Immunocytochemical Identification of

Transcription

1 THE JOURNAL OF COMPARATIVE NEUROLOGY 361: (1995) Immunocytochemical Identification of Cone Bipolar Cells in the Rat Retina THOMAS EULER AND HEINZ WASSLE Max-Planck-Insitut fur Hirnforschung, D Frankfurt, Germany ABSTRACT We studied the morphology of bipolar cells in fixed vertical tissue sections of the rat retina by injecting the cells with Lucifer Yellow and neurobiotin. In addition to the rod bipolar cell, nine different putative cone bipolar cell types were distinguished according to the position of their somata in the inner nuclear layer and the branching pattern and stratification level of their axon terminals in the inner plexiform layer. Some of these bipolar cell populations were labeled immunocytochemically in vertical and horizontal sections using antibodies against the calcium-binding protein recoverin, the glutamate transporter GLT-1, the alpha isoform of the protein kinase C, and the Purkinje cell marker L7. These immunocytochemically labeled cell types were characterized in terms of cell density and distribution. We found that rod bipolar cells and GLT-1-positive cone bipolar cells occur at higher densities in a small region located in the upper central retina. This area probably corresponds to the central area, which is the region of highest ganglion cell density. A second peak of rod bipolar cell density in the lower temporal periphery matches the retinal area of binocular overlap. The population densities of the immunocytochemically characterized bipolar cells indicate that at least 50% of all bipolar cells are cone bipolar cells. The variety and total number of cone bipolar cells is surprising because the retina of the rat contains 99% rods. Our findings suggest that cone bipolar cells may play a more important role in the visual system of the rat than previously thought. o 1995 Wiley-Liss, Inc. Indexing terms: mammalian retina, recoverin immunoreactivity, glutamate transporter (GLT-1) immunoreactivity, PKC immunoreactivity, L7 immunoreactivity Two basic types of bipolar cells are present in mammalian retinae: cone bipolar cells (CBs) and rod bipolar cells (RBs; Cajal, 1893). Bipolar cells have a dendritic tree in the outer plexiform layer (OPL) where they contact photoreceptor terminals and an axon that terminates in the inner plexiform layer (IPL) and makes synapses with ganglion cells and/or amacrine cells (Dowling and Boycott, 1966; Boycott and Dowling, 1969). Cone bipolar cells can be subdivided according to their morphology into several classes. These classes differ in the dendritic branching pattern, the number of cones contacted, shape and stratification of the axon terminal systems in the IPL, and the position of the cell body in the inner nuclear layer (INL; Boycott and Kolb, 1973; Famiglietti, 1981). Gene bipolar cells contact both amacrine cells and ganglion cells directly. In all vertebrate retinae, cone bipolar cells show a physiological dichotomy: ON-type CBs are depolarized, whereas OFF-type CBs are hyperpolarized by a light stimulus projected into their receptive field center (Kaneko, 1970; Toyoda, 1973; Famiglietti and Kolb, 1976; Nelson and Kolb, 1983; Saito, 1987). The physiological ONand OFF-type cells are probably correlated to an inner (sublamina b) and outer (sublamina a) stratification of their axon terminal systems, respectively (Famiglietti and Kolb, 1976; Nelson et al., 1978; Peichl and Wassle, 1981). Only one morphological type of rod bipolar cell seems to exist in the mammalian retina and can be distinguished from cone bipolar cells by several criteria. The dendritic trees of RBs are more bushy and their dendrites are finer and penetrate further into the outer nuclear layer (ONL), where they innervate rod spherules (Cajal, 1893; Polyak, 1941; Boycott and Dowling, 1969; Nelson and Kolb 1983). The somata of RBs are larger and are located in the outer half of the INL close to the OPL; their axons run through the IPL and end in large lobulated terminals at the border of the IPL and ganglion cell layer (GCL; Cajal, 1893). Rod bipolar cells are very likely to be of the physiological ON type (Dacheux and Raviola, 1986; Wassle et al., 1991). They make no direct output synapses onto ganglion cells, but rather make synaptic connections onto amacrine cells (Famiglietti and Kolb, 1975; Kolb and Famiglietti, 1976). One of these is the AII amacrine cell, which makes output Accepted May 8, Address reprints requests to Heinz Wassle, Max-Planck-Institut fur Hirnforschung, Deutschordenstrasse 46, D Frankfurt, Germany. o 1995 WILEY-LISS, INC.

2 462 T. EULER AND H. WASSLE synapses onto CB terminals (Kolb and Famiglietti, 1974, Famiglietti and Kolb, 1975). The ultrastructural details of the rod pathway have been analyzed in many different mammalian retinae, and it seems likely that all mammals have the same rod bipolar cell circuitry (Wassle et al., 1991). Various antibodies have been described to label bipolar cells in the mammalian retina. Barnstable (1980; Barnstable et al., 1983) stained nearly all bipolar cells in the rat retina by using antibodies raised ag9inst crude retinal membranes (RET-Bl). In addition, an antibody raised against rabbit olfactory bulb (ROB; MAb 115A10; Onoda and Fujita, 1987) stained almost all bipolar cells in the rat retina (Onoda, 1988). Antibodies raised against calbindin (CaBP D-28K) were shown to label bipolar cells along with other cell types in the monkey retina (Rohrenbeck et al., 1989; Martin and Grunert, 1992; Griinert et al., 1994). However, in the rat retina, only a few bipolar cells are weakly stained among other cell types (Feigenspan et al., 1993b; J.H. Brandstatter, personal communication). Some recently developed antibodies are known to label bipolar cell populations-in some cases together with other retinal neurons-in the mammalian retina. An antibody raised against protein kinase C (PKC) has been found to label rod bipolar cells in various mammalian species (Negishi et al., 1988; Greferath et al., 1990; Grunert and Martin, 1991; Muller and Peichl, 1991; Young and Vaney, 1991; Zhang and Yeh, 1991). An antiserum raised against the Purkinje cell marker L7 was also found to label rod bipolar cells (Berrebi et al., 1991; Grunert and Martin, 1991). Antibodies against the calcium-binding protein recoverin were found to label distinct populations of cone bipolar cells in different mammalian retinae and revealed two types of cone bipolar cells in the rat retina (Milam et al., 1993). An antibody raised against a glutamate transporter (GLT-1) purified from the rat brain (Danbolt et al., 1992) was recently found to label cone photoreceptors and two populations of putative cone bipolar cells in the rat retina (Rauen and Kanner, 1994). Although these antibodies are available as markers of bipolar cell populations in the rat retina, a systematic examination of the bipolar cell morphologies and their spatial distributions is still missing. A brief report by Muller et al. (1993) describing a variety of bipolar types in the rat retina has not yet been followed up by a full-length paper. The rat retina is becoming a model system for in vitro studies of the mammalian retina (Karschin and Wassle, 1990; Suzuki et al., 1990; Yeh et al., 1990; Boos et al., 1993; Feigenspan et al., 1993a,b), and for this reason it is also important to know more details of rat bipolar cells. The cone-dominated retina of the macaque monkey has been shown to have 10 different types of cone bipolar cells (Boycott and Wassle, 1991; Grunert et al., 1994). Here we have investigated whether the rat retina, which is so clearly rod dominated, also has several different types of cone bipolar cells. In this paper, we present evidence for 10 different types of bipolar cells in the rat retina that we distinguish morphologically after injecting the cells with Lucifer Yellow and neurobiotin. Five of these bipolar cell types could be labeled immunocytochemically in vertical and horizontal sections. Antibodies against the calcium-binding protein recoverin, the glutamate transporter GLT-1, an isoform of PKC, and the Purkinje cell marker L7 labeled distinct populations of these bipolar cells. The labeled cells were characterized in terms of their densities and spatial distri- butions. Our data indicate that at least 50% of all bipolar cells in the rat retina might be cone bipolar cells. METHODS Tissue preparation We used retinae from adult albino rats (lab strain). The animals were deeply anesthetized with halothane and then decapitated. The eyes were removed and opened by an encircling cut along the ora serrata, and the vitreous was dissected out in HEPES buffered physiological saline containing (in mm): 137 NaCl, 5 KC1, 2 CaC12, 1 MgS04, 1 Na2HP04, 15 glucose, and 10 HEPES (ph 7.4). For tissue slices, which were used for the injections of Lucifer Yellow and neurobiotin, the retina was removed immediately from the sclera, transferred to a plastic Petri dish, and cut into quarters. One of the quarters was held with a pair of forceps, and, by rolling a scalpel blade over the edge, thin slices of km were cut (Boos et al., 1993). The slices were fixed with 4% paraformaldehyde (PA) in 0.1 M phosphate buffer, ph 7.4 (PB), for minutes at room temperature and then rinsed several times (3 x 10 minutes) in 0.1 M phosphate buffer, ph 7.4. For immunocytochemistry, the eyecup was fixed for 30 minutes in 4% PA, then the retina was carefully dissected and fixed again for 2 hours at room temperature. A sketch of the retina was made to determine its shape and the position of the optic disk. After rinsing in PB, the retinas were immersed in 30% sucrose in PB overnight before sectioning. Vertical frozen sections of 12-pm thickness were taken on a cryostat, collected onto gelatinized slides, air dried, and stored at -20 C. Horizontal frozen sections of 50- or 70-pm thickness were taken with a sliding microtome and processed free floating. Injection of cells Between 6 and 8 of the fixed slices were transferred to a filling chamber and placed with the cut side up. The slices were held in place by a grid of parallel nylon strings (Konnerth et al., 1987; Edwards et al., 1989). The filling chamber was placed on the fixed stage of a microscope (Zeiss ACM). The microscope was equipped with epifluorescence illumination and differential interference contrast (Nomarski) optics. A water immersion objective was used (Zeiss, 40/0.75 W; for further details, see Boos et al., 1993). Sharp filling electrodes were pulled from borosilicate glass (outer diameter of 1.5 mm, wall thickness of 0.3 mm; Clark, Redding, UK) on a horizontal electrode puller (DMZ- Universal Puller, Zeitz Instrumente, Augsburg, Germany). The filling electrode contained 4% neurobiotin (NB; Vector; Vaney, 1991) and 1% Lucifer Yellow (LY, dilithium salt; Sigma) in 0.1 M Tris buffer, ph 7.6. The LY was used to control the filling procedure and to visualize the filled cells during the experiment, and NB was the intracellular marker for the subsequent permanent staining (Vaney 1991). For every experiment, the filling solution was freshly prepared from frozen stock solutions. The filling electrode was controlled by a Leitz micromanipulator for coarse movement and a piezoelectric manipulator (Physik Instrumente, Waldbronn, Germany) for fine adjustment. A positive holding current of less than 1 na was applied to the electrode (Micro-iontophoresis programmer, Model 160, W-P Instruments) to prevent LY from diffusing out of the electrode tip. After the filling electrode had penetrated the cell membrane, a negatively charged ejection current (-2 to -5 na)

3 BIPOLAR CELLS IN THE RAT RETINA 463 was applied for 2 to 4 minutes to inject LY into the cell. Under fluorescent light, the morphology of the filled cell became visible and photomicrographs could be taken. The direction of the current was then reversed to inject the positively charged neurobiotin molecules for about 5 minutes. We tried to fill bipolar cells in a systematic manner over large portions of each slice. To avoid an overlap of filled processes, only a few bipolar cells in a given area were injected, but the distance between these injection sites varied in an attempt to fill as many different types as possible. In addition, we avoided filling rod bipolar cells, which could be recognized by their soma size and position. Therefore, our data do not represent a truly random sample, and cell density cannot be determined solely by the number of each type of cell injected. Staining neurobiotin-filled cells Following the injection of LY and NB into cells, the slices were fixed once more in 4% PA for 1 hour and then rinsed in phosphate buffered saline (PBS). Then the slices were incubated in Extravidin-Peroxidase (Sigma, diluted 1:200 in PBS) for 1 hour. The NB-filled cells were visualized with 0.5% 3,3-diaminobenzidine tetrahydrochloride (DAB) and 0.01% HzOz in PB (Hsu et al., 1981). The slices were mounted between two coverslips in glycerol. Antibodies A polyclonal antibody against the 23-kD calcium-binding protein recoverin (Lambrecht and Koch, 1991) was used. The antibody was kindly provided by Dr. K.-W. Koch (Institut fur Biologische Informationsverarbeitung, Julich, Germany). It was raised in rabbit against purified recoverin isolated from bovine retinal photoreceptors (Lambrecht and Koch, 1992) and was used at dilutions of 1:2,000 or 1:4,000. A polyclonal antibody raised in rabbit against the glutamate transporter GLT-1 from rat brain (Danbolt et al., 1992) was a kind gift from Dr. B.J. Kanner (The Hebrew University, Jerusalem, Israel). It was used at a dilution of 1: 1,000. A mouse monoclonal antibody against PKC was purchased from Amersham (clone MC5, RPN.536). The information supplied with the antibody states that it recognizes both type I1 (p) and type I11 (a) of PKC, although Walker et al. (1990) showed that the MC5 antibody cross reacts only with the a isozyme. For brevity we will use the terms PKC immunoreactivity or PKC-positive cells instead of type- 11-PKC or a-pkc-like immunoreactivity. The PKC antibody was used at a dilution of 1: 100. The polyclonal antiserum against the Purkinje cell differentiation marker L7 was raised in rabbit against a synthetic peptide conjugate predicted from the L7 amino acid sequence (Oberdick at al., 1988). The antibody was kindly provided by Dr. J.I. Morgan (Roche Institute, Nutley, NJ, USA) and was used at a dilution of 1:1,000. Immunocytochemistry In vertical cryostat sections and horizontal sliding microtome sections, an ABC method (Hsu et al., 1981) was used to demonstrate immunoreactivity. For some vertical sections, an indirect immunofluorescence method was used. Regardless of which procedure was followed, all steps, except for the incubation of the horizontal sections, were carried out at room temperature. All sections were preincu- bated in a solution of 10% normal goat serum (NGS) and 5% Triton X-100 in PBS for 1-2 hours. The primary antibodies were diluted in a solution of 3% NGS, 0.5% Triton, and 0.05% sodium azide in PBS, and the sections were incubated in this solution for 1-2 hours at room temperature (vertical sections) or 5-7 days at 5 C (horizontal sections). After several washes in PBS, the sections were incubated in the secondary antibody diluted in 3% NGS and 0.5% Triton in PBS for 1-2 hours. For the ABC method, the following secondary antibodies were used: biotinylated goat anti-rabbit IgG (Sigma, B-8895, diluted 1:lOO or Vector, BA-1000, diluted 1:200) for recoverin, GLT-1, and L7 and biotinylated sheep anti-mouse IgG (Amersham, RPN 1001, diluted 1:200) for PKC. After rinsing in PBS, the sections were incubated in Extravidin- Peroxidase diluted 1:200 in PBS for 1 hour. For some staining, the Vectastain Elite ABC-Kit (Vector) was used instead of the Extravidin-Peroxidase. In these cases, the sections were processed according to the description supplied with the kit. The peroxidase was visualized with 0.5% DAB and 0.01% HzOz in PB. For visualizing immunoreactivity with fluorescence, the following secondary antibodies or fluorescence markers were used: goat anti-mouse fluorescein isothiocyanate (FITC; Dianova, , diluted 1:50), goat anti-rabbit carboxymethylindocyanin-3 (Cy3; Dianova, , diluted 1:1,000), and Texas Redstreptavidin conjugate (Amersham, RPN 1233, diluted 1: 100). For double immunofluorescence, vertical cryostat sections were incubated in a mixture of two of the primary antibodies. The sections were rinsed in PBS and then incubated in a mixture of two secondary antibodies. Vertical cryostat sections were coverslipped in Mowiol (Hoechst). For the fluorescence stainings, 2.5% 1,4- diazobicyclo-2,2,2-octane (DABCO; Aldrich) was added to the Mowiol solution to prevent bleaching. Horizontal sections were mounted onto gelatine-coated slides, either dehydrated and coverslipped with Permount (Fisher) or coverslipped with Mowiol. Fluorescence microscopy Retinal sections were viewed with a fluorescence microscope (Axiophot, Zeiss). Different combinations of filters were used to photograph the immunostained sections: , FT 510, and LP 520 for FITC; BP 546, FT 580, and LP 590 for Cy3; and , FT 600, and LP 615 for Texas Red. All micrographs were taken on 35-mm black-andwhite film (Kodak Tri-X pan 400). To show double labeling, micrographs taken with the appropriate filter combinations were drawn on transparent sheets at a final magnification of x 1,000. Then, the drawings were brought into register and the colocalization was analyzed. Light microscopy The DAB-processed NB-filled bipolar cells in the vertical slices were inspected and drawn with the aid of a light microscope equipped with a camera lucida (Zeiss, Oberkochen). The drawings were made at a final magnification of ~1,000. Although the surface of the slice was not always parallel to the focal level of the microscope, the morphology of the cells could be easily resolved in most cases by using different levels of focus. In this way, the cell could be drawn and classified by its morphology. Maps of the horizontal sections were drawn at a final magnifications of either ~ 6 or 4 ~ 42. To measure the cell density, fields were drawn at a final magnification of

4 464 T. EULER AND H. WASSLE x 1,000 (PKC, recoverin, and L7) or x 625 (GLT-1). All light micrographs were taken on 35-mm black-and-white film (Agfa APX 25). Cell density measurements Fields chosen for cell density measurements contained at least 50 objects (somata or axon profiles) or covered a minimal area of 50 x 100 pm. When counting axon profiles, usually all profiles could be recognized within a single horizontal section, hence reconstruction was not necessary. In most cases, the same was true for counting somata. In a few cases, areas with labeled cells in two adjacent sections were found, and therefore reconstruction was necessary. To prevent double counting, areas containing labeled cells in two adjacent horizontal sections were omitted and the stained cells in the remaining area of each section were counted. Then the two horizontal sections were brought into register by matching the edges of the sections, the position of the optic disk, and the areas with labeled cells in more than one section. For the density maps, at least 70 fields were counted. We tried to distribute these fields uniformly over the whole retina by using a regular grid (steps of 500 Fm). At positions with limited staining quality, we tried to count the cells in a neighboring field. In areas with steeper gradients, the cells in a field between two grid positions were also counted. The horizontal sections were fixed to gelatinized slides, where shrinkage has been found negligible (Wassle et al., 1975); therefore, the density data are not corrected for shrinkage. However, some shrinkage (5-10%) may occur as a result of the initial fixation and immunocytochemical processing prior to mounting on the slides. RESULTS Bipolar cell types in the rat retina More than 100 bipolar cells were injected with LY and NB in vertical tissue sections (Fig. 1A-G). They differed in the position of their cell bodies in the INL and in the shape and stratification levels of their axons in the IPL. Camera lucida drawings of the cells were made, and characteristic examples are shown in Figure 2. We observed one type of bipolar cell with the typical rod bipolar cell morphology (RB in Fig. 2) and nine additional bipolar cell types. In all other mammals examined to date, only one morphological type of RB has been found (Wassle et al., 1991). Assuming that this is also the case for the rat retina, the nine morphologically distinct bipolar cell types are probably cone bipolar cells. Confirmation of this identification with electron microscopy is currently lacking. For brevity, we will use the term cone bipolar cell (CB) to describe those bipolar cells that differ from the classical morphology of the rod bipolar cell and that display similarity to cone bipolar cells of other mammals. However, this designation does not preclude the possibility of some input from rods to these bipolar cells. The criteria used to group CBs into different morphological types were the position of the cell body in the INL and, most importantly, the branching pattern and the stratification level of the axon terminal system within the IPL. Because the dendrites are much finer than the axon terminals and because background staining in the OPL was relatively high, criteria such as shape and width of the dendritic trees could not normally be used for classification (Fig. 1). For the same reasons, the drawings in Figure 2 probably show only parts of the dendritic trees. However, unquestionably the dendritic trees of different types of bipolar cells differ substantially in size as reflected in Figure 2. In the classification scheme shown in Figure 2, the nine CB types are arranged according to the stratification level of their axon terminal systems in the IPL. We have subdivided the IPL into five strata of equal width, starting in the outermost part of the IPL. There is a striking relationship between soma position and stratification level in that the cell bodies that are closer to the OPL have axons that descend and stratify more deeply in the IPL. Type 1 CB (n = 6, Fig. 1B) is an outer cone bipolar cell with a flat stratification in stratum 1 of the IPL and only one ascending primary dendrite that ramifies sparsely. Type 2 (n = 12, Fig. 1C) and type 3 (n = 15, Fig. 1D) CBs have similar dendritic tree shapes but show diffuse stratification at different levels of the outer half of the IPL, that is, in strata 1-2 and 2, respectively. Whereas the axon terminal system of type 2 looks a bit disordered, type 3 shows a well-arranged treelike branching pattern. Type 4 (n = 4, Fig. 1E) is a diffuse CB stratifying in both strata 1 and 2. We could inject only a few cells of this type; therefore, we believe it to be very rare. Type 5 (n = 10, Fig. lf, left) and type 6 (n = 6, Fig. lf, right) both have a very narrow stratification. The branching pattern of their axon terminal systems is very similar; their axons descend into the middle of the IPL where they terminate in flat side branches in two distinct thin bands in stratum 3 (type 5) and at the border of strata 3 and 4 (type 6). By using Nomarski optics, the two cell types can be distinguished by their stratification level within the IPL relative to a band of higher optical density (Fig. la, arrow), which is localized at the border of strata 3 and 4 and represents the inner cholinergic band. In some of the injected cells, the lateral extent of the axon terminal system was found to be larger than shown in the drawing. Type 7 (n = 16, Fig. 1A) and type 8 (n = 10, Fig. 1A) are diffuse CBs with their axon terminal systems in the inner part of the IPL. Type 7 stratifies in strata 3 and 4, whereas type 8 stratifies in stratum 5 near the border of the GCL. Fig. 1. Photomicrographs show bipolar cells injected with Lucifer Yellow (LY) and neurobiotin (NB). Ten types of bipolar cells as diagrammed in Figure 2 are shown here following 3,3-diaminobenzidine tetrahydrochloride (DAB) processing and viewed with Nomarski optics. Throughout the figure, the large arrows point to cell bodies and arrowheads point to axons. A Layers of the retina, outer nuclear layer (ONL), etc., are distinguishable in this retinal slice that displays three injected bipolar cells. The bipolar cell on the left is a type 7 cell and stratifies in the inner inner plexiform layer (IPL) well above the ganglion cell layer (GCL). The bipolar cell in the middle is a type 5 cell, displaying a flat stratification directly above the optical dense band (small arrow), which represents the inner cholinergic band. The bipolar cell on the right is type 8 and shows a diffuse stratification in the inner IPL near the border of the GCL. B: This bipolar cell has a flat stratification at the border of the inner nuclear layer (INL) and IPL and is thus classified as a type 1 cell. A type 2 cell with a diffuse stratification in the outer IPL is illustrated in C. An example of a type 3 cell is shown in D. Like type 2 cells, these cells have diffuse stratification but at a lower level of the outer IPL. E: Two focal levels of the same section are shown. On the left, a type 4 cell is illustrated, which stratifies diffusely in the outer half of the IPL. On the right, a rod bipolar cell is visible. Two bipolar cells, each with flat stratification at different levels of the IPL, are shown in F. On the left, the axon terminals of the outer stratifying cell (type 5) are in focus; on the right, the axon terminals of the inner stratifying cell (type 6) are in focus (small arrows). G A type 9 bipolar cell, with an axon terminal system ramifying sparsely near the boarder of the GCL, is illustrated. Scale bars = 20 Fm.

5

6 466 T. EULER AND H. WASSLE + GCL RB Fig. 2. Classification scheme for bipolar cells in the rat retina. Camera lucida drawings from injected cells show the bipolar cell types present in the rat retina. RB is the rod bipolar cell. The nine putative cone bipolar cell types (labeled 1-9) are arranged according to the stratification level of their axon terminal systems within the IPL. The IPL is subdivided into five strata of equal width (stratum 1, outer IPL; stratum 5, inner IPL). The proportions of the cells and the layers are normalized to the same scale. OPL, outer plexiform layer. Scale bar = 50 bm. d Type 8 has lobular terminals. Both cell types have relatively slender cell bodies. Type 9 CB (n = 4, Fig. 1G) has a very sparse but large axon terminal system in the inner IPL (mainly stratum 5), with occasional processes penetrating into the GCL. The dendritic tree is also sparsely branched but covers a wide range of the OPL. Like type 4, type 9 CBs seem to be uncommon. Selective staining of bipolar cells by different antibodies The spatial density of distinct bipolar types cannot be inferred from the injection frequency. For an analysis of the bipolar cell types in terms of spatial cell density and distribution, immunocytochemical markers, which label distinct populations of bipolar cells, were utilized. The antibodies used in this study stained one or two types of bipolar cells, along with other retinal neurons. The classification scheme described in Figure 2 was used to provide a comparison between the types of cells filled by injection in tissue slices and the immunolabeled cells in vertical cryostat sections. On the basis of this comparison, 5 of the 10 bipolar cell types could be identified by different immunocytochemical markers and counted in horizontal sections. Spatial density maps were calculated from these data. Bipolar cells immunoreactive for recoverin In the primate retina, flat midget bipolar cells have been found to be immunoreactive for recoverin and their spatial mosaic has been analyzed (Milam et al., 1993, Grunert et al., 1994; Wassle et al., 1994). The labeling of the rat retina with this antibody is shown in Figures 3A and 4. In Figure 3A, the ONL and the OPL are heavily stained, whereas only some bipolar cells are labeled in the INL. The axons of these stained bipolar cells run into the IPL where their axon terminal systems form two diffuse bands: a denser band in the outer IPL and a sparser band in the inner IPL close to the GCL (Milam et al., 1993; Sassoe-Pognetto et al., 1994). The cell bodies of the recoverin-immunoreactive bipolar cells differ in size and staining intensity. The open arrowheads in Figure 3A point to bipolar cells with large, intensively stained somata. The axon terminal systems of these cells form the denser band in the outer IPL within sublamina a; thus, these cells are most likely to be OFF cone bipolar cells. The filled arrowheads point to weakly stained, more slender cell bodies with axon terminal systems that form the inner IPL band within sublamina b. These cells are most likely to be ON cone bipolar cells. Comparing the morphologies of the injected bipolar cell types (Fig. 2) with the recoverin immunostaining, the labeled cells in the INL could be identified as types 2 and 8 (Fig. 3B). A small gap (small arrow in Fig. 3A) in the outer band formed by the axon terminals raises the possibility that two types have been labeled. However, in camera lucida drawings made from immunostained sections (not shown), it was found that every labeled bipolar cell with an axon ending in the outer IPL sends processes into both substrata. Therefore these neurons belong to one type of OFF bipolar cell, whose axon subdivides into two narrow strata. Figure 4 shows horizontal sections immunostained with the recoverin antibody. A section through the outer INL (Fig. 4A) reveals two regular patterns of somata which differ in size and staining intensity: the larger, strongly stained cells (arrowheads) are the putative OFF CBs that send their axons to the outer IPL (Fig. 4B), and the smaller, weakly stained cells (arrows) are the putative ON CBs with axons stratifying in the inner IPL (Fig. 4C). We counted the immunolabeled cell bodies at different retinal locations to calculate the cell densities for both populations. The total density of the recoverin-positive bipolar cells ranged from 3,600 to 7,200 cells/mm2. The proportion of the weakly stained ON CBs was approximately 44% (2,900/mm2) in the high density area and about 30% (1,000/mm2) in the peripheral retina. The density distribution of recoverin-immunoreactive CBs is shown in Figure 5. A large area of high density forms a slight diagonal, streaklike band along the nasotemporal axis, with a shallow maximum in the nasal periphery. The density decreases gradually toward the periphery. Bipolar cells immunoreactive for the glutamate transporter (GLT-1) Immunostaining of bipolar cells with antibodies against the glutamate transporter (GLT-1) has been described elsewhere (Rauen and Kanner, 1994; Grunert et al., 1994; Brandstatter et al., 1995); however, an analysis of the cell types involved has not been done in rat. The labeling of the rat retina with this antibody is shown in Figure 6A. Cones and bipolar cells are strongly immunoreactive, whereas

7 BIPOLAR CELLS IN THE RAT RETINA 467 B ONL I -&- - IPL GCL Fig. 3. Recoverin-immunoreactive cone bipolar cells in a vertical section of the rat retina. A A vertical cryostat section immunostained by the recoverin antibody is shown. The open arrowheads point to the larger, more intensely labeled somata of putative OFF-type cone bipolar cells. The solid arrowheads point to the somata of more weakly labeled, slender cell bodies of putative ON-type cone bipolar cells. The labeled axons run into the IPL where the axon terminals form two diffuse bands. The small arrow points to the gap that subdivides the outer immunoreactive band. B: The camera lucida drawing shows the two corresponding injected bipolar cell types 2 and 8 from Figure 2. Scale bar = 50 pm. both the OPL and IPL show a diffuse labeling similar to that shown by Rauen and Kanner (1994) in their study. In accord with the putative membrane localization of the GLT-1 antigen, the DAB reaction product is confined to the cell membranes. The bipolar cell axons can be followed to the middle of the IPL, where they terminate in a narrow band in either stratum 3 or at the border of strata 3 and 4. While at the microscope, we sometimes could resolve single axon terminal systems that stratify either in the outer or the inner narrow immunolabeled band. This suggests that two different bipolar cell types might be stained. In addition, some less intensely labeled cell bodies can be found. The large arrow in Figure 6A points to the outermost cell layer of the INL, where these more weakly stained somata are located. In the IPL, a few weakly labeled axons can be followed through both GLT-1 bands. The axon terminals cannot be resolved because of the relatively strong background staining. The soma position and the axon terminal shape of these weakly labeled cells suggest that they are rod bipolar cells (Rauen and Kanner, 1994). To test this hypothesis we performed a double-labeling experiment for GLT-1 and PKC, which selectively stains rod bipolar cells in the rat retina (Greferath et al., 1990). We compared the staining patterns in vertical sections (not shown) and found that 13% (n = 426) of all stained cells in the INL showed immunoreactivity for both antibodies. Almost all of the more weakly stained GLT-1-immunoreactive cell bodies in the outermost INL were also immunoreactive for PKC. This suggests that the weakly GLT-1-immunoreactive bipolar cells are in fact rod bipolar cells. Although only about 40% of the RBs were labeled for GLT-1, we presume that the GLT-1 antibody generally cross reacts with RBs. The reason for the incomplete labeling might be a low level of GLT-1 expression by RBs. Nevertheless, we cannot exclude the possibility that there is a distinct subpopulation of rod bipolar cells that is GLT-1 positive. The ratio of PKClabeled RBs to GLT-1 CBs is estimated to be 10:9, based on the double-labeling experiments. The GLT-1 staining pattern (Fig. 6A) suggests the existence of two similar types of cone bipolar cells with flat stratification and therefore confirms our proposed classification scheme with the differentiation of type 5 and type 6 CBs (Fig. 6B). Horizontal sections immunolabeled by the GLT-1 antibody are shown in Figure 7. Figure 7A shows the cell mosaic in a section through the middle of the INL. The arrowheads point to GLT-1-immunoreactive cells that appear to be encircled by the DAB reaction product. In Figure 7B, the axons of the GLT-1-immunoreactive cells leaving the INL are visible. The axon profiles (arrowheads in Fig. 7B) are relatively constant in diameter and staining intensity. This suggests that only the strongly stained GLT-1 CB axons and not the weakly stained RB axons are visible. A section through the middle of the IPL (not shown), where the GLT-1 CB axons stratify, revealed granular labeling, but individual axon terminal systems could not be resolved. For the cell density measurements, we counted GLT-1- immunoreactive axon profiles in horizontal sections at the level of the outer IPL and calculated a GLT-1 CB density ranging from 8,600 to 16,400 cells/mm2. Figure 8 shows the corresponding density map. The cell distribution is quite uniform in the central and mid-peripheral regions, with a slight increase of cell density toward the central retina. A large density peak is located in the superior mid-peripheral

8 468 T. EULER AND H. WASSLE Fig. 4. Recoverin-immunoreactive cone bipolar cells in a horizontal section of the rat retina. A A horizontal section with the focal plane in the outer INL is shown. The arrowheads point to the more intensely stained somata of the putative OFF-type cone bipolar cells. The arrows point to the more weakly stained cell bodies of the putative ON-type cone bipolar cells. B: The same section with the focal plane in the outer IPL is shown. The outer recoverin-immunoreactive (putative OFF) dendritic plexus is visible. C: In the same section, with the focal plane in the inner IPL, the inner and sparser recoverin-immunoreactive (putative ON) dendritic plexus is visible. Scale bar = 50 pm. retina, about 1 mm from the optic disk, and two smaller local density maxima were found in the inferior retina. Immunoreactivity of rod bipolar cells Rod bipolar cells can be selectively labeled with antibodies against the CY isozyme of PKC in many mammalian retinae. The PKC staining pattern has been described in detail (Greferath et at., 1990), but an analysis of the RB density and distribution in the rat retina is still missing. Because PKC immunoreactivity was also found in some amacrine cells (Negishi et al., 19881, it was desirable to find a more selective marker of RBs to count them. An antibody, raised against a peptide corresponding to the predicted amino acid sequence of the Purkinje cell differentiation marker L7 (Oberdick et al., 1988), provides such a marker for RBs in a variety of species (Berrebi et al., 1991; Griinert and Martin, 1991). Double-labeling experiments with PKC and L7 antibodies in vertical sections revealed that in rat retina 95% of all labeled bipolar cells showed immunoreactivity for both antisera, whereas the remaining 5% were only stained by the L7 antibody (U. Griinert, personal communication). Therefore PKC, as well as L7, can be used as a marker for rod bipolar cells. However, the bipolar cells exclusively labeled by the L7 antibody might represent a subpopulation of a rare but distinct bipolar cell type. Vertical sections of the rat retina stained for L7 (Fig. 9A) and PKC (Fig. 9B) are shown side by side for comparison. Both antibodies label bipolar cell somata that are located in the outermost part of the INL, although occasionally bipolar cells in the middle of the INL are labeled. Whereas L7 immunoreactivity fills the cell body, the PKC staining is confined to the cell membrane. In both sections, the RB axons can be followed to the inner border of the IPL, where they form varicose terminals. Fewer axon terminal systems seem to be labeled in the L7-immunostained section than in the PKC-stained section. It has been reported that L7 immunoreactivity is reduced in the axons of RBs (Griinert and Martin, 1991; Martin and Griinert, 1992). A PKCimmunoreactive amacrine cell (arrow) is shown in Figure 9B. Figure 9C shows a camera lucida drawing of a LYinjected rod bipolar cell.

9 BIPOLAR CELLS IN THE RAT RETINA 469 UU INF. Fig. 5. Spatial density map for recoverin-immunoreactive cone bipolar cells in the rat retina. Higher cell densities are represented by darker gray levels. The cell densities are given in cellsimmz and the cross marks the optic disk. A large area of highest cell density forms a slightly diagonal, streaklike band along the nasotemporal axis, with a shallow maximum in the nasal periphery. Scale bar = 1 mm. A horizontal section labeled with the L7 antibody reveals the dense packing of rod bipolar cell somata in the outer INL (Fig. 10A). The arrows in Figure 10A point to rare but more strongly stained cell bodies that seem to have a regular distribution across the retina and might correspond to the bipolar cells exclusively stained by the L7 antibody. Figure 10B-C shows two different horizontal sections at the level of the inner INL. Figure 10B shows a PKCimmunostained section and Figure 1OC shows an L7- labeled section. In both micrographs, the cell bodies are interspersed with strongly immunolabeled RB axons. Amacrine cell somata (Fig. 10B, arrowheads), which seem to be arranged in clusters, show PKC immunoreactivity. For the cell density measurements, we counted labeled axon profiles in horizontal sections through the inner INL or the outer IPL at different retinal eccentricities. For both PKC- and L7-immunostained sections, the calculated rod bipolar cell density was in the same range, from 13,000 to 29,000 cells/mm2 (mean = 20,000) for PKC and from 12,100 to 27,900 cells/mm2 (mean = 20,300) for L7. The 5% exclusively L7-immunoreactive bipolar cells have a density ranging from 600 to 1,000 cells/mm2, although it is not clear whether these cells are a distinct population of bipolar cells. Figure 11 shows a cell density map of a retina immunostained by the L7 antibody. The cell density is relatively high in the central and mid-peripheral retina, and only in the far periphery is a significant decrease in cell density found. In the inferior temporal retina, a large region of high cell density can be observed, and a smaller peak of cell density is located in the upper mid-peripheral retina, approximately 1 mm from the optic disc. The general features of the RB distribution were similar in both L7- and PKC-immunostained retinae. DISCUSSION Bipolar cell types in the rat retina In this study, we presented morphological evidence for nine putative cone bipolar cell types in the rat retina. The density and distribution of four of these cone bipolar cell

10 470 T. EULER AND H. WASSLE B ONL GCL Fig. 6. GLT-1-immunoreactive cone bipolar cells in a vertical section of the rat retina. A A vertical cryostat section immunolabeled with the GLT-1 antibody is shown. Cone bipolar cell somata in the middle of the INL are labeled. The arrowheads point to immunostained axons that run into the IPL where the axon terminal systems form two dense bands (small arrows). The large arrow points to the outermost cell layer of the INL, where the more weakly stained somata of rod bipolar cells are located. B: A camera lucida drawing of the two corresponding injected bipolar cell types 5 and 6 (Fig. 2) is shown. Scale bar = 50 wm. populations, as well as the population of RBs, were characterized using immunocytochemical markers. By injecting bipolar cells with LY, we found nine putative cone bipolar cell types and one rod bipolar cell. They were distinguished by the position of their somata in the INL and the branching pattern and stratification level of their axon terminal systems within the IPL. Because detailed electron microscopical studies concerning the bipolar-photoreceptor synapses in the rat retina are lacking, it is unclear whether the nine putative cone bipolar cell types are exclusively cone connecting bipolar cells or whether they are mixed bipolar cells with contacts to both rods and cones. In the retinae of lower vertebrates, such mixed inputs to bipolar cells are common (Ishida et al., 1980; Saito et al., 1985). Among mammals, a mixed connecting bipolar cell type has been suggested for squirrels (West, 1978). Muller et al. (1993), using electron microscopy, presented preliminary evidence for the existence of at least one mixed rod-cone bipolar cell type in the rat retina. They found a bipolar cell type ramifying in strata 1-2 that received wide-cleft basal contacts from cone pedicles and narrow-cleft nonribbon contacts from rod spherules, which led them to propose the existence of additional pathways for rod information in the rat retina. Muller et al. (1993) also reported, in addition to the rod bipolar cell, seven other bipolar cell types. As of yet, they have not published a full-length report; thus, a comparison to the bipolar types found in our study is not possible. In the cat retina, the border between the OFF and the ON sublamina divides the IPL into the outer one-third (sublamina a) and the inner two-thirds (sublamina b; Nelson et al., 1978; Peichl and Wassle, 1981; Cohen and Sterling, 1990a,b). Although electrophysiological evidence for functional sublaminae within the IPL of the rat retina is still wanting, the IPL stratification levels of ganglion cells suggest that the subdivision of the IPL into an OFF and ON sublamina might exist and, similar to the cat retina, that the border might be shifted toward the outer retina (Peichl, 1989). The reason for this shift may be the broad band of rod bipolar cell terminals, which occupies a large portion of the inner part of the INL (Voigt, 1986; Greferath et al., 1990). The IPL stratification levels of the nine cone bipolar cell types suggest that types 1-4, with stratification in the outer IPL, could be physiological OFF-type bipolar cells, whereas types 6-9, with stratification in the inner IPL, could be ON-type bipolar cells. For the type 5 bipolar cell, which stratifies in the middle of the IPL, it cannot be clearly predicted whether the cell is of the physiological ON or OFF type. A unifying concept in anatomical studies of the retina suggests that cell types with similar morphology form functional ON/OFF pairs (Famiglietti and Kolb, 1976; Peichl, 1989). In the rat retina, Peichl (1989) showed that different types of ganglion cells form ON/OFF pairs in such a way that outer parts of the OFF sublamina correspond with outer parts of the ON sublamina. It is very likely that at least some types of cone bipolar cells follow the stratification pattern of ganglion cells (Famiglietti and Kolb, 19761, but exceptions from this scheme have also been reported (Nelson and Kolb, 1983; Saito, 1987). In the rat retina, the following ON/OFF pairs can be defined according to morphological similarities: types 1/5 or 6, types 2/7, and types

11 BIPOLAR CELLS IN THE RAT RETINA 471 Fig. 7. GLT-1-immunoreactive cone bipolar cells in horizontal sections of the rat retina. A A horizontal cryostat section with the focal plane in the middle of the INL is shown. The arrowheads point to GLT-1-immunoreactive cone bipolar cell somata. B: A horizontal section with the focal plane in the inner INL is shown. The regular mosaic of unstained cell bodies is interspersed by the GLT-1- immunolaheled axon profiles of cone bipolar cells (arrowheads). Scale bar = 50 pm For the type 4 CBs and either type 5 or type 6 CBs, no morphological counterparts could be assigned. Immunocytochemical labeling of bipolar cells in the rat retina The morphologies of single LY-/NB-injected bipolar cells were compared with immunolabeled bipolar cells. We were able to correlate types 2, 5, 6, and 8 of Figure 2 to different immunolabeled bipolar cell populations. Types 2 and 8 probably correspond to the two bipolar cell populations labeled by the recoverin antibody. The cell bodies of type 2 bipolar cells are large and intensely stained by recoverin. The position of their axon terminals, which form a diffuse band within the outer part of the IPL, suggests that these cells may be OFF bipolar cells. Type 8 bipolar cells have more weakly stained, slender somata and their axon terminals form the inner recoverin-immunoreactive IPL band. Thus, these cells are probably ON bipolar cells. The morphologies of types 5 and 6 suggest that these cells correspond with the bipolar cell populations labeled by the GLT-1 antibody. Their axon terminal systems stratify in strata 3 and 415, respectively, forming two thin, but dense, bands in close proximity to the dendritic plexus of displaced cholinergic amacrine cells (Voigt, 1986; Brandstatter et al., 1995). Because cholinergic amacrine cells are involved in direction selectivity (Masland and Ames, 1976; Ariel and Daw, 19821, Brandstatter et al. (1995) suggested that GLT-1-positive bipolar cells contribute to the circuitry underlying direction selectivity. With the antibodies used in the present study, bipolar cell populations corresponding to LY-injected cell types 1, 3, 4, 7, and 9 were not immunostained. Type 1 bipolar cells show a flat stratification in stratum 1 of the IPL close to the border to the INL. Although we found no recoverinimmunoreactive bipolar cells stratifying exclusively in the outer part of the outer IPL band, it cannot be excluded that type 1 cells might contribute to this band. Relatively few of the filled cells were of types 4 and 9, thus we propose that these cells each comprise a sparse population of bipolar cells. Type 4 shows a diffuse stratification in both strata 1 and 2. It was shown in other mammalian retinae (Boycott and Wassle, 1991; Young and Vaney, 1991; Mills and Massey, 1992) that many types of bipolar cells show territorial behavior. Assuming that this is the case for the type 4 CB suggests that the narrow axon terminal system has a higher cell density than predicted from the filling frequency. The dendritic tree of type 9 is relatively broad, and the sparse axon terminal system ramifies in

12 472 T. EULER AND H. WASSLE > < r AP. Fig. 8. Spatial density map for GLT-1-immunoreactive cone bipolar cells in the rat retina. Higher cell densities are represented by darker gray levels. The cell densities are given in cells/mm2 and the cross marks the optic disk. The largest cell density peak is located in the superior mid-peripheral retina approximately 1 mm from the optic disk. Scale bar = 1 mm. stratum 5. In other mammals, a bipolar cell type with a comparable morphology has been described: the blue cone bipolar cell (Famiglietti, 1981; Mariani, 1984; Kouyama and Marshak, 1992). The rat retina has very few blue cones, comprising only 0.05% of all photoreceptors (Szel and Rohlich, 1992); thus, the type of bipolar cells connecting to blue cones might also be very rare. When considering the morphological similarities and the presumed low density, type 9 cells are reasonable candidates to be the blue cone bipolar cell of the rat retina. Only one type of interplexiform cell has been described in the rat retina (Perry and Walker, 1980). Because the somata of these interplexiform cells are positioned in the inner INL close to the border to the IPL, it is unlikely that these cells have been mistakenly identified as bipolar cells. Bipolar cell densities Rod bipolar cells were labeled with antibodies against PKC and L7. The density for both immunostained populations agreed fairly well and ranged from about 12,000 to 28,000 cells/mm2 (Table 1). Double-labeling experiments with the two antibodies suggested the possibility that an additional bipolar cell subpopulation was stained exclusively by the L7 antibody. If this is the case, then the cell density for this presumed subpopulation would range from 600 to 1,000 cells/mm2, corresponding to 5% of all L7- immunoreactive bipolar cells. In the horizontal sections, some L7-positive bipolar cells display a more intensely stained soma and seem to be distributed in a regular pattern in horizontal sections. These cells could be the 5% PKC-negative bipolar cells found in the double-labeling experiments. In vertical sections, however, these cells were not detected; neither more strongly stained somata nor different branching pattern of axons could be observed. The GLT-1-immunoreactive CBs with a density ranging from 8,600 to 16,400 cells/mm2 comprise the largest CB population in the rat retina found in this study (Table 1). The density of the recoverin-immunoreactive CBs ranged from 3,600 to 7,200 cells/mm2 (Table 1). Putative ON-type and OFF-type recoverin CBs could be distinguished by the staining intensity of their cell bodies and the stratification level of their axons. Putative ON-type CBs made up approxi-

13 BIPOLAR CELLS IN THE RAT RETINA 4 73 C ONL GCL Fig. 9. Rod bipolar cells in vertical sections of the rat retina. A,B: Two different vertical cryostat sections with immunolabeled rod bipolar cells (RBs) are shown. Their axons run through the IPL and terminate close to the border of the GCL. A RBs immunostained by the L7 antibody are visible. B: RBs immunolabeled with the protein kinase C (PKC) antibody are shown. The arrowhead points to a PKC-immunoreactive amacrine cell. C: A camera lucida drawing of the corresponding injected bipolar cell type RB (Fig. 2) is shown. Scale bars = 50 km. mately 37% of all recoverin-immunoreactive bipolar cells, equivalent to 1,000 to 2,900 cells/mm2. In Table 1. the densities of RBs and CBs are comdared. Despite the fact that only a fraction of the morpholo$cally identified CBs could be counted from our immunocytochemical experiments, they comprise 48% of all immunocytochemically labeled bipolar cells. The density of the bipolar cell populations immunoreactive for recoverin, GLT- 1, and L7/PKC is relatively high, and the space within the outer INL for the other cone bipolar cell types is limited. This suggests that the remaining cone bipolar cell populations may be much smaller and the total proportion of cone bipolar cells would not be larger than 60% of all bipolar cells. Additional, but more indirect, evidence for the existence of a large proportion of cone bipolar cells in the rat retina comes from the immunocytochemical staining of dissociated retinal cells. It was shown, in a preparation of cells freshly dissociated from the rat retina, that all cells with typical bipolar cell morphology were labeled by the PKC antibody (Greferath et al., 1990; Karschin and Wassle, 1990). Because only RBs are immunoreactive for PKC, this suggests that only rod bipolar cells maintain their typical morphology during the dissociation process, whereas cone bipolar cells do not. Greferath et al. (1990) provided evidence for a loss of dendrites and axons of CBs during dissociation. They applied the ROB antibody (MAb 115A10), which was raised against a homogenate of rabbit olfactory bulb (Onoda and Fujita, 1987), to retinal sections and dissociated retinae. In the retinal sections, both rod bipolar cells and ON cone bipolar cells were labeled. In the dissociates, rod bipolar cells were labeled, as were many round cell bodies, which obviously had lost their processes. These round cell bodies may well represent the populations of cone bipolar cells labeled by ROB in the retinal sections (Greferath et al., 1990; Griinert et al., 1994). Spatial distribution of bipolar cell types Here we compare the spatial distributions of rod bipolar cells, GLT- 1-immunoreactive cone bipolar cells, and recoverin-positive cone bipolar cells with the distribution of other cell types in the rat retina. In the nasal and inferior portions of the retina, the density of middle-to-long wavelength sensitive cones (M cones) is slightly increased. The other photoreceptor types, rods and short wavelength-sensitive cones (S cones), show a relatively even distribution over the whole rat retina (Coers, 1983; Szel and Rohlich, 1992). The distribution of recoverin-immunoreactive CBs follows, to some extent, that of the M cones. Horizontal cells are distributed more uniformly across the rat retina (Peichl and Gonzdez- Soriano, 1994), so a comparison with bipolar cell distributions is not useful. A peak of ganglion cell density is located in the upper retina at a distance of about 20" ( - 1 mm in adult rats) from the optic disk, slightly shifted to the temporal retina. In this so-called central area, the ganglion cell density reaches 4,000 cells/mm2 (Fukuda, 1977; Jeffery, 1985). The peak densities of both rod bipolar cells and GLT-1-positive cone bipolar cells are probably colocalized with the central area. Wassle et al. (1993) found that the density of AII amacrine cells also increases in the upper retina. The increase in density of RBs, A11 cells, and ganglion cells within the central area suggests that spatial resolution is improved as a result of a lower convergence of rod signals onto ganglion cells. It should be emphasized, however, that the central area in the rat is not as pronounced as the central area of

14 4 74 T. EULER AND H. WASSLE Fig. 10. Rod bipolar cells in horizontal sections of the rat retina. Horizontal cryostat sections immunostained with either the L7 or the PKC antibody are shown. A: In the horizontal section, with the focal plane in the outer INL, the L7-immunoreactive rod bipolar cell (RBI somata are visible. The arrows point to more intensely stained cell bodies that seem to have a regular distribution across the retina and might correspond to a small subpopulation of cone bipolar cells. B,C: Two different horizontal sections with the focal plane at the level of the border of IPL and GCL are shown. Immunolabeled axon profiles of RBs are visible. B: The section was immunostained with the PKC antibody. PKC-labeled amacrine cell bodies (arrowheads), interspersed with axon profiles of RBs, are visible. C: In this section, the axon profiles of RBs immunostained with the L7 antibody are visible. Scale bar = 50 pm. the cat, the visual streak of the rabbit, or the fovea of the primate (Hughes, 1977). We found a second, larger peak of RB density located in the lower temporal and peripheral retina. Jeffery (1985) found, at the same retinal location, a crescent-shaped area where approximately 10% of the a ganglion cells project to the ipsilateral visual cortex. He concluded that in this area binocular vision might be possible. It is possible that the increased density of rod bipolar cells in the lower temporal retina coincides with this area of binocular vision. It has been shown for the rabbit retina (Young and Vaney, 1991; Mills and Massey, 1992) and for the macaque monkey retina (Boycott and Wassle, 1991) that the axon terminals of many types of bipolar cells show territorial behavior and therefore their axonal field coverage approaches 1. By assuming that the axon terminals of the injected bipolar cells were completely labeled, the cell density can roughly be estimated from the diameter of their axon terminal systems. For both the recoverin-positive type 2 and type 8 cone bipolar cells, this estimation matches the results from the cell density measurements, indicating an axonal field coverage of approximately 1. For the GLT-1- immunoreactive cone bipolar cells, the ratio of type 5 to type 6 cells could not be evaluated, but the diameter of their axon terminal systems suggests that their axonal fields might overlap. Because specific immunocytochemical markers are not available for the other types of cone bipolar cells, an estimation of their cell density is of limited use. Further detailed analysis of single axon terminal systems in horizontal sections is needed to answer the question of axonal coverage in the rat retina. Cone bipolar cells in mammalian retina In the mammalian retina, many different types of cone bipolar cells have been recognized by using various methods. In the rabbit retina, 9 types have been described from Golgi-stained material (Famiglietti, 1981). Other studies have confirmed the existence of two of these types and recognized one additional type (Mills and Massey, 1992). In the cat retina, 8 to 11 types of cone bipolar cells have been

15 BIPOLAR CELLS IN THE RAT RETINA 475 h SUP < das. INF. - Fig. 11. Spatial density map for rod bipolar cells of the rat retina. The spatial density of rod bipolar cells (RBs; L7 immunolabeled) was counted from their descending axons. In the inferior temporal periphery, the RB density is increased in a large area. A smaller peak of cell density is located in the upper mid-peripheral retina about 1 mm from the optic disc. The general features of the RB distribution were similar in both L7- and PKC-immunostained retinae. Scale bar = 1 mm. TABLE 1. Ranges of Cell Density (cells/mm2) Across the Retina (Maxima and Minima) for Different Cell Types Directly Involved in Rod and Cone Pathway of the Rat Retina Photoreceptors Rod Cones Bipolar cells Rod bipolar cells Cone bipolar cells icbs) GLT-1-positive CBs Recoverin-positive CBs z Other CB types A11 amacrine cells Ganglion cells 'From Coers (1983). 'From Sz6l and Rohlich (1992) 3From Wassle et al. (1993). 'From Jeffery (1985). 206, ,000' 3,800 5, ,000 28,000 8,600 16,400 3,000 7,200 > 11,600 23,600 '? 2,000 7,0003 2,500 4,0004 described (Famiglietti, 1981; Kolb et al., 1981; McGuire et al., 1984; Cohen and Sterling, 1990a,b). In primates, mainly in the macaque monkey retina, 10 different types of cone bipolar cells were distinguished in Golgi-prepared tissue (Boycott and Wassle, 1991). This classification was confirmed by using immunocytochemical methods (Griinert et al., 1994; Wassle et al., 1994). In the rat retina, we could recognize 9 different types of cone bipolar cells. Taken together, these studies suggest that there are about 10 types of cone bipolar cells in the mammalian retina. Although cat, rabbit, monkey, and rat retinae have some notable differences in retinal circuitry, similarities exist between many of the cell types from all classes of retinal cells. However, a comparison between cell types in different mammals is difficult, especially when it is based only on morphology and immunoreactivity. With some caution applied, both the morphology and immunoreactivity of a certain cell type can be useful to find counterparts in other mammalian retinae. This has been demonstrated for the rod bipolar cell, whose morphology and immunoreactivity seem to be well preserved in different mammals (Negishi et al., 1988; Greferath et al., 1990; Griinert and Martin, 1991; Muller and Peichl, 1991; Young and Vaney, 1991; Zhang andyeh, 1991).

16 476 T. EULER AND H. WASSLE Bipolar cell types can be classified by the branching pattern and the stratification level of their axon terminal systems within the IPL. In the rat retina, we recognized cell types with a flat stratification in the IPL (types 1,5,6) and cell types whose axon terminals ramify more broadly (types 3, 8) or even in more than one stratum of the IPL (types 2, 4, 7,9). A similar rough classification can be made with the cone bipolar cells that have been described for the macaque monkey (Boycott and Wassle, 1991), the cat, and the rabbit retina (Famiglietti, 1981). Similarities of bipolar cells across species can be noted in vertical sections. For example, the type 1 CBs in the rat retina resemble the DBl cell in the macaque monkey retina (Boycott and Wassle, 1991) and the CBal and the DAPI Bal cells in the rabbit retina (Cohen and Sterling, 1990a; Mills and Massey, 1992). There are also resemblances between the type 3 CBs (rat), DB2 (macaque monkey), DAPI Ba2/ CBa2 (rabbit), and the varicose flat bipolar cell in the cat retina (Famiglietti, 1981). Except for differences in the stratification level within the IPL, type 5 and type 6 CBs resemble DB3 and DB5 in the macaque monkey retina and the various flat stratifying bipolar cells in the cat and the rabbit retina (Famiglietti, 1981). Our study thus supports the view that bipolar cell architecture in the mammalian retina might be highly conserved (Famiglietti, 1981; Wade and Boycott, 1991). Using immunocytochemical results for a comparison between cone bipolar cell types in different mammals is rather questionable. Immunocytochemical similarities between cell types do not necessarily imply functional similarities, given the diversity of protein isoforms available for antibody binding. In addition, a cursory examination of immunocytochemical similarities between bipolar cells in this study and bipolar cells characterized in other mammals suggest that these comparisons are inappropriate in our case. For example, in the rat retina, two diffuse cone bipolar cells, an inner (type 2) and an outer (type 8) stratifying type, are recoverin immunoreactive, whereas, in the macaque monkey retina, only an outer stratifying midget bipolar cell (FMB in Wassle et al., 1994) is labeled by this antibody. The same midget bipolar cell is also immunoreactive to GLT-1, whereas the two CB types of the rat retina, which stratify narrowly in the middle of the IPL, are GLT-1 positive. Thus, it is difficult to make morphological and immunocytochemical comparisons between the cone bipolar cells described in mammalian retina. A more reliable, and appropriate, basis for such comparisons will come from a combination of anatomical and functional studies. At this time, very little is known about functional properties of different cone bipolar cells. Anatomy offers some clues: the axon stratification level within the IPL possibly indicates whether or not the cell will show a physiological ON or OFF type of response, and the cell density may give an indication of the highest spatial frequency that the bipolar cell could transmit. Physiological recordings from ganglion cells indicate that the light signal is separated into different channels, which differ in their temporal, spatial, and chromatic transfer characteristics (Levick, 1975; Lee et al., 1989). It is likely that the bipolar cells that provide input to these ganglion cells play a role in determining these channels. Supporting this view are in situ hybridization studies and immunocytochemistry that have revealed differential expression patterns of glutamate receptor subtypes (Hughes et al., 1992; Hamassaki-Britto et al., 1993; Brandstatter et al., 1994) and GABAA and glycine receptor subunits (Brecha, 1992; Greferath et al., 1993, 1994) in the INL and IPL. These findings may reflect functional differences in bipolar cell populations that contribute to the separation of signal properties. Thus, different types of bipolar cells could represent the first element of parallel input channels with distinct functional properties. Photopic vision of the rat Rats are nocturnal animals and, accordingly, about 99% of their photoreceptors are rods. Early electrophysiological experiments suggested the existence of both scotopic and photopic mechanisms in the rat retina (Green, 1971,1973). More recently, Jacobs et al. (1991) described two photopic sensitivities in the electroretinogram of rats, indicating a functional two-cone system, although behavioral tests failed to produce any evidence for the existence of color vision in rat (Neitz and Jacobs, 1986). In agreement with the electrophysiological data, two types of cones have been demonstrated in the rat retina by using cone-specific antibodies directed against visual pigments (Szel and Rohlich, 1992; Szel et al., 1993). Seven percent of all cones are presumed to be sensitive to short wavelengths, whereas the remaining cones probably contain pigments that are sensitive to middle-to-long wavelengths. In the present study, we have given strong evidence for the existence of a variety of different putative cone bipolar cell types, which probably outnumber rod bipolar cells and may contribute to the photopic system of the rat. These results are consistent with the existence of a two-cone system in the rat retina and further supports the idea of photopic vision in the rat. In addition, the diversity of cone bipolar cell types is consistent with the model proposed by Strettoi et al. (1990, 1994) that cone bipolar cells may serve as interneurons in the rod pathway. Although they receive no direct input from rods, cone bipolar cells would, in such a model, play an important role in scotopic vision. ACKNOWLEDGMENTS We thank Dr. B.J. Kanner, Dr. K.-W. Koch, and Dr. J.I. Morgan for providing antibodies. We are grateful to U. Pieper and F. Boij for excellent technical assistance and to Dr. U. Griinert for practical hints and helpful discussion. We thank Prof. B.B. Boycott and Dr. M.L. Veruki for critically reading and improving the manuscript. LITERATURE CITED Ariel, M., and N.W. Daw (1982) Pharmacological analysis of directionally sensitive rabbit retinal ganglion cells. J. Physiol. (London) 324: Barnstable, C.J. (1980) Monoclonal antibodies which recognize different cell types in the rat retina. Nature 286: Barnstable, C.J., K. Akagawa, R. Hofstein, and J.P. Horn (1983) Monoclonal antibodies that label discrete cell types in the mammalian nervous system. Cold Spring Harbor Quant. Biol. 48t Berrebi, A.S., J. Oherdick, L. Sangameswaian, S. Christakos, J.I. Morgan, and E. Mugnaini (1991) Cerebellar purkinje cell markers are expressed in retinal bipolar neurons. J. Comp. Neurol. 308: Boos, R., H. Schneider, and H. Wassle (1993) Voltage- and transmitter-gated currents of AII-amacrine cells in a slice preparation of the rat retina. J. Neurosci. 13: Boycott, B.B., and J.E. Dowling (1969) Organization of the primate retina: Light microscopy. Phil. Trans. R. SOC. Lond. B 799Ci551:

17 , I BIPOLAR CELLS IN THE RAT RETINA 477 Boycott, B.B., and H. Kolb (1973) The connections between bipolar cells and photoreceptors in the retina of the domestic cat. J. Comp. Neurol. 148: Boycott, B.B., and H. Wassle (1991) Morphological classification of bipolar cells in the macaque monkey retina. Eur. J. Neurosci. 3: Brandstatter, J.H., E. Hartveit, M. Sassoe-Pognetto, and H. Wassle (1994) Expression of NMDA and high-affinity kainate receptor subunit mrnas in the adult rat retina. Eur. J. Neurosci. 6:llOO Brandstatter, J.H., U. Greferath, T. Euler, and H. Wassle (1995) Costratification of GABA, receptors with the directionally selective circuitry of the rat retina. Vis. Neurosci Brecha, N.C. (1992) Expression of GABA, receptors in the vertebrate retina. Prog. Brain Res. 90:3-28. Cajal, S.R. y (1893) La retine des vertebres. La cellule 9: Coers, M. (1983) Vergleichende Untersuchung an Augen von pigmentierten und albinotischen Laborratten. Dissertation, Tierarztliche Hochschule Hannover. Cohen, E., and P. Sterling (1990a) Demonstration of cell types among cone bipolar neurons of cat retina. Phil. Trans. R. SOC. Lond. B 330~ Cohen, E., and P. Sterling (1990b) Convergence and divergence of cones onto bipolar cells in the central area of the cat retina. Phil. Trans. R. SOC. Lond. B 330~ Dacheux, R.F., and E. Raviola (1986) The rod pathway in the rabbit retina: A depolarizing bipolar and amacrine cell. J. Neurosci. 6~21: Danbolt, N.C., J. Storm-Mathisen, and B.I. Kanner (1992) An [Na++K+l coupled L-glutamate transporter purified from rat brain is located in glial cell processes. Neuroscience Dowling, J.E., and B.B. Boycott (1966) Organization of the primate retina: Electron microscopy. Proc. R. SOC. Lond. B 166~ Edwards, F.A., A. Konnerth, B. Sakmann, and T. Takahashi (1989) A thin slice preparation for patch clamp recordings from neurons of the mammalian central nervous system. Pflug. Arch. 414, Famiglietti, E.V. (1981) Functional architecture of cone bipolar cells in mammalian retina. Vis. Res. 21: Famiglietti, E.V., and H. Kolb (1975) A bistratified amacrine cell and synaptic circuitry in the inner plexiform layer of the retina. Brain Res Famiglietti, E.V., and H. Kolb (1976) Structural basis for ON- and OFFcenter responses in retinal ganglion cells. Science 194~ Feigenspan, A., H. Wassle, and J. Bormann (1993a) Pharmacology of GABA receptor C1- channels in rat retinal bipolar cells. Nature 361(6408): Feigenspan, A., J. Bormann, and H. Wassle (1993b) Organotypic slice culture of the mammalian retina. Vis. Neurosci. 10: Fukuda, Y. (1977) A three-group classification of rat retinal ganglion cells: Histological and physiological studies. Brain Res. 119: Green, D.G. (1971) Light adaptation in the rat retina: Evidence for two receptor mechanisms. Science 174t Green, D.G. (1973) Scotopic and photopic components of the rat electroretinogram. J. Physiol. (London) 228~ Greferath, U., U. Grunert, and H. Wade (1990) Rod bipolar cells in the mammalian retina show protein kinase C-like immunoreactivity. J. Comp. Neurol. 301t Greferath, U., F. Muller, H. Wiissle, B. Shivers, and P. Seeburg (1993) Localization of GABAA receptors in the rat retina. Vis. Neurosci. 10: Greferath, U., J.H. Brandstatter, H. Wassle, J. Kirsch, J. Kuhse, and U. Grunert (1994) Differential expression of glycine receptor subunits in the retina of the rat: A study using immunohistochemistry and in situ hybridization. Vis. Neurosci. 11: Griinert, U., and P.R. Martin (1991) Rod bipolar cells in the macaque monkey retina: Immunoreactivity and connectivity. J. Neurosci. 11 (9): Griinert, U., P.R. Martin, and H. Wassle (1994) Immunocytochemical analysis of bipolar cells in the macaque monkey retina. J. Comp. Neurol. 348t Hamassaki-Britto, D.E., I. Hermans-Borgmeyer, S. Heinemann, and T.E. Hughes (1993) Expression of glutamate receptor genes in the mammalian retina: The localization of GluRl through GluR7 mrnas. J. Neurosci. 13f51: Hsu, S.M., L. Raine, and H. Fanger (1981) Use of avidin-biotin peroxidase complex (ABC) in immunoperoxidase techniques. J. Histochem. Cytochem. 29~ Hughes, A. (1977) The topography of vision in mammals of contrasting life style: Comparative optics and retinal organization. In F. Crescitelli (ed): Handbook of Sensory Physiology. Berlin: Springer-Verlag, pp Hughes, T., I. Hermans-Borgmeyer, and S. Heinemann (1992) Differential expression of glutamate receptor genes (GluR1-5) in the rat retina. Vis. Neurosci. 8: Ishida, A.T., W.K. Stell, and D.O. Lightfoot (1980) Rod and cone inputs to bipolar cells in goldfish retina. J. Comp. Neurol. 191t Jacobs, G.H., J. Neitz, and J.F. Deegan (1991) Retinal receptors in rodents maximally sensitive to ultraviolet light. Nature 353(6345): Jeffery, G. (1985) The relationship between cell density and the nasotemporal division in the rat retina. Brain Res. 347~ Kaneko, A. (1970) Physiological and morphological identification of horizontal, bipolar and amacrine cells in goldfish retina. J. Physiol. (London) 207: Karschin, A., and H. Wassle (1990) Voltage- and transmitter-gated currents in isolated rod bipolar cells of rat retina. J. Neurophysiol. 63(41; Kolb, H., and E.V. Famiglietti (1974) Rod and cone pathways in the inner plexiform layer of cat retina. Science 186~4749. Kolb, H., and E.V. Famiglietti (1976) Rod and cone pathways in the retina of the cat. Invest. Ophthalmol. Vis. Sci. 15t Kolb, H., R. Nelson, and A. Mariani (1981) Amacrine cells, bipolar cells and ganglion cells of the cat retina: a golgi study. Vis. Res. 21~ Konnerth A,, A.L. Obaid, and B.M. Salzberg (1987) Optical recording of electrical activity from parallel fibers and other cell types in skate cerebellar slices in uitro. J. Physiol. (London) 393: Kouyama, N., and D. Marshak (1992) Bipolar cells specific for blue cones in the macaque retina. J. Neurosci. 12(4): Lambrecht, H.-G., and K.-W. Koch (1991) A 26 kd calcium binding protein from bovine rod outer segments as modulator of photoreceptor guanylate cyclase. EMBO J. 1Ot Lambrecht, H.-G., and K.-W. Koch (1992) Recoverin, a novel calciumbinding protein from vertebrate photoreceptors. Biochim. Biophys. Acta. 1160(1): Lee, B.B., P.R. Martin, and A. Valberg (1989) Sensitivity of macaque retinal ganglion cells to chromatic and luminance flicker. J. Physiol Levick, W.R. (1975) Form and function of cat retina ganglion cells. Nature 254: Mariani, A.P. (1984) Bipolar cells in monkey retina selective for the cones likely to be blue-sensitive. Nature 308: Martin. P.R.. and U. Griinert (1992) Soatial densitv and immunoreactivitv of bipolar cells in the macaque monkey retina. J. Comp. Neurol Masland, R.H., and A. Ames (1976) Responses to acetylcholine of ganglion cells in an isolated mammalian retina. J. Neurophysiol. 39: McGuire, B.A., J.K. Stevens, and P. Sterling (1984) Microcircuitry ofbipolar cells in the cat retina. J. Neurosci. 4t Milam, A.H., D.M. Dacey, and A.M. Dizhoor (1993) Recoverin immunoreactivity in mammalian cone bipolar cells. Vis. Neurosci. 1O:l-12. Mills, S.L., and S.C. Massey (1992) Morphology of bipolar cells labeled by DAPI in the rabbit retina. J. Comp. Neurol. 321: Miiller, B., and L. Peichl (1991) Rod bipolar cells in the cone-dominated retina of the tree shrew Tupaia belangeri. Vis. Neurosci. 6t Muller, J.F., P.D. Lukasiewicz, and M.S. Silverman (1993) Rod and cone inputs to bipolar cells in the rat retina [abstract Invest. Ophthalmol. 34t984. Negishi, K., S. Kato, and T. Teranishi (1988) Dopamine cells and rod bipolar cells contain protein kinase C-like immunoreactivity in some vertebrate retinas. Neurosci. Lett. 94(3): Neitz, J., and G.H. Jacobs (1986) Reexamination of spectral mechanisms in the rat (Rattus noruegicus). J. Comp. Psychol. 100(1): Nelson, R., and H. Kolb (1983) Synaptic patterns and response properties of bipolar and ganglion cells in the cat retina. Vis. Res. 23(10): Nelson, R., E.V. Famiglietti, and H. Kolb (1978) Intracellular staining reveals different levels of stratification for ON- and OFF-center ganglion cells in cat retina. J. Neurophysiol. 41: Oberdick, J., F. Levinthal, and C. Levinthal(1988) A purkinje cell differentiation marker shows a partial DNA sequence homology to the cellular sislpdgf2 gene. Neuron 1: Onoda, N. (1988) A monoclonal antibody specific for a subpopulation of retinal bipolar cells in vertebrates. Neurosci. Res. 8(Suppl.):S113-S125. Onoda, N., and S.C. Fujita (1987) A monoclonal antibody specific for a subpopulation of retinal bipolar cells in the frog and other vertebrates. Brain Res

18 478 Peichl, L. (1989) Alpha and delta ganglion cells in the rat retina. J. Comp. Neurol. 286: Peichl, L., and J. Gonzalez-Soriano (1994) Morphological types of horizontal cell in rodent retinae: A comparison of rat, mouse, gerbil and guinea pig. Vis. Neurosci. 11: Peichl, L., and H. Wassle (1981) Morphological identification of ON- and OFF-centre brisk transient (Y) cells in the cat retina. Proc. R. SOC. Lond. B 212: Perry, V.H., and M. Walker (1980) Amacrine cells, displaced amacrine cells and interplexiform cells in the retina of the rat. Proc. R. Soc. Lond. B 208: Polyak, S.L. (1941) The Retina. Chicago: University of Chicago Press. Rauen, T., and B. Kanner (1994) Localization of the glutamate transporter GLT-1 in rat and macaque monkey retina. Neurosci. Lett. 169: Rohrenbeck, J., H. Wassle, and B.B. Boycott (1989) Horizontal cells in the monkey retina: Immunocytochemical staining with antibodies against calcium binding proteins. Eur. J. Neurosci. 1: Saito, T. (1987) Physiological and morphological differences between ONand OFF-center bipolar cells in the vertebrate retina. Vis. Res. 27t Saito, T., T. Kujiraoka, T. Yonaka, and Y. Chino (1985) Reexamination of photoreceptor-hipolar connectivity patterns in carp retina: HRP-EU and golgi-em studies. J. Comp. Neurol. 236t Sassoe-Pognetto, M., H. Wassle, and U. Griinert (1994) Glycinergic synapses in the rod pathway of the rat retina: Cone bipolar cells express the alpha 1 subunit of the glycine receptor. J. Neurosci. 14(81: Strettoi, E., R.F. Dacheux, and E. Raviola (1990) Synaptic connections of rod bipolar cells in the inner plexiform layer of the rabbit retina. J. Comp. Neurol. 295: , Strettoi, E., R.F. Dacheux, and E. Raviola (1994) Cone bipolar cells as interneurons in the rod pathway of the rabbit retina. J. Comp. Neurol. 347: Suzuki, S., M. Tachibana, and A. Kaneko (1990) Effects ofglycine and GABA on isolated bipolar cells of the mouse retina. J. Physiol. (London) 421 : Szel, A,, and P. Rohlich (1992) Two cone types of rat retina detected by anti-visual pigment antibodies. Exp. Eye Res. 55: T. EULER AND H. WASSLE SzB1, A., P. Rohlich, and T. Van Veen (1993) Short-wave sensitive cones in the rodent retina. Exp. Eye Res. 57: Toyoda, J. (1973) Membrane resistance changes underlying the bipolar cell responses in the carp retina. Vis. Res. 13: Vaney, D.I. (1991) Many diverse types of retinal neurons show tracer coupling when injected with biocytin or neurobiotin. Neurosci. Lett. 125: Voigt, T. (1986) Cholinergic amacrine cells in the rat retina. J. Comp. Neurol. 248: Walker, J.M., E.C. Homan, and J.J. Sando (1990) Differential activation of protein kinase C isozymes by short chain phosphatidylserines and phosphatidylcholines. J. Biol. Chem. 265: W&sle, H., and B.B. Boycott (1991) Functional architecture of the mammalian retina. Phys. Rev. 71t W&sle, H., W.R. Levick, and B.G. Cleland (1975) The distribution of the alpha type of ganglion cells in the cat s retina. J. Comp. Neurol. 159t Wassle, H., M. Yamashita, U. Greferath, U. Griinert, and F. Muller (1991) The rod bipolar cell of the mammalian retina. Vis. Neurosci. 7: Wassle, H., U. Grunert, and J. Rohrenbeck (1993) Immunocytochemical staining of AH-amacrine cells in the rat retina with antibodies against parvalbumin. J. Comp. Neurol. 332: Wassle, H., U. Griinert, P.R. Martin, and B.B. Boycott (1994) Immunocytochemical characterization and spatial distribution of midget bipolar cells in the Macaque monkey retina. Vis. Res. 34(5): West, R.W. (1978) Bipolar and horizontal cells of the gray squirrel retina: Golgi morphology and receptor connections. Vis. Res. 18: Yeh, H.H., M.B. Lee, and J.E. Cheun (1990) Properties of GABA-activated whole-cell currents in bipolar cells of the rat retina. Vis. Neurosci. 4: Young, H.M., and D.I. Vaney (1991) Rod-signal interneurons in the rabbit retina: 1. Rod bipolar cells. J. Comp. Neurol. 310: Zhang, D.R., and H.H. Yeh (1991) Protein kinase C-like immunoreactivity in rod bipolar cells of the rat retina: A developmental study. Vis. Neurosci. 6: