Exp Brain Res (1992) 89:115-124

Experimental Brain Research �9 Springer-Verlag 1992

Calcium binding protein (calbindin D28k) immunoreactivity in the hamster superior colliculus: ultrastructure and lack of co-localization with GABA M. Behan*, A. Jourdain, and G.M. Bray

Center for Research in Neuroscience, McGill University and the Montreal General Hospital, Montreal, Canada

Received September 14, 1991 / Accepted December 19, 1991

Summary. The expression of specific calcium binding proteins is being used increasingly as a potential neu- roanatomical marker for neurons with similar functions. In this study, the distribution of calbindin D28k in the superior colliculus (SC) of adult hamsters was examined by light and electron microscopy. Calbindin immuno- reactivity was prominent in specific regions and laminae of the SC throughout its rostrocaudal extent, and was found to label horizontal, vertical and stellate cell types. In addition, calbindin label highlighted "bridges" of neuronal processes in the intermediate layers. The most frequent calbindin-immunoreactive profiles seen in the electron microscope were dendrites, some of which were post-synaptic to apparent retinal ganglion cell axon ter- minals. Labelled axons and axon terminals were less frequently encountered. There was considerable overlap between the size distribution of calbindin D28k-immuno- reactive neurons and that of GABA-immunoreactive or Nissl stained neurons in the SC. However, using a double fluorescent labelling technique, and examination of the tissue with confocal laser microscopy, no neurons were observed in the hamster SC that showed immunoreactiv- ity for both catbindin and GABA. In this regard, the SC is similar to the mammalian lateral geniculate nucleus and the pretectum, but differs from the neocortex, where calbindin and GABA are colocalized. The demonstration in the SC, as well as other parts of the nervous system, of sub-populations of neurons that contain distinct cal- cium-binding proteins suggests that these neurons have different functional properties. Correlative studies may clarify the relevance of these cytoplasmic components as cell markers, as well as their different patterns of associa- tion with neurotransmitters and peptides.

Key words: Immunocytochemistry - Visual pathways Electron microscopy - Hamster

* Present address: Department of Comparative Biosciences, Uni- versity of Wisconsin, Madison, WI 53706, USA

Offprint requests to: M. Behan

Introduction

The mammalian superior colliculus (SC) is organized into alternating fibrous and cellular laminae. These lami- nae are of particular interest as they appear to delineate specific afferents and efferents to the SC from a large number of cortical and subcortical structures (for review, see Huerta and Hatting 1984). Several recent studies have attempted to segregate regions and/or populations of neurons in the SC based upon their immunoreactivity to identified or putative neurotransmitters and neuro- modulators (Graybiel et al. 1984; Bennett-Clarke et al. 1989; Miguel-Hidalgo et al. 1989; Mize 1989).

Distinct subpopulations of structurally homogeneous neurons in the visual system have also been defined on the basis of their content of calcium-binding proteins, particularly calbindin D28k. In the retina of several spe- cies, horizontal, amacrine and retinal ganglion cells con- tain calbindin D28k (Rohrenbech et al. 1987; Pasteels et al. 1990). In the cat SC, different classes of interneurons as well as a small population of projection neurons are calbindin-immunoreactive (Mize et al. 1991). Calbindin immunoreactivity has also been observed in the lateral geniculate nucleus of cats (Stichel et al. 1986; Demeule- meester et al. 1989; Jones and Hendry 1989). Moreover, in monkeys (Celio et al. 1986; Hendry et al. 1989) and cats (Stichel et al. 1987), calbindin immunoreactivity labels the non-geniculate recipient areas of primary visual cortex.

In some parts of the nervous system, neurons that contain calcium-binding proteins such as calbindin D28k or parvalbumin are also GABAergic. Co-localization of GABA and calbindin has been demonstrated or can be inferred in monkey visual cortex (Defilipe et al. 1989; Van Brederode et al. 1990), rat neocortex (Celio 1990), and rat striatum (Gerfen et al. 1985; Oertel and Mug- naini 1984). However, only 4% of calbindin-containing neurons in the cat SC are also GABAergic (Mize et al. 1991). Other populations of calbindin-immunoreactive neurons that do not contain GABA are found in the monkey LGN (Jones and Hendry 1989), cat pretectum

116

Fig. 1A, B. Photomicrographs showing the distribution of calbindin D28k immuno- reactivity in coronal sections through ros- tral (A) and caudal (B) regions of the hamster superior colliculus. A Open arrow indicates a leaflet of labelled cells in the stratum griseum intermedium (SGI). Closed arrow indicates large, lightly la- belled cells in the deep layers. Stratum zonale (SZ); Stratum griseum superficiale (SGS); stratum opticum (SO); stratum al- bum intermedium (SAI); stratum griseum profundum (SGP); stratum album profun- dum (SAP). B Closed arrow indicates cal- bindin labelled axons in the deep layers of the SC

(Nabors and Mize 1990), and pyramidal cells of the CAI and CA2 regions of the h ippocampus (Woodson et al. 1989; Celio 1990).

The hamster SC has been the subject of extensive structure-function correlations (Schneider 1969; Chalupa 1981 ; Rhoades et al. 1987). The reorganization (Schneider 1973; Mooney et al. 1985) and regeneration (Carter et al. 1989) of afferent connections to the SC after experiment- al lesions have also been investigated in hamsters. The aim of the present study was to determine the distribu- tion of calbindin immunoreactivi ty in the hamster SC, particularly as calbindin D28k localization varies con- siderably among species (Pasteels et al. 1990). Specifically we sought to determine: i) if calbindin-containing neu- rons in the SC were associated with any particular lamina that might shed light on their association with specific

afferent or efferent pathways, ii) if there was co-localiza- tion of calbindin and G A B A in SC neurons, and iii) the ultrastructure of calbindin-immunoreactive profiles in the SC neuropile.

Material and methods

Perfusion and tissue processin9

Adult golden hamsters were anesthetized deeply with intraperi- toneal 7% chloral hydrate in saline (0.42 mg/g body weight), and perfused intracardially. For calbindin immunocytochemistry, ani- mals were perfused with a saline wash followed by 300 ml of 4% paraformaldehyde, 0.1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. For GABA immunocytochemistry, or combined GABA

117

and calbindin immunocytochemistry, animals were perfused with a saline and heparin perfusion (2.5 I.U. heparin/ml) followed by 300 ml of 2% paraformaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer. After postfixation for one hour, coronal 50 ~tm thick sections of the midbraln were cut with a vibrating microtome into 0.1 M phosphate buffer.

Immunocytochemistry

Tissue sections to be processed for GABA, or for combined calbin- din D28k and GABA immunoreactivity, were placed in 1% sodium borohydride for 30 min. All sections were rinsed in 0.1 M phosphate buffered saline (PBS), preblocked for 30 min with 10% normal goat serum (Vector Labs) containing 0.3% Triton X-100 in PBS for 30 min, and incubated in primary antiserum with 2% normal goat serum and 0.3% Triton X-100 in PBS for 18 h at 4 ~ C.

For calbindin D28k immunoreactivity, we used a polyclonal antibody (Baimbridge and Miller 1982) at a dilution of 1 : 1000. For GABA immunoreactivity, we used either a polyclonal antibody (Incstar) at a dilution of 1 : 5000 or a monoclonal antibody (Sith- igorngul et al. 1989) at a dilution of 1 : 10,000. After rinsing in PBS for 30 min, the sections were processed for immunofluorescence or immuno-peroxidase visualization of the reaction product. For dou- ble-labelling immunofluorescence, sections were incubated with combinations of the primary antibodies to calbindin and GABA, rinsed in 0.1 M PBS, and simultaneously incubated for 1 hour with FITC-labelled goat anti-rabbit IgG (1:50) and RITC-labelled sheep anti-mouse IgG (1:50). For calbindin or GABA immuno- peroxidase reactions, sections were incubated with biotinylated goat anti-rabbit IgG (Vector Labs) for one hour at room temperature, rinsed for 30 min in PBS, incubated for one hour in avidin-biotiny- lated horseradish peroxidase, rinsed again in PBS, and treated with 0.05% 3,3'-diaminobenzidine tetrachloride (DAB, BDH Chemical) in PBS with 0.003% hydrogen peroxide for 5-15 min.

As immunocytochemical controls, sections were processed with- out primary antibodies, or with the calbindin D28k antibody re- placed with pooled rabbit serum. As controls for the double-label- ling studies, sections incubated with antibodies to calbindin or GABA were then incubated with the inappropriate secondary anti- bodies (sheep anti-mouse or goat anti-rabbit IgG, respectively). Readsorption controls for the calbindin antibody were reported previously to be negative (Buchan and Balmbridge 1988). Although another calcium-binding protein, calretinin D29k, shows 58% homology to calbindin D28k (Rogers 1987), the antibody used in the present study does not cross react with calretinin D29k (Buchan and Baimbridge 1988).

After histochemical processing, immunoperoxidase-reacted sec- tions were mounted on coated glass slides, dehydrated in a graded series of alcohols, cleared in Hemo-D R (Fisher Scientific). Sections for immunofluorescence were rinsed in PBS and mounted on ovalbumin-coated slides with an anti-fading solution (0.04% para- phenylenediamine in 0.1 M sodium carbonate, pH 9.0, diluted 1 : 3 with glycerol). Two series of Nissl stained sections were also prepared.

For analysis of immunofluorescence, sections were examined with a Reichert-Jung Polyvar Fluorescence Microscope (incident light fluorescence module IB2 for fluorescein, IG2 for rhodamine), or a Confocal Laser Scanning Microscope (Wild-Leitz Instruments) using an excitation filter of 488 nm and barrier filter OG 515 nm for fluorescein (calbindin D28k); for rhodamine (GABA), the ex- citation filter was 514.5 nm and barrier filter OG 550 nm.

Sections for electron microscopy were processed for immuno- reactivity without Triton X-100, postfixed in 1% osmium tetroxide in 0.1 M phosphate buffer for 30 min, dehydrated in a graded series of alcohols and propylene oxide, and embedded in epoxy resin (TAAB R, Marovac Ltd.). Thin sections were cut onto mesh grids, stained with lead citrate and viewed in a Philips CM10 electron microscope.

Data analysis

To correlate SC laminae with calbindin immunoreactivity, alternate sections of the midbrain from two hamsters were prepared with and without immunoperoxidase processing, embedded in epoxy resin, and compared in camera lucida drawings. For quantitative studies, a 350 ~tm wide strip was projected onto the calbindin D28k or GABA immunoperoxidase-stained sections. Each strip, which was centered mediolaterally and positioned orthogonal to the surface, extended from the surface of the SC to the border of the periaque- ductal grey. To record the depth of each labelled neuron, the strips were divided into thirteen bins measuring 100 gm in depth and enclosing an area of 35,000 Ixm 2. With a camera lucida, all immu- nocytochemically labelled neurons that could be outlined clearly were recorded within each bin. Quantitative data for calbindin- immunoreactive neurons were obtained from both SC in two sec- tions through the rostral half of the tectum and one section through the caudal half of the tectum in each of two animals. Quantitative data for GABA immunoreactive neurons were obtained from both SC in one rostral and one caudal section in each of two animals. Because of the large number of neurons in the Nissl stained sections, a narrower strip (110 gm wide) was projected onto each section.

Fig. 2. Photomicrograph of a single bridge. Calbindin D28k-labelled axons and dendrites contribute to the bridge, and a few labelled neurons are also associated with it. Dashed lines indicate the ap- proximate boundaries of the bridge with the SO dorsally and the SAI ventrally

118

Quantitative data for Nissl-stained neurons were obtained from both SC in one rostral and one caudal section in each of two animals. Only profiles with a clearly stained nucleus were measured. All quantitative data were analyzed by means of an IBAS I tablet digitizer. A total of 474 calbindin-immunoreactive neurons, 591 GABA-immunoreactive neurons and 1036 Nissl stained neurons were plotted and measured.

Results

Distribution of calbindin immunoreactivity

In sections incubated with the calbindin D28k antibody, most labelled neuronal cell bodies and their processes showed intense peroxidase or fluorescence immunoreac- tivity (Figs. 1-3). In addition, there was a population of more lightly stained neurons at the ventrolateral margin of the SC (Fig. 1A); these neurons were among the largest in the hamster SC. Omitting the primary anti- body, or replacing it with rabbit serum abolished both these types of immunostaining. Such control sections (not illustrated) showed only diffuse, faint cytoplasmic staining that did not have the selective pattern observed with the calbindin D28k antibody.

Calbindin D28k immunoreactivity in the hamster superior colliculus showed distinctive patterns (Fig. 1). Labeling of neurons and neuropile was most dense in three areas of the SC: the stratum zonale (SZ) and upper part of the stratum griseum superficiale (SGS); the stra- tum opticum (SO), where a population of larger neurons was labelled; and the stratum griseum profundum (SGP), particularly medially. Little label was found in the lower part of the SGS, the stratum griseum intermedium (SGI), the stratum album intermedium (SAI) and the stratum album profundum (SAP) (Fig. 1).

A striking feature of calbindin D28k immunoreactiv- ity was the presence of labelled "bridges" connecting the SO and the SGP (Figs. 1 and 2). These bridges extended through the relatively label free SGI. The bridges, which were found in the mediolateral central one-third of the SC throughout its rostrocaudal extent, occupied a similar location in each of the animals examined. Bridges varied in thickness, and contained axons, dendrites and vertical- ly oriented neurons (Fig. 2).

In addition to the bridges extending through the SGI, calbindin D28k-immunoreactive axons were present in the coUicular commissure together with dendrites of hot-

Fig. 3A-D. Photomicrographs of calbindin D28k-immunoreactive neurons in the hamster SC. A Hor- izontal neuron, located in the SGS with a process extending into the collicular commissure (left). B Ver- tically oriented neuron located on the SGI/SAI border. C Multipolar neuron located in the SGP. D Mul- tipolar neuron located on the SAI/SGP border

119

izontal cells. The latter were particularly clear in the caudal part of the colliculus, directly ventral to the median fissure (Fig. 3A). Calbindin immunoreactive axons were also seen extending ventrolaterally from the SAI, particularly in the caudal half of the colliculus (Fig. 1B).

Calbindin irnmunoreactive neurons

Neurons stained with the calbindin D28k antibody vari- ed in their size and morphology. The size distribution of these neurons ranged from 41.3 to 308.4 gm 2 in area, with a mean of 119.2 gm 2 (S.D. 41.0), with the smallest neurons located in the first 200 gm from the collicular surface, in the stratum zonale, SGS and upper SO (Figs, 4 and 5). The largest calbindin-immunoreactive neurons were located in the intermediate layers (Fig. 5). Although the sample area measured in this study did not include the distinctive large neurons located in the ventrolateral SAI and SGP (Fig. 1A), these neurons ranged from 213.1 to 536.2 gm 2 (mean= 391.3 gm 2, S.D. 70.8, n = 42).

Several types of calbindin D28k-immunoreactive neurons could be identified, including horizontal, vertical and stellate cells (Fig. 3). There was no discernable rela- tionship between cell type and laminar distribution. Several of the neurons located in or adjacent to bridges in the stratum album intermedium were of the vertical or wide-field vertical type (Figs. 1 and 2).

The frequency distribution and size of calbindin D28k-immunoreactive neurons were compared with GABA-immunoreactive neurons and, as an indication of the total populat ion of SC neurons, with Nissl stained neurons in similar areas of the hamster SC (Fig. 4). GABA-immunoreact ive neurons were found throughout the SC, although slightly more were present in the inter- mediate than in the superficial or deep layers (Fig. 6). GABA-immunoreactive neurons ranged in size 27.2 to 158.5 gm 2 with a mean of 71.1 gm 2 (S.D. 23.3) (Fig. 4). Thus, there was considerable overlap between the sizes of the neurons showing calbindin and GABA immuno- reactivity. Compared to calbindin, the size distribution

20

>" I o

e ........ Calbindin cr - - - Gaba ~- Nissl

b_ 10

tD

13_

0 ' i �9 I " ' i I ' �9 I

0 100 200 3 0 0 4 0 0 500

Cell Area (H,m 2)

Fig. 4. The size distribution of calbindin D28k-immunoreactive, GABA-immunoreactive and Nissl stained neurons in the ham- ster SC

E v

<

o

160

140

120

100

8O

60

40

20

0

100

80

60

40

20

0

160

140

120

100

80

60

40

20

0

11 T T T

T r

500 1000

Calbindin

_T_ T I

i

Gaba

-

Depth f rom Surface (#m)

Fig. 5. The size distribution with increasing depth from the col- licular surface of calbindin D28k-immunoreactive, GABA-immu- noreactive and Nissl stained neurons. Error bars indicate standard error of the mean

t ~

E 80

o o o

u3 60 K3

69

-6 (~ 40

2 0 .t3

E

z o

o Calbindin z~ Gaba

u ' - - ' u j a - - U ~ a--u [] Nissl

~ \oG/- []

. . . . i . . . . I ' '

500 1 0 0 0

Depth f rom Surface (#m)

Fig. 6. Number of calbindin D28k-immunoreactive, GABA-immu- noreactive and Nissl stained cells per 35,000 gm z with increasing depth from the collicular surface

120

C A L B I N D I N - F I T C G A B A - R I T e

Fig. 7A-D. Confocal laser scanning photomicrographs of FITC- positive (calbindin D28k) and RITC-positive (GABA) neurons in the hamster SC. A, B Are FITC and RITC images respectively of the same field, located 150 gm ventral to the collicular surface. C, D Are FITC and RITC images respectively of the same field,

located 500 gm ventral to the collicular surface. (Asterisk) location of calbindin-immunoreactive neurons; (Star) location of GABA- immunoreactive neurons. While neurons are labelled for calbindin and GABA, there are no double labelled neurons

o f G A B A - i m m u n o r e a c t i v e neurons relative to depth f rom the surface o f the SC was remarkably consistent t h r o u g h o u t all layers (Fig. 5). The size distr ibution o f Nissl-stained neurons, which ranged f rom 16.9 to 460.6 lam a with a mean o f 95.6 lain 2 (S.D. 44.5), had a greater range than either the calbindin- or G A B A - i m m u - noreactive neurons (Fig. 4). The smallest Nissl-stained neurons were located in the superficial layers, and there was a slight increase in size in the SGI ; the largest Nissl

stained neurons were present in the SGP and SAP (Fig. 5).

Co-localization of calbindin and GABA

To check for double labelling in the same neurons, we examined sections that were reacted for the presence o f bo th labels using different secondary antibodies to vi-

121

Fig. 8A-C. Electron micrographs of calbindin D28k-immunoreactive post-synaptic processes. A Terminal containing round synaptic vesicles makes an asymmetrical contact with a calbindin- labelled profile. B Synaptic profile containing flat- tened vesicles makes a symmetrical contact with a calbindin-immunoreactive profile. C This labelled profile is contacted by an en-passant synaptic bou- ton that forms an asymmetrical synapse

sualize clabindin D28k (FITC) and GABA (RITC) immu- noreactivity. None of the labelled neurons in the SZ/SGS (n = 222), SO (n = 126), or SGP (n = 46) showed co-loca- lization of calbindin and G A B A immunoreact ivi ty (Fig. 7).

Ultrastructure of calbindin labelled neurons and processes

Processed for EM immunocytochemistry, the neuropile of the hamster SC showed several types of calbindin D28k-immunoreact ive profiles. In addition to some mye- linated axons and the cell bodies or pr imary dendrites of SC neurons, there were many calbindin-stained dendritic processes that were post-synaptic to terminal and en passant boutons (Fig. 8). Some terminals contained round vesicles and made asymmetrical contacts (Fig. 8A), suggesting they were the terminals of retinal ganglion cell (RGC) axons; confirmation with an ante- rogradely-transported tracer would be necessary, how- ever, to establish this point with certainty (Carter et al.

<

Fig. 9A, B. Electron micrographs of calbindin D28k-immunoreac- tire synaptic terminals. A Labelled axon terminal, containing round synaptic vesicles, synapses with a small, unlabelled dendritic shaft, which is also contacted by an unlabelled axon terminal that contains pleomorphic vesicles. B This labelled profile contains pleomorphic vesicles and makes symmetrical contacts with two post-synaptic profiles

122

1991). Other terminals that contacted calbindin- immunoreactive dendrites contained flattened or pleo- morphic vesicles and formed synapses that appeared to be symmetrical (Fig. 8B), an indication that they were not of retinal origin.

Calbindin D28k-immunoreactive axon terminals (Fig. 9) were less frequently observed than the immuno- reactive post-synaptic profiles. Some of these terminals contained round vesicles (Fig. 9A), while others had pleomorphic vesicles (Fig. 9B). Although the cell bodies of these axon terminals were not identified, those con- taining round vesicles may originate from a small popu- lation of retinal ganglion cells that have been shown to be calbindin-immunoreactive in several species (Pasteels et al. 1990) including hamsters (Berkelaar M, Bray GM, Villegas-Perez M, Aguayo A J, unpublished observa- tions). There were also occasional examples in which both pre- and post-synaptic profiles were calbindin- immunoreactive.

Discussion

This immunocytochemical study has demonstrated that subpopulations of neurons throughout the rostrocaudal extent of the hamster superior colliculus (SC) are labelled by an antibody to the calcium binding protein, calbindin D28k. These immunoreactive neurons are predominant- ly located in the stratum zonale and upper stratum gri- scum superficiale (SGS), in the SO, and in the stratum grsieum profundum (SGP). Although there is overlap in the size distribution of calbindin and GABA immuno- reactive neurons (Fig. 4), we did not observe any colo- calization of catbindin and GABA immunoreactivity in individual neurons in the SC. In the superficial SC, ultra- structural analysis revealed many calbindin-labelled den- dritic processes that were post-synaptic to apparent reti- nal ganglion cell (RGC) axon terminals; calbindin-la- belled axon terminals were less frequently observed.

Distribution of calbindin-immunoreactive neurons in the hamster SC

The observed laminar arrangement of calbindin D28k immunoreactivity may reflect the known segregation of specific afferent and efferent connections to the SC. The superficial layers of the SC project to, and receive input from, primarily visually-related areas (Huerta and Hart- ing 1984; Albers 1990). A major source of input to the superficial layers is the contralateral retina. Although orthogradely-transported labels were not used to prove the retinal origin of the terminals that formed synapses with calbindin-immunoreactive dendrites in the super- ficial SGS, many of these terminals had the characteristic morphology of retinal ganglion cell axon terminals (Car- ter et al. 1991). Thus, it is possible that many calbindin- immunoreactive neurons in the colliculus are the targets of retinal ganglion cell axons. In this regard, it is interest- ing to note that in the cat pretectum, groups of calbindin- immunoreactive neurons are found exclusively in retino-

recipient areas of this nuclear complex (Nabors and Mize 1990), and many receive direct retinal input (Emami et al. 1991).

In rhesus monkeys, monocular enucleation causes reduced expression of another calcium-binding protein, parvalbumin, in the neuropile of the lateral geniculate nucleus (Tigges and Tigges 1991). However, such lesions did not affect calbindin D28k immunoreactivity in the SC (Luo and Mize 1991). These results suggest differences in the responses of the geniculostriate and collicular sys- tems to visual deprivation; it has not yet been determined if similar changes in calcium binding proteins occur in species other than primates.

Many intrinsic neurons in the SGS and SO project to the deeper layers of the SC (Rhoades et al. 1989). The "bridges" between the SO and the SGP that we observed in the present study suggest that at least some of the superficial-to-deep projections are calbindin D28k-im- munoreactive (Figs. 1 and 2). We also identified a group of calbindin-labelled horizontal neurons in the superficial layers of the SC whose axons project immediately ventral to the midline fissure (Fig. 3A), presumably to terminate in the contratateral SC. Thus, calbindin immunoreactiv- ity in the superficial SC appears to be present in both intrinsic and commissural neurons.

Mize et al. (1991) argued that most calbindin D28k- immunoreactive neurons in the cat SC project within the colliculus because they could not be retrogradely labelled with injections of tracer into the dorsolateral tegmentum, the predorsal bundle, or the pulvinar, although a few were retrogradely labelled from the lateral geniculate nucleus. Although the relatively few calbindin- immunoreactive pre-synaptic profiles observed in the present study could also have arisen from interneurons, those that contained round synaptic vesicles could have been terminals of retinal ganglion cells, some of which show calbindin immunoreactivity (Pasteels et al. 1990).

Most calbindin D28k immunoreactivity in the deep layers of the SC is found in the stratum griseum profun- dum. This area receives input from a variety of sources, notably the somatosensory and auditory cortices, spinal cord and cerebellum. However, no single input to the deep SC is restricted precisely to the territory containing calbindin-labelled neurons (Rhoades 1981; Huerta and Harting 1984; Albers 1990). Calbindin-stained neurons in the deep layers may contribute to the intercollicular pathway (Fish et al. 1982). They do not appear to project in the medial efferent bundle, as no labelled axons were found in the predorsal bundle. However, in the caudal half of the SC, a distinct bundle of calbindin-immuno- reactive fibers was located in the deep layers (Fig. 1B), which may be part of the ipsilateral lateral efferent bundle. The lateral efferent bundle innervates targets in the lateral midbrain and pons (Murray and Coulter 1982; Redgrave et al. 1987).

Neurochemically distinct subgroups of neurons are abundant throughout the brain, and a variety of substan- ces have been shown to localize to specific SC laminae including GABA, enkephalin, substance P, serotonin, norepinephrine, somatostatin, choline acetyltransferase, acetylcholinesterase, calcitonin gene related peptide and

123

adenosine deaminase (see Albers 1990 for a review). However, there is no obvious correlation between the distribution of any neurotransmitter in the rat SC and the distribution of calbindin-labelled neurons. Nonethe- less, the bridges that are seen with calbindin D28k immu- noreactivity between the stratum opticum and the stra- tum griseum profundum correspond closely with sheets of substance P and enkephalin-immunoreactive fiber bundles described in the rat (Miguel-Hidalgo et al. 1989).

Possible functions of calcium-bindin9 proteins in SC n e u y o n s

Calcium has been implicated in many biochemical processes in neuronal cells including neurotransmitter release, the activation of intracellular second messenger systems and use-dependent long-term modification of neuronal excitability (Kennedy 1989). Intraneuronal cal- cium levels are modulated by voltage-regulated or trans- mitter-linked transmembrane channels, intracellular storage mechanisms, and calcium-binding proteins such as calmodulin, calbindin D28k or parvalbumin. In con- trast to calmodulin which is found in most neurons, calbindin and parvalbumin tend to be found in distinct subgroups of neurons. Although the precise functions of these calcium-binding proteins are unknown, it has been postulated that they act as intraneuronal calcium ion buffers (Baimbridge and Miller 1982; McBurney and Neering 1987) that are involved in the generation of calcium "spikes", and in providing protection against prolonged stimulation (Scharfman and Schwartzkroin 1989), or ischemic injury (Meyer 1989). The significance of subpopulations of apparently similar neurons with distinct intracellular calcium-binding mechanisms has not been determined but could reflect functional differences such as type of afferent connections or second messenger systems.

The functional significance of the frequent co- localization of G A B A and calcium-binding proteins is intriguing. In Old World monkeys, for example, GABA- ergic neurons can be categorized by their immunoreactiv- ity for parvalbumin or calbindin D28k (Hendry et al. 1989). However, a correlation between G A B A and cal- bindin does not appear to be important in the cat (Mize et al. 1991) or hamster SC (present study). It remains to be determined if there is a correspondence between GABA and parvalbumin content in the mammalian superior colliculus.

Acknowledgements. The authors wish to thank K.G. Baimbridge for a kind gift of the calbindin antibody, and P. Sithigorngul for a gift of the GABA monoclonal antibody. We thank W. Wilcox, S. Shinn, J. Trecarten and M. David for excellent technical assistance. The work was supported by a grant from the National Eye Institute (EYO4478) to M. Behan, and the Medical Research Council of Canada (G.M.B.).

References

Albers FJ (1990) Structure and organization of the superior colli- culus of the rat. Thesis Publishers, Amsterdam, pp 1-52

Baimbridge KG, Miller JJ (1982) Immunohistochemical localiza- tion of calcium-binding protein in the cerebellum, hippocampal formation, and olfactory bulb of the rat. Brain Res 245: 223-229

Bennett-Clarke C, Mooney RD, Chiaia NL, Rhoades RW (1989) A substance P projection from the superior colliculus to the parabigeminal nucleus in the rat and hamster. Brain Res 500: 1-11

Buchan AM J, Baimbridge KG (1988) Distribution and co-localiza- tion of calbindin D-28k with VIP and Neuropeptide Y but not somatostatin, galanin and substance P in the enteric nervous system of the rat. Peptides 9:333-338

Carter DA, Bray GM, Aguayo AJ (1989) Regenerated retinal gan- glion cell axons form well-differentiated synapses in the superior colliculus of adult hamsters. J Neurosci 9 : 4042-4050

Carter DA, Aguayo AJ, Bray GM (1991) Retinal ganglion cell terminals in the hamster superior colliculus: an ultrastructural study. J Comp Neurol 311:97-107

Celio MR, Scharer L, Morrison JH, Norman AW, Bloom FE (1986) Calbindin immunoreactivity alternates with cytochrome c-oxidase-rich zones in some layers of the primate visual cortex. Nature 323: 715-717

Cello MR (1990) Calbindin D-28K and parvalbumin in the rat nervous system. Neuroscience 35:375-475

Chalupa LM (1981) Some observations on the functional organiza- tion of the golden hamster's visual system. Behav Brain Res 3 : 189-200

DeFelipe J, Hendry SHC, Jones EG (1989) Synapses of double boquet cells in monkey cerebral cortex visualized by calbindin immunoreactivity. Brain Res 503:49-54

Demeulemeester H, Vandesande F, Orban GA, Heizmann CW, Pochet R (1989) Calbindin D-28K and parvalbumin immuno- reactivity is confined to two separate subpopulations in the cat visual cortex, whereas partial coexistence is shown in the dorsal lateral geniculate nucleus. Neurosci Lett 99:6-11

Emami S, Butler GD, Nabors B, Luo Q, Spencer RF, Mize RR (1991) Calbindin labelled neurons receive significant synaptic input from retinal terminals within the calbindin cell clusters of the cat pretectum. Soc Neurosci Abstr 17:113

Fish SE, Goodman DG, Kuo DC, Polcer JD, Rhoades RW (1982) The intercollicular pathway in the golden hamster: an anatomi- cal study. J Comp Neurol 204:6-20

Gerfen CR, Baimbridge KG, Miller JJ (1985) The neostriatal mo- saic: compartmental distribution of calcium-binding protein and parvalbumin in the basal ganglia of the rat and monkey. Proc Nail Acad Sci USA 82:8780-8784

Graybiel AM, Brecha N, Karten HJ (1984) Cluster-and-sheet pat- tern of enkephalin-like immunoreactivity in the superior colli- culus of the cat. Neuroscience 12:191-214

Hendry SH, Jones EG, Emson, PC, Lawson DE, Heizmann CW, Streit P (1989) Two classes of cortical GABA neurons defined by different calcium binding protein immunoreactivities. Exp Brain Res 76: 467-472

Huerta M J, Harting JK (1984) The mammalian superior colliculus: studies of its morphology and connections. In: Vanegas H (ed) Comparative neurology of the optic tectum. Plenum Press, New York, pp 687-773

Jones EG, Hendry SHC (1989) Differential calcium binding protein immunoreactivity distinguishes classes of relay neurons in mon- key thalamic nuclei. Eur J Neurosci 1 : 222-246

Kennedy MB (1989) Regulation of neuronal function of calcium. TINS 12:417-420

Luo Q, Mize RR (1991) Calbindin and GABA immunoreactivities are not reduced by monocular deprivation in the superior colli- culus of the rhesus monkey. Invest Ophthalmol Vis Sci. 32:1035

McBurney RN, Neering IR (1987) Neuronal calcium homeostasis. TINS 10:164-169

Meyer FB (1989) Calcium, neuronal hyperexcitability and ischemic injury. Brain Res Rev 14:227-243

Miguel-Hidalgo J-J, Senba E, Matsutani S, Takatsuji K, Fukui H, Tohyama M (1989) Laminar and segregated distribution of immunoreactivities for some neuropeptides and adenosine de-

124

aminase in the superior colliculus of the rat. J Comp Neurol 280: 410-423

Mize RR (1989) Enkephalin-like immunoreactivity in the cat superior colliculus: distribution, ultrastructure, and colocaliza- tion with GABA. J Comp Neurol 285:133-155

Mize RR, Jeon C-J, Butler GD, Luo Q, Emson PC (1991) The calcium-binding protein calbindin-D28k reveals subpopula- tions of projection and interneurons in the cat superior collicu- lus. J Comp Neurol 307:417-436

Mooney RD, Klein BG, Szczepanik AM, Rhoades RW (1985) Extensive recrossing of retinotectal axons after neonatal unilat- eral superior colliculus lesions in hamster. Dev Brain Res 19:297-313

Murray EA, Coulter JD (1982) Organization of tectospinal neurons in the cat and rat superior colliculus. Brain Res 243:361-365

Nabors LB, Mize RR (1990) Antibodies to calbindin delineate retinal projection zones in the pretectum. Soc Neurosci Abstr 16:109

Oertel WH, Mugnaini E (1984) Immunocytochemical studies of GABAergic neurons in rat basal ganglia and their relations to other neuronal systems. Neurosci Lett 47:233-238

Pasteels B, Rogers J, Blachier F, Pochet R (1990) Calbindin and calretinin localization in retina from different species. Vis Neurosci 5:1-16

Redgrave P, Mitchell I, Dean P (1987) Descending projections from the superior colliculus in rat. Exp Brain Res 68 : 147-167

Rhoades RW (1981) Organization of somatosensory input to the deep collicular laminae in hamster. Behav Brain Res 3:201-222

Rhoades RW, Mooney RD, Klein BG, Jacquin MF, Szczepanik AM, Chiaia NL (1987) The structural and functional charac- teristics of tectospinal neurons in the golden hamster. J Comp Neurol 255:451-465

Rhoades RW, Mooney RD, Rohrer WH, Nikoletsias MM, Fish SE (1989) Organization of the projection from the superficial to the deep layers of the hamster's superior colliculus as demonstrated by the anterograde transport of Phaseolus vulgaris leucoaggluti- nin. J Comp Neurol 283:54-70

Rogers JH 1987 Calretinin: A gene for a novel calcium-bind- ing protein expressed principally in neurons. J Cell Biol 105:1343-1353

Rohrenbech J, W/issle H, Heizmann CW (1987) Immunohis- tochemical labeling of horizontal' cells in mammalian retina using antibodies against calcium-binding proteins. Neurosci Lett 77 : 255-260

Scharfman HE, Schwartzkroin PA (1989) Protection of dentate hilar cells from prolonged stimulation by intracellular calcium chelation. Science 246:257-260

Schneider GE (1969) Two visual systems. Science 163:895-902 Schneider GE (1973) Early lesions in superior colliculus: Factors

affecting the formation of abnormal retinal projections. Brain Behav Evol 8:73-109

Sithigorngul P, Cowden C, Guastetla J, Stretton AO (1989) Genera- tion of monoclonal antibodies against a nematode peptide ex- tract: another approach for identifying unknown neuropep- tides. J Comp Neurol 284:389-397

Stichel CC, Singer W, Heizmann CW, Norman AW (1986) On- togeny of calcium-binding proteins, parvalbumin and calbindin- D 28K in the dorsal lateral geniculate nucleus (LGNd) of the cats. Neurosci Lett Suppl 26:$55

Stichel CC, Singer W, Heizmann CW, Norman AW (1987) Immu- nohistochemical localization of calcium-binding proteins, parvalbumin and calbindin-D 28k, in the adult and developing cortex of cats: a fight and electron microscopic study. J Comp Neurol 262: 563-577

Tigges M, Tigges J (1991) Parvalbumin immunoreactivity of the lateral geniculate nucleus in adult rhesus monkeys after mono- cular eye enucleation. Vis Neurosci 6:375-382

Van Brederode JFM, Mulligan KA, Hendrickson AE (1990) Calcium-binding proteins as markers for subpopulations of GABAergic neurons in monkey striate cortex. J Comp Neurol 298 : 1-22

Woodson W, Nitecka L, Ben-Ari Y (1989) Organization of the GABAergic system in the rat hippocampal formation: A quan- titative immunocytochemical study. J Comp Neuro1280 : 254 271