- Email: [email protected]
Patterns of glutamate-like immunoreactive bipolar cell axons in the retina of the marine teleost, the dragonet
Vision Res. Vol. 31, No. 3,pp. 451-462,1991 Printed in Great Britain. All rights reserved
0042-6989/91 $3.00+ 0.00
Copyright 0 1991Pergamon Press plc
PATTERNS OF GLUTAMATE-LIKE IMMUNOREACTIVE BIPOLAR CELL AXONS IN THE RETINA OF THE MARINE TELEOST, THE DRAGONET E. VAN HAJSENDONC~ and L. MSSOTTEN Eye Research Laboratory, Catholic University of Leuven, Kapucijnenvoer 33, B-3000 Leuven, Belgium (Received 28 December
1989)
Ahatraet-Oblique 1 pm-sections through the dorsal inner plexiform layer of the light-adapted dragonet retina were processed for ~stem~ding, silver-enhanced immunogold labeling after incubation with a glutamate-specific antiserum. Light microscopy showed strongly immunolabeled boutons grouped into distinct square patterns. These patterns were compared with the succesive grids of bipolar axonal boutons revealed by electron microscope analysis of serial, oblique sections through the entire dorsal inner plexiform layer. With one exception, all types of patterned bipolar synaptic boutons could be clearly identified in the immunoreactive staining pattern. The elevated levels of endogenous glutamate in most bipolar synaptic boutons suggest that the large majority of bipolar cell types use glutamate as their neurotransmitter. However, some bipolar synaptic boutons displaying low levels of glutamate indicate that a small number of bipolar cells may contain another neuroactive substance. Retina
Bipolar cells
Inner plexiform layer
Glutamate
A line of electrophysiological and pharmacological evidence suggests that ON- and OFFcenter bipolar cells in the vertebrate retina are excitatory in nature and that L-glutamate (Glu) is the most likely neurotransmitter of bipolar cell synapses in the inner plexiform layer (IPL) (Naka, 1976, 197’7; Miller & Dacheux, 1976a, b; Miller, 1979; Slaughter & Miller, 198 1, 1983a, b; Toyoda & Fujimoto, 1984; Bloomfield & Dowling, 1985; Lukasiewicz & McReynolds, 1985; Coleman, Massey & Miller, 1986; Massey & Miller, 1988). Consistent with this hypothesis, antisera against Glu strongly label bipolar cells and their axonal synaptic boutons. This indicates that significantly enhanced levels of endogenous Glu are present, a prerequisite for glutamatergic synaptic transmission (Ehinger, Ottersen, Storm-Mathisen & Dowling, 1988; Marc, Massey, Kalloniatis & Basinger, 1989). By virtue of its regular pattern arrangements and the certainty with which bipolar axonal boutons can be identified, the retina of the marine teleost, the dragonet, is a very suitable model to study bipolar cell neurotransmitter candidates. The dragonet retina consists of a ventral mixed and a dorsal pure-cone half. In the dorsal half, where higher neuron densities occur, bipolar cell axons display three distinct square patterns at defined levels in the IPL.
Immunocytochemistry
Teleost
These patterns mirror the photoreceptor cell disposition: the narrow S-pm pattern corresponds to the arrangement of the double cone pairs; of the two wider 7-pm patterns, one corresponds to the single cone distribution, while the other coincides with the intersections of the rows of double cone pairs (Van Haesendonck & Missotten, 1983a). We have shown that bipolar cells in the ventral and the dorsal region of the dragonet retina are strongly immunolabeled with antiGlu. On semithin vertical sections of the dorsal retina, the five levels with geomet~cally arrayed bipolar synaptic boutons are clearly identifiable by their prominent Glu-like immunoreactivity (GLIR) (Van Haesendonck & Missotten, 1990). In the present report, we complement our previous results by determining the extent to which bipolar cell types in the dorsal retina are immunoreactive for anti-Glu. We show the localization of endogenous Glu on semithin oblique sections of the dorsal IPL, processed for postembedding Glu-immunocytochemistry. We compare the arrangement of immunolabeled synaptic boutons with the consecutive square patterns of bipolar axonal boutons revealed by electron microscope analysis of serial, ultrathin oblique sections through the entire IPL. The correlation allows us to judge whether all the geomet~~lly arrayed bipolar cells are immunoreactive for anti-Glu. We find that all but one
451
E. VAN HAESENDONCK and L. MKSOTTEN
452
Dragonets (Callionymus lyra L.) were caught in the North Sea. Light-adapted fish were killed by cervical transection and pithing. After enucleation, the retinas were isolated and fixed for 2 hr by immersion in cold 2.5% glutaraldehyde on 0.1 M sodium cacodylate buffer (pH 7.4) supplemented with 3% sucrose. They were then rinsed overnight in 0.1 M phosphate buffer (PB) (pH 7.4) supplemented with 3% sucrose (PBS). All retinas intended for light microscopic immunocytochemistry were dehydrated over increasing concentrations of ethanol and, with propylene oxide as intermediary, embedded in epoxyresin (Ladd, U.S.A.). Retinas from other animals, intended for electron microscopy, were postfixed for 2 hr in 2% 0~0, in PB before dehydration and embedding in epoxyresin.
showed selective recognition of the Gluconjugate. The immunoreaction was abolished by preabsorption of anti-Glu with Gluconjugate, and when anti-Glu was omitted in the immunocytochemical procedure. Sections were first etched with sodium ethoxide for 15 min. After extensive washing in absolute ethanol, PB and distilled water, they were preincubated for 2 hr in 3% normal goat serum in PBS with 0.3% triton X-100 (PSBTx). Subsequently, the sections were washed in PBSTx (2 x 1 hr) and incubated for 3 hr in goat anti-rabbit antiserum conjugated to 5 nm colloidal gold particles (Auroprobe EM G5, Janssen, Belgium), diluted 1: 50 in the same diluent as the primary antiserum. The sections were washed in PBS (2 x 1 hr) and fixed in 2.5% glutaraldehyde in PBS for 10 min. After several rinses in PB and distilled water, they were treated with IntenSE M (Janssen, Belgium) to silverenhance the immunogold labeling. All steps in this immunocytochemical procedure were performed at room temperature. After a final rinse in three changes of distilled water, the sections were lightly counterstained with buffered toluidine blue and coverslipped with epoxyresin. They were examined with a x 40 oil immersion planapochromatic objective (Zeiss, F.R.G.).
Immunocytochemistry
Electron microscopy
Vertical and oblique l-pm sections were collected on glass slides and processed acccording to a method (modified from Danscher & Norgaard, 1983; Holgate, Jackson, Cowen & Bird, 1983) for postembedding, silver-intensified immunogold labeling after incubation with an anti-Glu antiserum (anti-Glu). The anti-Glu (Biosys, France) was raised in rabbit against a Glu-glutaraldehyde-thyroglobulin conjugate. Tested in dot-immunoassays with serial dilutions of synthesized Glu, aspartate, glutamine, GABA and glycine conjugates, the anti-Glu
Serial, ultrathin oblique sections from the inner nuclear layer (INL) to the ganglion cell layer (GCL) were collected on formvar-coated, single-slot grids. They were stained with lead citrate for 8 min (Venable & Coggeshall, 1965) and examined with a Zeiss 9S-2 electron microscope.
of these bipolar cell types in the dragonet retina contain elevated Glu levels. Furthermore, bipolar cell types with axons terminating in fibrous layers of the dorsal IPL, particularly the most distal and proximal, also have synaptic boutons enriched with endogenous Glu. MATERIALS
AND METHODS
Tissue preparation
RESULTS
Figure la serves to illustrate the location of the pattern levels (P-levels) in the IPL on a
Fig. I (Opposite). (a) Distribution of glutamate-like immunoreactivity (GLIR) on a vertical I-pm section through the dorsal dragonet retina. Strong labeling is seen over the photoreceptor cell terminals (PCT) and over the bipolar and ganglion cell bodies. Among the bipolar cell bodies, a number of moderately labeled profiles occur (white arrows). In the inner plexiform layer (IPL), sequences of regularly spaced immunoreactive boutons are observed at five levels (PI-PS). The three larger bouton types at the P3 level are immunostained. A BP3, 1 bouton (small arrowhead) lies adjacent to a BP3, 2d bouton (arrow). On the left and the right of this pair lies a BP3, 2p bouton (large arrowheads). GCL, ganglion cell layer; INL, inner nuclear layer. Scale bar = I2 pm. (b) Distribution of GLIR on an oblique I-pm section through the IPL in the dorsal dragonet retina. The inner plexiform layer (IPL) is moderately labeled. Five levels (PI-PS) with distinct types of strongly immunoreactive and square-patterned boutons are observed. Bouton profiles in the Fl and particularly in the F6 fibrous layer are also strongly immunoreactive. No labeling is seen over the inner nuclear layer (INL) or the distal row of small cell bodies in the ganglion cell layer (GCL). Scale bar = I2 pm.
INL _ Fl Pi.
Pi.
P3. IPL P4
P5
F6
GCL Fig. 1
Fig. 2. Bipolar cell organization at the PI level. (a) Electron micrograph of the Pt , 1 layer. Bipolar synaptic boutons with high synaptic vesicle density (BP& 1 encircled) am grouped by a 5-pm square. Scale bar = 2 pm. (b) Electron micrograph of the PI, 2 layer. Bipolar synaptic boutons are less densely tilled with synaptic vesicles (BPl. 2 encircled) and grouped by a 7-pm square. The 5-pm and the 7-pm square differ in orientation by 45 deg (indicated by the arrows). Scale bar = 2 pm. (c). Light micrograph of the GLIR pattern at the Pl level. Two types of strongly labeled boutons are seen: small boutons arranged according to a narrow square and, more proximally, larger boutons arranged in a wider square. The orientation of both squares also differs over 45 deg. l-Km section, scale bar = 12 pm. 454
Fig. 3. Bipolar cell organization at the P2 level. (a) Electron micrograph of the Pl ,2 layer. Bipolar synaptic boutons with high synaptic vesicle density (BP2, 1 encircled) are grouped by the 5-pm square. Scale bar = 2 pm. (b) Electron ~cro~aph of the P2,2 Iayer. Irregularly shaped boutons with broad collaterals (BP2, 2 encircled) are also grouped by the S-pm square (neigh~~ng BP2, 1 boutons in the upper right hand comer are indicated by a dashed line). Scale bar = 2 pm. (c) Light micrograph of the GLIR pattern at the P2 level. A single pattern of small, round boutons grouped by the narrow square is observed. I-pm section, scale bar = 12 pm.
455
Fig. 4. Bipolar cell organization at the P3 level. (a) Electron micrograph of the P3 level. Four types of bipolar synaptic bouton are observed. Most distally, boutons with low synaptic vesicle density (BP3, I encircled by dotted line) are grouped by a 7-pm square. Immediately adjacent proximally are boutons with high synaptic vesicle density (BP3,2d encircled by dashed line), arrayed according to the same 7-pm square pattern. Larger boutons, less densely filled with synaptic vesicles (BP3, 2p encircled by solid line) are grouped by another, distinct 7-pm square. Finally, small boutons with high synaptic vesicle density (BP3, 2s encircled by solid line) show S-pm square arrangement. Scale bar = 2 pm. (b) Light micrograph of the GLIR staining at the P3 level. Square patterns with small and large, strongly labeled boutons interfere. Arrowheads indicate BP3, 2s boutons. I-ym section, scale bar = 12 pm.
456
Fig. 5. Bipolar cell organization at the P4 level. (a) Electron micrograph of the P4 level. A single type of bipolar synaptic boutons, densely filled with synaptic vesicles and showing Sprn square arrangement is observed (BP4 encircled). Scale bar = 2 pm. (b) Light micrograph of the GLIR at the P4 level. Likewise, a single narrow square pattern of small, strongly labeled boutons is noted. l-pm section, scale bar = 12pm.
Glutamate-like immuroreactive bipolar cell axons
vertical l-pm section of the dorsal dragonet retina which has been processed for anti-Glu immunocytochemistry. Five levels with regularly grouped and strongly labeled boutons are discerned (PI-PS). Prominent GLIR is also observed over the bipolar cell bodies in the INL, the photoreceptor terminals (cone pedicles) and the ganglion cell bodies (Van Haesendonck & Missotten, 1990). A number of cell body profiles in the bipolar cell half of the INL appear to be less intensely labeled (white arrows). Figure lb shows the distribution of GLIR on an oblique 1 pm-section of the dorsal IPL. The IPL is moderately labeled. At successive levels, grids of strongly immunoreactive boutons are observed. The location of these grid layers in the IPL coincides with the P-levels of squarepatterned bipolar axonal boutons (Pl-PS). In some layers (e.g. at the P2 and the P4 level) a single square pattern occurs, in others (e.g. at the P3 and the P5 level) overlapping square patterns are seen. In addition, globular immunoreactive bouton profiles are observed in some fibrous layers (F-layers) that separate the P-levels. They are clearly noted in the thicker Fl and F6 layer. Particularly in the F6 layer, they are large and numerous. These boutons are not geometrically arrayed. The electron micrographs of the layers displaying square-patterned bipolar axonal boutons in the dorsal IPL are presented in Figs 2-6. A variety of boutons are observed and their bipolar nature is established by the presence of synaptic ribbons. The terminology of the boutons is the same as in previous studies (Van Haesendonck Jz Missotten, 1983a, b). Boutons are named after the layer in which they are located and, in cases where more than one type occurs in a single layer, reference is made to a major morphological feature (e.g. d, boutons with dark appearance due to high synaptic vesicle density). For comparison with the GLIR patterns, the accompanying light
459
micrographs (Figs 2c, 3c, 4b, 5b and 6c) in each case illustrate the disposition of immunoreactive boutons in that particular P layer. At the Pl level (at 15-20% in the IPL), two types of bipolar boutons occur, each arranged according to a different square pattern (Fig. 2). The medium-sized, round boutons (BP1 , I ; approx. 2.8 pm dia.) in the distal layer have a high synaptic vesicle density and show the j-pm square disposition (Fig. 2a). The larger, round boutons (BPl, 2; approx. 4.5 pm dia.) in the proximal layer are less densely filled with synaptic vesicles and show a 7 pm square disposition (Fig. 2b). The GLIR pattern at the PI level (Fig. 2c) includes two types of intensely labeled boutons, grouped by two distinct square patterns. Smaller immunoreactive boutons are arranged according to a narrow (5 pm) square, larger immunoreactive boutons are arranged according to a wider (7 pm) square. On the electron micrographs of the Pl layers, as well as in the GLIR pattern, a 45 deg difference in orientation between the two square patterns is noted. The PZlevel (at 30-35% in the IPL) likewise consists of two layers, each containing a single bouton type (Fig. 3). The distally located, round boutons (BP2, 1; approx. 2.8 pm dia.) with high synaptic vesicle density resemble the BPl, 1 boutons and are also grouped by the j-pm square (Fig. 3a). Immediately proximal lies a second layer, containing boutons with the same j-pm square disposition. These boutons (BP2,2) are flattened, with broad, irregularly shaped collaterals, and have a low synaptic vesicle density (Fig. 3b). Figure 3c shows the distribution of GLIR boutons at the P2 level. Important to note is the absence of two distinct GLIR grids at this level. Only a single, narrow square pattern of small, round and strongly labeled boutons is observed. The P3 level (at 45-50% in the IPL) is complex. At this level, four types of boutons
Fig. 6 (Opposite). Bipolar cell organization at the PS-level. (a) Electron micrograph of the P5, I layer. Two types of bipolar synaptic boutons are observed. Boutons with low synaptic vesicle density are grouped by the 5-pm square (BPS, lp encircled by solid line). Boutons with high synaptic vesicle density are group by a 7-pm square (BPS, Id encircled by dashed line). These bouton types mutually contact (arrows). Scale bar = 2 pm. (b) Electron micrograph of the PS, 2 layer. Again, two types of bipolar synaptic boutons occur. Large boutons with variable synaptic vesicle density (BP5,2b encircled by solid line) have the 5-pm square arrangement. Small boutons with high synaptic vesicle density (BP5,2s encircled by dashed line) have a 7-pm square arrangement. Scale bar = 2 pm. (c) Light micrograph of the GLIR pattern at the P5 level. In the distal P5, 1 layer, a lattice-like arrangement of strongly labeled boutons is seen. In the proximal P5, 2 layer, small and large strongly immunostained boutons with distinct square arrangement are observed. l-pm section, scale bar = 12 pm.
460
E. VAN HAESENWNCKand L. MISSOTTEN
occur, and the three distinct square patterns interfere. Two poorly separated layers can be distinguished. A distal layer contains mediumsized, round boutons (BP3, 1; approx. 3.0pm dia.) with low synaptic vesicle density and grouped by a 7-pm square pattern. The closely adjoining proximal layer has been described in detail previously (Van Haesendonck & Missotten, 1983a). This layer contains three bouton types. The largest boutons, with moderate synaptic vesicle density (BP3, 2p; approx. 4pm dia.) and the medium-sized boutons with high synaptic vesicle density (BP3, 2d; approx. 3.0,um dia.) are patterned according to two different 7-pm squares. The small boutons with high synaptic vesicle density (BP3, 2s; approx. 1.3 pm dia.) form a 5-pm grid. Figure 4b illustrates the distribution of GLIR boutons at the P3 level. Boutons of different sizes are visible whose dispositions are unmistakably based upon square patterns. Their overall arrangement, however, appears somewhat disordered. The P4 level (at 5560% in the IPL) is the sole pattern level at which single type of boutons occurs (Fig. 5). These boutons are rather small (BP4; approx. 2pm dia.), contain many synaptic vesicles (Fig. 5a), and have the 5-pm square disposition. Figure 5b shows the arrangement of boutons immunolabeled with anti-Glu at this level. A narrow square pattern of small GLIR boutons is observed. The P5 level (at 70-75% in the IPL) is the most extensive pattern level (Fig. 6). It consists of a distal layer (P5, 1) with two bouton types and a proximal layer (P5,2), also containing two boutons types. Both layers are separated by a very thin fibrous layer. In the P5, 1 layer (Fig. 6a), the two bouton types are quite large (approx. 3.0 pm dia.). One type has a moderate to low synaptic vesicle density (BP5, lp) and 5-pm square disposition. The other type has a high synaptic vesicle content (BPS, Id) and a 7-pm square disposition. These dissimilar boutons have considerable contact zones with specialized cell membranes (Fig. 6a, arrows; Van Haesendonck & Missotten, 1983b). The corresponding pattern of GLIR boutons is shown in Fig. 6c. The immunoreactive boutons also appear to contact one another, forming a lattice-like structure. Labeled boutons in narrow squares are more often observed in this layer. In the P5, 2 layer (Fig. 6b), the large boutons (BP5, 2b; approx 4pm dia.) have a variable synaptic vesicle content. They
are grouped into the 5-pm square pattern, The small boutons in this layer (BP5, 2s; approx. 1.3 ,um dia.) have a high synaptic vesicle density and are arranged according to a 7-pm square pattern. The immunostaining pattern of the P5, 2 layer (Fig. 6c) shows a large and a very small type of labeled boutons. The former displays the narrow square pattern, while the latter is arranged according to the wider square pattern. DISCUSSION
As anticipated, the present results confirm that the regularly spaced GLIR boutons, observed on vertical sections of the dorsal IPL in the dragonet retina (Van Haesendonck & Missotten, 1990), display square patterns. The different levels with geometrical disposition of immunolabeled boutons correspond to the levels with square patterned bipolar synaptic boutons, observed by conventional electron microscopy of the dorsal IPL. These observations emphasize the bipolar cell origin of the immunostained boutons. To exactly what extent can the individual types of square-patterned bipolar boutons be identified in the GLIR patterns? Careful comparison of GLIR and electron microscopically observed grids at the successive P-levels indicates that all but one type of patterned boutons is strongly immunoreactive for anti-Glu and can be readily identified by the size differences and the distinct spacings. The boutons at the Pl, P4 and P5 level are the most easily correlated. The size and the disposition of the boutons in the GLIR patterns at these levels effectively correspond to those of the bipolar synaptic boutons in the electron microscopically observed patterns. Also, the 45 deg difference in orientation between the narrow and the wide square patterns is obvious in the arrangement of the GLIR boutons. At the P3 level, interposition of the three square patterns results in a mingled arrangement of bipolar axonal boutons. Likewise, the corresponding pattern of immunoreactive boutons has a complicated, rather confusing appearance. Moreover, the similarity in size among some bouton hampers identification of different types within the GLIR pattern, and makes correlation with EM images difficult. However, combining observations on vertical and oblique sections allows us to conclude that all four bouton types that occur at the P3 level show GLIR. Figure la indicates that the three
Glutamate-like immuroreactive bipolar cell axons
larger types are strongly labeled. The two vertically adjacent boutons (small arrowhead and arrow) can be correlated with the BP3, I and the BP3, 2d boutons. The boutons on the left and the right of these (large arrowheads) most likely correspond to BP3, 2p boutons. Furthermore, Fig. 4c (arrowheads) shows small, regularly distributed, immoreactive boutons that interfere with the arrangement of the large boutons. These correspond to the BP3, 2s boutons. One type of patterned boutons, namely the BP2, 2 boutons at the P2 level, are not clearly discernible. We correlate the round, labeled boutons in the GLIR pattern to the distal, round BP2, 1 boutons. It can be argued that the BP2, 1 and the BP2, 2 boutons have the same square pattern and, thus, are not easily distinguished from one another. Because of the characteristic shape of the BP2, 2 boutons with broad collaterals though, this type should be recognizable as a distinct GLIR lattice. Such a lattice is not observed. Also, a single layer of prominently immunostained boutons is observed at the P2 level on vertical sections. We conclude that BP2, 2 boutons are not labeled, or only very weakly labeled with antiGLU. The absence of intense immunostaining could be explained by the low synaptic vesicle content of these boutons. However, BP3, 1 and BP5, lp boutons have similar low synaptic vesicle densities and yet label much more strongly than the BP2, 2 boutons. Our results indicate that in the light-adapted retina, the endogenous Glu level in the BP2, 2 boutons is not markedly elevated. Interestingly, Ehinger et al. (1988) also observed a small number of bipolar cell terminals in the turtle IPL that showed little or no labeling. A number of the weakly labeled cell bodies in the distal half of the INL may belong to the bipolar cells with BP2, 2 axon terminals. The weakly labeled population of cell bodies probably includes somata of Miiller cells as well. In addition to the square-patterned bipolar boutons, large GLIR bouton profiles also occur in the fibrous layers between the P levels. These boutons are predominantly located in the Fl and most abundantly in the F6 layer. They most likely correspond to the synaptic boutons of bipolar axons without the square pattern disposition, observed by electron microscopy (Van Haesendonck & Missotten, 1983a, 1990). Such bipolar boutons have also been found in the F2 and the F3 layer. In these thinner fibrous layers, the less numerous and smaller
461
boutons are hard to detect by GLIR. Therefore, we do not exclude the possibility that some of these boutons are not immunoreactive for anti-Glu. If we generalize the electrophysiological results of Famiglietti, Kaneko and Tachibana (1977) in the carp retina and accept the subdivision into a distal OFF-sublamina and a proximal ON-sublamina for the dragonet IPL, then our present results once again show that GLIR boutons occur in both sublaminae (Van Haesendonck & Missotten, 1990). This suggests that ON- as well as OFF-center bipolar cell synaptic terminals are enriched with endogenous Glu and agrees with Glu-immunocytochemistry studies in other vertebrate retinas (Ehinger et al., 1988; Marc et al., 1989). In conclusion, most bipolar axonal boutons in the dorsal half of the light-adapted retina label strongly for anti-Glu. This indicates that elevated levels of endogenous Glu are present in the boutons and supports the idea that the large majority of bipolar cell types in this purecone region use Glu as their neurotransmitter. Our results leave open the possibility that a patterned bipolar cell type with axonal terminal (BP2, 2 boutons) in the distal sublamina of the IPL (OFF-sublamina) may contain a different or an additional neuroactive substance. In the teleost retina, no other classical neurotransmitter has been demonstrated in bipolar cells by immunocytochemistry. In other vertebrate classes, on the other hand, serotonin-, glycineand GABA-like immunoreactivity has been observed in bipolar cells (Bruun, Ehinger & Sytsma, 1984; Weiler & Schiitte, 1985; Mosinger, Yazulla & Studholme, 1986; Agardh, Ehinger & Wu, 1987; Wassle, Voigt & Patel, 1987; Millar, Winder, Ishimoto & Morgan, 1988; Yang & Yazulla, 1988a, b; Hendrickson, Koontz, Pourcho, Sarthy & Goebel, 1988; WZissle & Chun, 1989). As of yet, the identity of a possible second neurotransmitter substance in some bipolar cells of the dragonet retina remains to be determined. Acknowledgements-We thank Professor F. Ollevier and Dr P. Van Damme, K. U. Leuven Zoological Institute, for their cooperation with the collection of dragonets, Mrs A. Loenders for technical assistance and Dr S. Raiguel for helpful discussions. REFERENCES Agardh E., Ehinger B. & Wu J.-Y. (1987). GABA and GAD-like immunoreactivity in the primate retina. Histochemistry, 86, 485490.
462
E. VAN HAE~ENDONCK and L. MIS~OTTEN
Bloomfield, S. A. & Dowling, J. E. (1985). Roles of aspartate and glutamate in synaptic transmission in rabbit retina II. Inner plexiform layer. Journal of Neurophystology, 53, 714125.
Bruun, A., Ehinger, B. & Sytsma, U. M. (1984). Neurotransmitter localization in the skate retina. Brain Research, 295, 233-248.
Coleman, R. A., Massey, S. C. & Miller, R. F. (1986). Kynuremc acid distinguishes kainate and glutamate receptors in the vertebrate retina. Brain Research, 381, 172-175.
Danscher, G. & Norgaard, J. 0. D. (1983). Light microscopic vizuahzation of colloidal gold on resin-embedded tissue. Journal of Histochemistry and Cytochemistry, 31, 13941398.
Ehinger, B., Ottersen, 0. P., Storm-Mathisen, J. & Dowling, J. (1988). Bipolar cells in the turtle retina are strongly immunoreactive for glutamate. Proceedings of the National Academy of Sciences, U.S.A., g&8321-8325.
Famiglietti, E. V. Jr, Kaneko, A. & Tachibana, M. (1977). Neuronal architecture of ON and OFF pathways to ganglion cells in carp retina. Science, 198, 126771269.
Hendrickson, A. E., Koontz, M. A., Pourcho, R. G., Sarthy, P. V. & Goebel, D. (1988). Localization of glycine-containing neurons in the Macaca monkey retina. Journal of Comparative Neurology, 273, 473487.
Holgate, C. S., Jackson, P., Cowen, P. N. & Bird, C. C. (1983). Immunogold-silver staining: A new method of immunostaining with enhanced sensitivity. Journal of Histochemistry and Cytochemistry, 31, 938-944.
Lukasiewicz, P. D. & McReynolds, J. S. (1985). Synaptic transmission at N-methyl-o-aspartate receptors in the proximal retina of the mudpuppy. Journal of Physiology, London, 367, 99-116.
Marc, R. E., Massey, S. C., Kalloniatis, M. & Basinger, S. F. (1989). Immunochemical evidence that the fast neurotransmitter of rods, cones, bipolar and ganglion cells is glutamic acid. Investigation Opthalmology & Visual Science (Suppl.), 30, 320. Massey, S. M. & Miller, R. F. (1988). Glutamate receptors of ganglion cells in the rabbit retina: Evidence for glutamate as a bipolar cell transmitter. Journal of Physiology, London, 40.5, 6355655.
Millar, T. J., Winder, C., Ishimoto, I. SC Morgan, I. G. (1988). Putative serotonergic bipolar and amacrine cells in the chicken retina. Brain Research, 439, 77-87. Miller, R. F. (1979). The neuronal basis of ganglion cell receptive field organization and the physiology of amacrine cells. In Schmitt, F. 0. & Worden, F. G. (Eds.), The neuroscience fourth study program (pp. 227-245). Cambridge, MA: MIT Press. Miller, R. F. & Dacheux, R. F. (1976a). Synaptic organization and ionic basis of On and Off channels in the mudpuppy retina. Journal of General Physiology, 67, 661678.
Miller, R. F. & Dacheux, R. F. (1976b). Synaptic organization and ionic basis of On and Off channels in the
mudpuppy
retina. Journal of General Physiology, 67,
679690.
Mosinger, J. L., Yazulla, S. & Studholme K. M. (1986). GABA-like immunoreactivity in the vertebrate retina: A species comparison. Experimental Eye Research, 42, 63 1644.
Naka, K.-I. (1976). Neural circuitry in the catfish retina. Investigative Ophthalmology, 15, 926.
Naka, K.-I. (1977). Functional organization of the catfish retina. Journal of Neurophysiology, 40, 2643. Slaughter, M. M. & Miller, R. F. (1981) 2-Amino-4phosponobutyric acid: A new pharmacological tool for retina research. Science, 2f I, 1822185. Slaughter, M. M. & Miller, R. F. (1983a). Bipolar cells in the mudpuppy retina use an excitatory ammo acid neurotransmitter. Nature, London, 303, 537-538. Slaughter, M. M. & Miller, R. F. (1983b). The role of excitatory amino acid transmitters in the mudpuppy retina: An analysis with kainic acid and N-methyl aspartate. Journal of Neuroscience, 3, 1701-171 I. Toyoda, J.-I. St Fujimoto, M. (1984). Application of transretinal current stimulation for the study of bipolaramacrine transmission. Journal qf General Physiology, 84, 915-925.
Van Haesendonck, E. & Missotten, L. (1983a). Stratification and square pattern arrangements in the dorsal inner plexiform layer in the retma of Callionymus lyra L. Journal of Ultrastructure Research, 83, 296302. Van Haesendonck, E. & Missotten, L. (1983b). Interbipolar contacts in the dorsal inner plexiform layer in the retina of Callionymus lyra L. Journal of Ultrastructure Research, 83, 303-3 11. Van Haesendonck, E. & Missotten, L. (1990). Glutamatelike immunoreactivity in the retina of the marine teleost, the dragonet. Neuroscience Letters, III, 281-286. Venable, J. H. & Coggeshall, R. (1965). A simplified lead citrate stain for use in electron microscopy. Journal of Cell Biology, 25, 407-408.
Wlssle, H. & Chun, M. H. (1989). GABA-like immunoreactivity in the cat retina: Light microscopy. Journal of Comparative Neurology, 279, 43-54.
Wiissle, H., Voigt, T. & Patel, B. (1987). Morphological and immunocytochemical identification of indoleamineaccumulating neurons in the car retina. Journal of Neuroscience,
7, 1574-l 585.
Weiler, R. & Schtitte, M. (1985). Morphological and pharmacological analysis of putative serotonergic bipolar and amarcrine cells in the retina of a turtle, Pseudemys scripta elegans. Cell and Tissue Research, 241, 373-382.
Yang, C.-Y. & Yazulla, S. (1988a). Light microscopic localization of putative glycinergic neurons in the larval tiger salamander retina by immunocytochemical and autoradiographic methods. Journal of Comparative Neurology, 272, 343-357.
Yang, C.-Y. & Yazulla, S. (1988b). Locahzation of putative GABA-ergic neurons in the larval tiger salamander retina by immunocytochemical and autoradiographic methods. Journal of Comparative Neurology, 277, 96108.
0042-6989/91 $3.00+ 0.00
Copyright 0 1991Pergamon Press plc
PATTERNS OF GLUTAMATE-LIKE IMMUNOREACTIVE BIPOLAR CELL AXONS IN THE RETINA OF THE MARINE TELEOST, THE DRAGONET E. VAN HAJSENDONC~ and L. MSSOTTEN Eye Research Laboratory, Catholic University of Leuven, Kapucijnenvoer 33, B-3000 Leuven, Belgium (Received 28 December
1989)
Ahatraet-Oblique 1 pm-sections through the dorsal inner plexiform layer of the light-adapted dragonet retina were processed for ~stem~ding, silver-enhanced immunogold labeling after incubation with a glutamate-specific antiserum. Light microscopy showed strongly immunolabeled boutons grouped into distinct square patterns. These patterns were compared with the succesive grids of bipolar axonal boutons revealed by electron microscope analysis of serial, oblique sections through the entire dorsal inner plexiform layer. With one exception, all types of patterned bipolar synaptic boutons could be clearly identified in the immunoreactive staining pattern. The elevated levels of endogenous glutamate in most bipolar synaptic boutons suggest that the large majority of bipolar cell types use glutamate as their neurotransmitter. However, some bipolar synaptic boutons displaying low levels of glutamate indicate that a small number of bipolar cells may contain another neuroactive substance. Retina
Bipolar cells
Inner plexiform layer
Glutamate
A line of electrophysiological and pharmacological evidence suggests that ON- and OFFcenter bipolar cells in the vertebrate retina are excitatory in nature and that L-glutamate (Glu) is the most likely neurotransmitter of bipolar cell synapses in the inner plexiform layer (IPL) (Naka, 1976, 197’7; Miller & Dacheux, 1976a, b; Miller, 1979; Slaughter & Miller, 198 1, 1983a, b; Toyoda & Fujimoto, 1984; Bloomfield & Dowling, 1985; Lukasiewicz & McReynolds, 1985; Coleman, Massey & Miller, 1986; Massey & Miller, 1988). Consistent with this hypothesis, antisera against Glu strongly label bipolar cells and their axonal synaptic boutons. This indicates that significantly enhanced levels of endogenous Glu are present, a prerequisite for glutamatergic synaptic transmission (Ehinger, Ottersen, Storm-Mathisen & Dowling, 1988; Marc, Massey, Kalloniatis & Basinger, 1989). By virtue of its regular pattern arrangements and the certainty with which bipolar axonal boutons can be identified, the retina of the marine teleost, the dragonet, is a very suitable model to study bipolar cell neurotransmitter candidates. The dragonet retina consists of a ventral mixed and a dorsal pure-cone half. In the dorsal half, where higher neuron densities occur, bipolar cell axons display three distinct square patterns at defined levels in the IPL.
Immunocytochemistry
Teleost
These patterns mirror the photoreceptor cell disposition: the narrow S-pm pattern corresponds to the arrangement of the double cone pairs; of the two wider 7-pm patterns, one corresponds to the single cone distribution, while the other coincides with the intersections of the rows of double cone pairs (Van Haesendonck & Missotten, 1983a). We have shown that bipolar cells in the ventral and the dorsal region of the dragonet retina are strongly immunolabeled with antiGlu. On semithin vertical sections of the dorsal retina, the five levels with geomet~cally arrayed bipolar synaptic boutons are clearly identifiable by their prominent Glu-like immunoreactivity (GLIR) (Van Haesendonck & Missotten, 1990). In the present report, we complement our previous results by determining the extent to which bipolar cell types in the dorsal retina are immunoreactive for anti-Glu. We show the localization of endogenous Glu on semithin oblique sections of the dorsal IPL, processed for postembedding Glu-immunocytochemistry. We compare the arrangement of immunolabeled synaptic boutons with the consecutive square patterns of bipolar axonal boutons revealed by electron microscope analysis of serial, ultrathin oblique sections through the entire IPL. The correlation allows us to judge whether all the geomet~~lly arrayed bipolar cells are immunoreactive for anti-Glu. We find that all but one
451
E. VAN HAESENDONCK and L. MKSOTTEN
452
Dragonets (Callionymus lyra L.) were caught in the North Sea. Light-adapted fish were killed by cervical transection and pithing. After enucleation, the retinas were isolated and fixed for 2 hr by immersion in cold 2.5% glutaraldehyde on 0.1 M sodium cacodylate buffer (pH 7.4) supplemented with 3% sucrose. They were then rinsed overnight in 0.1 M phosphate buffer (PB) (pH 7.4) supplemented with 3% sucrose (PBS). All retinas intended for light microscopic immunocytochemistry were dehydrated over increasing concentrations of ethanol and, with propylene oxide as intermediary, embedded in epoxyresin (Ladd, U.S.A.). Retinas from other animals, intended for electron microscopy, were postfixed for 2 hr in 2% 0~0, in PB before dehydration and embedding in epoxyresin.
showed selective recognition of the Gluconjugate. The immunoreaction was abolished by preabsorption of anti-Glu with Gluconjugate, and when anti-Glu was omitted in the immunocytochemical procedure. Sections were first etched with sodium ethoxide for 15 min. After extensive washing in absolute ethanol, PB and distilled water, they were preincubated for 2 hr in 3% normal goat serum in PBS with 0.3% triton X-100 (PSBTx). Subsequently, the sections were washed in PBSTx (2 x 1 hr) and incubated for 3 hr in goat anti-rabbit antiserum conjugated to 5 nm colloidal gold particles (Auroprobe EM G5, Janssen, Belgium), diluted 1: 50 in the same diluent as the primary antiserum. The sections were washed in PBS (2 x 1 hr) and fixed in 2.5% glutaraldehyde in PBS for 10 min. After several rinses in PB and distilled water, they were treated with IntenSE M (Janssen, Belgium) to silverenhance the immunogold labeling. All steps in this immunocytochemical procedure were performed at room temperature. After a final rinse in three changes of distilled water, the sections were lightly counterstained with buffered toluidine blue and coverslipped with epoxyresin. They were examined with a x 40 oil immersion planapochromatic objective (Zeiss, F.R.G.).
Immunocytochemistry
Electron microscopy
Vertical and oblique l-pm sections were collected on glass slides and processed acccording to a method (modified from Danscher & Norgaard, 1983; Holgate, Jackson, Cowen & Bird, 1983) for postembedding, silver-intensified immunogold labeling after incubation with an anti-Glu antiserum (anti-Glu). The anti-Glu (Biosys, France) was raised in rabbit against a Glu-glutaraldehyde-thyroglobulin conjugate. Tested in dot-immunoassays with serial dilutions of synthesized Glu, aspartate, glutamine, GABA and glycine conjugates, the anti-Glu
Serial, ultrathin oblique sections from the inner nuclear layer (INL) to the ganglion cell layer (GCL) were collected on formvar-coated, single-slot grids. They were stained with lead citrate for 8 min (Venable & Coggeshall, 1965) and examined with a Zeiss 9S-2 electron microscope.
of these bipolar cell types in the dragonet retina contain elevated Glu levels. Furthermore, bipolar cell types with axons terminating in fibrous layers of the dorsal IPL, particularly the most distal and proximal, also have synaptic boutons enriched with endogenous Glu. MATERIALS
AND METHODS
Tissue preparation
RESULTS
Figure la serves to illustrate the location of the pattern levels (P-levels) in the IPL on a
Fig. I (Opposite). (a) Distribution of glutamate-like immunoreactivity (GLIR) on a vertical I-pm section through the dorsal dragonet retina. Strong labeling is seen over the photoreceptor cell terminals (PCT) and over the bipolar and ganglion cell bodies. Among the bipolar cell bodies, a number of moderately labeled profiles occur (white arrows). In the inner plexiform layer (IPL), sequences of regularly spaced immunoreactive boutons are observed at five levels (PI-PS). The three larger bouton types at the P3 level are immunostained. A BP3, 1 bouton (small arrowhead) lies adjacent to a BP3, 2d bouton (arrow). On the left and the right of this pair lies a BP3, 2p bouton (large arrowheads). GCL, ganglion cell layer; INL, inner nuclear layer. Scale bar = I2 pm. (b) Distribution of GLIR on an oblique I-pm section through the IPL in the dorsal dragonet retina. The inner plexiform layer (IPL) is moderately labeled. Five levels (PI-PS) with distinct types of strongly immunoreactive and square-patterned boutons are observed. Bouton profiles in the Fl and particularly in the F6 fibrous layer are also strongly immunoreactive. No labeling is seen over the inner nuclear layer (INL) or the distal row of small cell bodies in the ganglion cell layer (GCL). Scale bar = I2 pm.
INL _ Fl Pi.
Pi.
P3. IPL P4
P5
F6
GCL Fig. 1
Fig. 2. Bipolar cell organization at the PI level. (a) Electron micrograph of the Pt , 1 layer. Bipolar synaptic boutons with high synaptic vesicle density (BP& 1 encircled) am grouped by a 5-pm square. Scale bar = 2 pm. (b) Electron micrograph of the PI, 2 layer. Bipolar synaptic boutons are less densely tilled with synaptic vesicles (BPl. 2 encircled) and grouped by a 7-pm square. The 5-pm and the 7-pm square differ in orientation by 45 deg (indicated by the arrows). Scale bar = 2 pm. (c). Light micrograph of the GLIR pattern at the Pl level. Two types of strongly labeled boutons are seen: small boutons arranged according to a narrow square and, more proximally, larger boutons arranged in a wider square. The orientation of both squares also differs over 45 deg. l-Km section, scale bar = 12 pm. 454
Fig. 3. Bipolar cell organization at the P2 level. (a) Electron micrograph of the Pl ,2 layer. Bipolar synaptic boutons with high synaptic vesicle density (BP2, 1 encircled) are grouped by the 5-pm square. Scale bar = 2 pm. (b) Electron ~cro~aph of the P2,2 Iayer. Irregularly shaped boutons with broad collaterals (BP2, 2 encircled) are also grouped by the S-pm square (neigh~~ng BP2, 1 boutons in the upper right hand comer are indicated by a dashed line). Scale bar = 2 pm. (c) Light micrograph of the GLIR pattern at the P2 level. A single pattern of small, round boutons grouped by the narrow square is observed. I-pm section, scale bar = 12 pm.
455
Fig. 4. Bipolar cell organization at the P3 level. (a) Electron micrograph of the P3 level. Four types of bipolar synaptic bouton are observed. Most distally, boutons with low synaptic vesicle density (BP3, I encircled by dotted line) are grouped by a 7-pm square. Immediately adjacent proximally are boutons with high synaptic vesicle density (BP3,2d encircled by dashed line), arrayed according to the same 7-pm square pattern. Larger boutons, less densely filled with synaptic vesicles (BP3, 2p encircled by solid line) are grouped by another, distinct 7-pm square. Finally, small boutons with high synaptic vesicle density (BP3, 2s encircled by solid line) show S-pm square arrangement. Scale bar = 2 pm. (b) Light micrograph of the GLIR staining at the P3 level. Square patterns with small and large, strongly labeled boutons interfere. Arrowheads indicate BP3, 2s boutons. I-ym section, scale bar = 12 pm.
456
Fig. 5. Bipolar cell organization at the P4 level. (a) Electron micrograph of the P4 level. A single type of bipolar synaptic boutons, densely filled with synaptic vesicles and showing Sprn square arrangement is observed (BP4 encircled). Scale bar = 2 pm. (b) Light micrograph of the GLIR at the P4 level. Likewise, a single narrow square pattern of small, strongly labeled boutons is noted. l-pm section, scale bar = 12pm.
Glutamate-like immuroreactive bipolar cell axons
vertical l-pm section of the dorsal dragonet retina which has been processed for anti-Glu immunocytochemistry. Five levels with regularly grouped and strongly labeled boutons are discerned (PI-PS). Prominent GLIR is also observed over the bipolar cell bodies in the INL, the photoreceptor terminals (cone pedicles) and the ganglion cell bodies (Van Haesendonck & Missotten, 1990). A number of cell body profiles in the bipolar cell half of the INL appear to be less intensely labeled (white arrows). Figure lb shows the distribution of GLIR on an oblique 1 pm-section of the dorsal IPL. The IPL is moderately labeled. At successive levels, grids of strongly immunoreactive boutons are observed. The location of these grid layers in the IPL coincides with the P-levels of squarepatterned bipolar axonal boutons (Pl-PS). In some layers (e.g. at the P2 and the P4 level) a single square pattern occurs, in others (e.g. at the P3 and the P5 level) overlapping square patterns are seen. In addition, globular immunoreactive bouton profiles are observed in some fibrous layers (F-layers) that separate the P-levels. They are clearly noted in the thicker Fl and F6 layer. Particularly in the F6 layer, they are large and numerous. These boutons are not geometrically arrayed. The electron micrographs of the layers displaying square-patterned bipolar axonal boutons in the dorsal IPL are presented in Figs 2-6. A variety of boutons are observed and their bipolar nature is established by the presence of synaptic ribbons. The terminology of the boutons is the same as in previous studies (Van Haesendonck Jz Missotten, 1983a, b). Boutons are named after the layer in which they are located and, in cases where more than one type occurs in a single layer, reference is made to a major morphological feature (e.g. d, boutons with dark appearance due to high synaptic vesicle density). For comparison with the GLIR patterns, the accompanying light
459
micrographs (Figs 2c, 3c, 4b, 5b and 6c) in each case illustrate the disposition of immunoreactive boutons in that particular P layer. At the Pl level (at 15-20% in the IPL), two types of bipolar boutons occur, each arranged according to a different square pattern (Fig. 2). The medium-sized, round boutons (BP1 , I ; approx. 2.8 pm dia.) in the distal layer have a high synaptic vesicle density and show the j-pm square disposition (Fig. 2a). The larger, round boutons (BPl, 2; approx. 4.5 pm dia.) in the proximal layer are less densely filled with synaptic vesicles and show a 7 pm square disposition (Fig. 2b). The GLIR pattern at the PI level (Fig. 2c) includes two types of intensely labeled boutons, grouped by two distinct square patterns. Smaller immunoreactive boutons are arranged according to a narrow (5 pm) square, larger immunoreactive boutons are arranged according to a wider (7 pm) square. On the electron micrographs of the Pl layers, as well as in the GLIR pattern, a 45 deg difference in orientation between the two square patterns is noted. The PZlevel (at 30-35% in the IPL) likewise consists of two layers, each containing a single bouton type (Fig. 3). The distally located, round boutons (BP2, 1; approx. 2.8 pm dia.) with high synaptic vesicle density resemble the BPl, 1 boutons and are also grouped by the j-pm square (Fig. 3a). Immediately proximal lies a second layer, containing boutons with the same j-pm square disposition. These boutons (BP2,2) are flattened, with broad, irregularly shaped collaterals, and have a low synaptic vesicle density (Fig. 3b). Figure 3c shows the distribution of GLIR boutons at the P2 level. Important to note is the absence of two distinct GLIR grids at this level. Only a single, narrow square pattern of small, round and strongly labeled boutons is observed. The P3 level (at 45-50% in the IPL) is complex. At this level, four types of boutons
Fig. 6 (Opposite). Bipolar cell organization at the PS-level. (a) Electron micrograph of the P5, I layer. Two types of bipolar synaptic boutons are observed. Boutons with low synaptic vesicle density are grouped by the 5-pm square (BPS, lp encircled by solid line). Boutons with high synaptic vesicle density are group by a 7-pm square (BPS, Id encircled by dashed line). These bouton types mutually contact (arrows). Scale bar = 2 pm. (b) Electron micrograph of the PS, 2 layer. Again, two types of bipolar synaptic boutons occur. Large boutons with variable synaptic vesicle density (BP5,2b encircled by solid line) have the 5-pm square arrangement. Small boutons with high synaptic vesicle density (BP5,2s encircled by dashed line) have a 7-pm square arrangement. Scale bar = 2 pm. (c) Light micrograph of the GLIR pattern at the P5 level. In the distal P5, 1 layer, a lattice-like arrangement of strongly labeled boutons is seen. In the proximal P5, 2 layer, small and large strongly immunostained boutons with distinct square arrangement are observed. l-pm section, scale bar = 12 pm.
460
E. VAN HAESENWNCKand L. MISSOTTEN
occur, and the three distinct square patterns interfere. Two poorly separated layers can be distinguished. A distal layer contains mediumsized, round boutons (BP3, 1; approx. 3.0pm dia.) with low synaptic vesicle density and grouped by a 7-pm square pattern. The closely adjoining proximal layer has been described in detail previously (Van Haesendonck & Missotten, 1983a). This layer contains three bouton types. The largest boutons, with moderate synaptic vesicle density (BP3, 2p; approx. 4pm dia.) and the medium-sized boutons with high synaptic vesicle density (BP3, 2d; approx. 3.0,um dia.) are patterned according to two different 7-pm squares. The small boutons with high synaptic vesicle density (BP3, 2s; approx. 1.3 pm dia.) form a 5-pm grid. Figure 4b illustrates the distribution of GLIR boutons at the P3 level. Boutons of different sizes are visible whose dispositions are unmistakably based upon square patterns. Their overall arrangement, however, appears somewhat disordered. The P4 level (at 5560% in the IPL) is the sole pattern level at which single type of boutons occurs (Fig. 5). These boutons are rather small (BP4; approx. 2pm dia.), contain many synaptic vesicles (Fig. 5a), and have the 5-pm square disposition. Figure 5b shows the arrangement of boutons immunolabeled with anti-Glu at this level. A narrow square pattern of small GLIR boutons is observed. The P5 level (at 70-75% in the IPL) is the most extensive pattern level (Fig. 6). It consists of a distal layer (P5, 1) with two bouton types and a proximal layer (P5,2), also containing two boutons types. Both layers are separated by a very thin fibrous layer. In the P5, 1 layer (Fig. 6a), the two bouton types are quite large (approx. 3.0 pm dia.). One type has a moderate to low synaptic vesicle density (BP5, lp) and 5-pm square disposition. The other type has a high synaptic vesicle content (BPS, Id) and a 7-pm square disposition. These dissimilar boutons have considerable contact zones with specialized cell membranes (Fig. 6a, arrows; Van Haesendonck & Missotten, 1983b). The corresponding pattern of GLIR boutons is shown in Fig. 6c. The immunoreactive boutons also appear to contact one another, forming a lattice-like structure. Labeled boutons in narrow squares are more often observed in this layer. In the P5, 2 layer (Fig. 6b), the large boutons (BP5, 2b; approx 4pm dia.) have a variable synaptic vesicle content. They
are grouped into the 5-pm square pattern, The small boutons in this layer (BP5, 2s; approx. 1.3 ,um dia.) have a high synaptic vesicle density and are arranged according to a 7-pm square pattern. The immunostaining pattern of the P5, 2 layer (Fig. 6c) shows a large and a very small type of labeled boutons. The former displays the narrow square pattern, while the latter is arranged according to the wider square pattern. DISCUSSION
As anticipated, the present results confirm that the regularly spaced GLIR boutons, observed on vertical sections of the dorsal IPL in the dragonet retina (Van Haesendonck & Missotten, 1990), display square patterns. The different levels with geometrical disposition of immunolabeled boutons correspond to the levels with square patterned bipolar synaptic boutons, observed by conventional electron microscopy of the dorsal IPL. These observations emphasize the bipolar cell origin of the immunostained boutons. To exactly what extent can the individual types of square-patterned bipolar boutons be identified in the GLIR patterns? Careful comparison of GLIR and electron microscopically observed grids at the successive P-levels indicates that all but one type of patterned boutons is strongly immunoreactive for anti-Glu and can be readily identified by the size differences and the distinct spacings. The boutons at the Pl, P4 and P5 level are the most easily correlated. The size and the disposition of the boutons in the GLIR patterns at these levels effectively correspond to those of the bipolar synaptic boutons in the electron microscopically observed patterns. Also, the 45 deg difference in orientation between the narrow and the wide square patterns is obvious in the arrangement of the GLIR boutons. At the P3 level, interposition of the three square patterns results in a mingled arrangement of bipolar axonal boutons. Likewise, the corresponding pattern of immunoreactive boutons has a complicated, rather confusing appearance. Moreover, the similarity in size among some bouton hampers identification of different types within the GLIR pattern, and makes correlation with EM images difficult. However, combining observations on vertical and oblique sections allows us to conclude that all four bouton types that occur at the P3 level show GLIR. Figure la indicates that the three
Glutamate-like immuroreactive bipolar cell axons
larger types are strongly labeled. The two vertically adjacent boutons (small arrowhead and arrow) can be correlated with the BP3, I and the BP3, 2d boutons. The boutons on the left and the right of these (large arrowheads) most likely correspond to BP3, 2p boutons. Furthermore, Fig. 4c (arrowheads) shows small, regularly distributed, immoreactive boutons that interfere with the arrangement of the large boutons. These correspond to the BP3, 2s boutons. One type of patterned boutons, namely the BP2, 2 boutons at the P2 level, are not clearly discernible. We correlate the round, labeled boutons in the GLIR pattern to the distal, round BP2, 1 boutons. It can be argued that the BP2, 1 and the BP2, 2 boutons have the same square pattern and, thus, are not easily distinguished from one another. Because of the characteristic shape of the BP2, 2 boutons with broad collaterals though, this type should be recognizable as a distinct GLIR lattice. Such a lattice is not observed. Also, a single layer of prominently immunostained boutons is observed at the P2 level on vertical sections. We conclude that BP2, 2 boutons are not labeled, or only very weakly labeled with antiGLU. The absence of intense immunostaining could be explained by the low synaptic vesicle content of these boutons. However, BP3, 1 and BP5, lp boutons have similar low synaptic vesicle densities and yet label much more strongly than the BP2, 2 boutons. Our results indicate that in the light-adapted retina, the endogenous Glu level in the BP2, 2 boutons is not markedly elevated. Interestingly, Ehinger et al. (1988) also observed a small number of bipolar cell terminals in the turtle IPL that showed little or no labeling. A number of the weakly labeled cell bodies in the distal half of the INL may belong to the bipolar cells with BP2, 2 axon terminals. The weakly labeled population of cell bodies probably includes somata of Miiller cells as well. In addition to the square-patterned bipolar boutons, large GLIR bouton profiles also occur in the fibrous layers between the P levels. These boutons are predominantly located in the Fl and most abundantly in the F6 layer. They most likely correspond to the synaptic boutons of bipolar axons without the square pattern disposition, observed by electron microscopy (Van Haesendonck & Missotten, 1983a, 1990). Such bipolar boutons have also been found in the F2 and the F3 layer. In these thinner fibrous layers, the less numerous and smaller
461
boutons are hard to detect by GLIR. Therefore, we do not exclude the possibility that some of these boutons are not immunoreactive for anti-Glu. If we generalize the electrophysiological results of Famiglietti, Kaneko and Tachibana (1977) in the carp retina and accept the subdivision into a distal OFF-sublamina and a proximal ON-sublamina for the dragonet IPL, then our present results once again show that GLIR boutons occur in both sublaminae (Van Haesendonck & Missotten, 1990). This suggests that ON- as well as OFF-center bipolar cell synaptic terminals are enriched with endogenous Glu and agrees with Glu-immunocytochemistry studies in other vertebrate retinas (Ehinger et al., 1988; Marc et al., 1989). In conclusion, most bipolar axonal boutons in the dorsal half of the light-adapted retina label strongly for anti-Glu. This indicates that elevated levels of endogenous Glu are present in the boutons and supports the idea that the large majority of bipolar cell types in this purecone region use Glu as their neurotransmitter. Our results leave open the possibility that a patterned bipolar cell type with axonal terminal (BP2, 2 boutons) in the distal sublamina of the IPL (OFF-sublamina) may contain a different or an additional neuroactive substance. In the teleost retina, no other classical neurotransmitter has been demonstrated in bipolar cells by immunocytochemistry. In other vertebrate classes, on the other hand, serotonin-, glycineand GABA-like immunoreactivity has been observed in bipolar cells (Bruun, Ehinger & Sytsma, 1984; Weiler & Schiitte, 1985; Mosinger, Yazulla & Studholme, 1986; Agardh, Ehinger & Wu, 1987; Wassle, Voigt & Patel, 1987; Millar, Winder, Ishimoto & Morgan, 1988; Yang & Yazulla, 1988a, b; Hendrickson, Koontz, Pourcho, Sarthy & Goebel, 1988; WZissle & Chun, 1989). As of yet, the identity of a possible second neurotransmitter substance in some bipolar cells of the dragonet retina remains to be determined. Acknowledgements-We thank Professor F. Ollevier and Dr P. Van Damme, K. U. Leuven Zoological Institute, for their cooperation with the collection of dragonets, Mrs A. Loenders for technical assistance and Dr S. Raiguel for helpful discussions. REFERENCES Agardh E., Ehinger B. & Wu J.-Y. (1987). GABA and GAD-like immunoreactivity in the primate retina. Histochemistry, 86, 485490.
462
E. VAN HAE~ENDONCK and L. MIS~OTTEN
Bloomfield, S. A. & Dowling, J. E. (1985). Roles of aspartate and glutamate in synaptic transmission in rabbit retina II. Inner plexiform layer. Journal of Neurophystology, 53, 714125.
Bruun, A., Ehinger, B. & Sytsma, U. M. (1984). Neurotransmitter localization in the skate retina. Brain Research, 295, 233-248.
Coleman, R. A., Massey, S. C. & Miller, R. F. (1986). Kynuremc acid distinguishes kainate and glutamate receptors in the vertebrate retina. Brain Research, 381, 172-175.
Danscher, G. & Norgaard, J. 0. D. (1983). Light microscopic vizuahzation of colloidal gold on resin-embedded tissue. Journal of Histochemistry and Cytochemistry, 31, 13941398.
Ehinger, B., Ottersen, 0. P., Storm-Mathisen, J. & Dowling, J. (1988). Bipolar cells in the turtle retina are strongly immunoreactive for glutamate. Proceedings of the National Academy of Sciences, U.S.A., g&8321-8325.
Famiglietti, E. V. Jr, Kaneko, A. & Tachibana, M. (1977). Neuronal architecture of ON and OFF pathways to ganglion cells in carp retina. Science, 198, 126771269.
Hendrickson, A. E., Koontz, M. A., Pourcho, R. G., Sarthy, P. V. & Goebel, D. (1988). Localization of glycine-containing neurons in the Macaca monkey retina. Journal of Comparative Neurology, 273, 473487.
Holgate, C. S., Jackson, P., Cowen, P. N. & Bird, C. C. (1983). Immunogold-silver staining: A new method of immunostaining with enhanced sensitivity. Journal of Histochemistry and Cytochemistry, 31, 938-944.
Lukasiewicz, P. D. & McReynolds, J. S. (1985). Synaptic transmission at N-methyl-o-aspartate receptors in the proximal retina of the mudpuppy. Journal of Physiology, London, 367, 99-116.
Marc, R. E., Massey, S. C., Kalloniatis, M. & Basinger, S. F. (1989). Immunochemical evidence that the fast neurotransmitter of rods, cones, bipolar and ganglion cells is glutamic acid. Investigation Opthalmology & Visual Science (Suppl.), 30, 320. Massey, S. M. & Miller, R. F. (1988). Glutamate receptors of ganglion cells in the rabbit retina: Evidence for glutamate as a bipolar cell transmitter. Journal of Physiology, London, 40.5, 6355655.
Millar, T. J., Winder, C., Ishimoto, I. SC Morgan, I. G. (1988). Putative serotonergic bipolar and amacrine cells in the chicken retina. Brain Research, 439, 77-87. Miller, R. F. (1979). The neuronal basis of ganglion cell receptive field organization and the physiology of amacrine cells. In Schmitt, F. 0. & Worden, F. G. (Eds.), The neuroscience fourth study program (pp. 227-245). Cambridge, MA: MIT Press. Miller, R. F. & Dacheux, R. F. (1976a). Synaptic organization and ionic basis of On and Off channels in the mudpuppy retina. Journal of General Physiology, 67, 661678.
Miller, R. F. & Dacheux, R. F. (1976b). Synaptic organization and ionic basis of On and Off channels in the
mudpuppy
retina. Journal of General Physiology, 67,
679690.
Mosinger, J. L., Yazulla, S. & Studholme K. M. (1986). GABA-like immunoreactivity in the vertebrate retina: A species comparison. Experimental Eye Research, 42, 63 1644.
Naka, K.-I. (1976). Neural circuitry in the catfish retina. Investigative Ophthalmology, 15, 926.
Naka, K.-I. (1977). Functional organization of the catfish retina. Journal of Neurophysiology, 40, 2643. Slaughter, M. M. & Miller, R. F. (1981) 2-Amino-4phosponobutyric acid: A new pharmacological tool for retina research. Science, 2f I, 1822185. Slaughter, M. M. & Miller, R. F. (1983a). Bipolar cells in the mudpuppy retina use an excitatory ammo acid neurotransmitter. Nature, London, 303, 537-538. Slaughter, M. M. & Miller, R. F. (1983b). The role of excitatory amino acid transmitters in the mudpuppy retina: An analysis with kainic acid and N-methyl aspartate. Journal of Neuroscience, 3, 1701-171 I. Toyoda, J.-I. St Fujimoto, M. (1984). Application of transretinal current stimulation for the study of bipolaramacrine transmission. Journal qf General Physiology, 84, 915-925.
Van Haesendonck, E. & Missotten, L. (1983a). Stratification and square pattern arrangements in the dorsal inner plexiform layer in the retma of Callionymus lyra L. Journal of Ultrastructure Research, 83, 296302. Van Haesendonck, E. & Missotten, L. (1983b). Interbipolar contacts in the dorsal inner plexiform layer in the retina of Callionymus lyra L. Journal of Ultrastructure Research, 83, 303-3 11. Van Haesendonck, E. & Missotten, L. (1990). Glutamatelike immunoreactivity in the retina of the marine teleost, the dragonet. Neuroscience Letters, III, 281-286. Venable, J. H. & Coggeshall, R. (1965). A simplified lead citrate stain for use in electron microscopy. Journal of Cell Biology, 25, 407-408.
Wlssle, H. & Chun, M. H. (1989). GABA-like immunoreactivity in the cat retina: Light microscopy. Journal of Comparative Neurology, 279, 43-54.
Wiissle, H., Voigt, T. & Patel, B. (1987). Morphological and immunocytochemical identification of indoleamineaccumulating neurons in the car retina. Journal of Neuroscience,
7, 1574-l 585.
Weiler, R. & Schtitte, M. (1985). Morphological and pharmacological analysis of putative serotonergic bipolar and amarcrine cells in the retina of a turtle, Pseudemys scripta elegans. Cell and Tissue Research, 241, 373-382.
Yang, C.-Y. & Yazulla, S. (1988a). Light microscopic localization of putative glycinergic neurons in the larval tiger salamander retina by immunocytochemical and autoradiographic methods. Journal of Comparative Neurology, 272, 343-357.
Yang, C.-Y. & Yazulla, S. (1988b). Locahzation of putative GABA-ergic neurons in the larval tiger salamander retina by immunocytochemical and autoradiographic methods. Journal of Comparative Neurology, 277, 96108.