABA) transaminase inhibitors and GABA uptake inhibitors to investigate the influence of GABA neurons on dopamine-containing amacrine cells of the rat retina. J. Pharmacol. Exp. Ther. 227t627-632. Sternberger, L.A. (1979) Immunocytochemistry, 2nd Ed. New York John Wiley and Sons, Inc. Teranishi, T., K. Negishi, and S. Kato (1983) Dopamine modulates S potential amplitude and dye coupling between external horizontal cells in carp retina. Nature (London) 301:243-246. Thier, P., and V. Alder (1984) Action of iontophorectically applied dopamine on cat retinal ganglion cells. Brain Res. 292109-121. Vaney, D.I., and H.M. Young (1988) GABA-like immunoreactivity in cholinergic amacrine cells of the rabbit retina. Brain Res. 438t369-373. Wassle, H., and M.H. Chun (1988) Dopaminergic and indoleamine-accumulating amacrine cells express GABA-like immunoreactivity in the cat retina. J. Neurosci. 8:3383-3394. Wassle, H., U. Grunert, J. Rohrenbeck, and B.B. Boycott (1989) Cortical magnification factor and the ganglion cell density of the primate retina. Nature 341r643-646. Yazulla, S. (1985)Evoked efflux of 3H-GABA from goldfish retina in the dark. Brain Res. 325:171-180. Yazulla, S., and C.L. Zucker (1988) Synaptic organization of dopaminergic interplexiform cells in the goldfish retina. Visual Neuroscience Ir13-29.
