THE JOURNAL OF COMPARATIVE NEUROLOGY 298:l-22 (1990)

Calcium-BindingproteinSas Markersfor Subpopulatio& of GABAergic Neurons inMonkey Striate Cortex J.F.M. VAN BREDERODE, JLA. MUJLIGAN, AND A.E. HENDR.ICKS0N Departments of Biological Structure (J.F.M.V.B., K.A.M., A.E.H.) and Ophthalmology (A.E.H.), University of Washington School of Medicine, Seattle, Washington 98195

ABSTRACT Recent studies have shown that the presence of immunoreactivity for parvalbumin (PV-IR) and calbindin-D 28k (Cal-IR) can be used as markers for certain types of gamma-aminobutyric acid (GABA) immunoreactive interneurons in monkey cerebral cortex. Little quantitative information is available regarding the features that distinguish these two subpopulations, however. Therefore, in this study we localized PV-IR and Cal-IR neurons in Mucaca monkey striate cortex and analyzed quantitatively their laminar distribution, cell morphology, and co-localization with GABA by double-labeling immunocytochemistry. PV-IR was found in nonpyramidal cells in all layers of the cortex, although PV-IR cells in layer 1 were rare. In contrast, Cal-IR was found mainly in nonpyramidal cells in two bands corresponding to layers 2-3 and 5-6. We found very few double-labeled PV-IR/Cal-IR cells but confirmed that almost all PV-IR and Cal-IR cells are GABAergic. Overall, 74% of GABA neurons in striate cortex displayed PV-IR compared to only 12%that displayed Cal-IR and 14% that were GABA-IR only. Quantitative analysis indicated that the relative proportion of GABA cells that displayed PV-IR or Cal-IR showed conspicuous laminar differences, which were often complementary. Cell size measurements indicated that PV-IR/GABA cells in layers 2-3 and 5-6 were significantly larger than Cal-IR/GABA cells. Analysis of the size, shape, and orientation of stained cell bodies and proximal dendrites further demonstrated that each subpopulation contained several different types of smooth stellate cells, suggesting that Cal-IR and PV-IR are found in functionally and morphologically heterogeneous subpopulations of GABA neurons. There was a thick bundle of PV-IR axons in the white matter underlying the striate but not prestriate cortex. PV-IR punctate labeling matched the cytochrome oxidase staining pattern in layers 4A and 4C, suggesting that PV-IR is present in geniculocortical afferents as well as intrinsic neurons. Cal-IR neuropil staining was high in layers 1,2,4B,and 5, where cytochrome oxidase staining is relatively low. We did not find a preferential localization of either PV-IR or Cal-IR cell bodies in any cytochrome oxidase compartments in layers 2-3 of the cortex. These findings indicate that PV and Cal are distributed into different neuronal circuits. Key words: immunocytochemistry,laminar analysis, cell size, GABA, interneurons

In monkey cortex neurons that display immunoreactivity van Brederode et al., '89) have described two Ca2+-binding for gamma-aminobutyric acid (GABA-IR)can be subdivided proteins, calbindin-D 28k (or 28 kd Vitamin D dependent and classified according to differences in morphology as Ca2+-bindingprotein) and parvalbumin, whose distribution revealed by Golgi preparations (Lund, '87; Lund et al., '88), is limited to distinct populations of nonpyramidal neurons. co-localization with neuropeptides (Hendry et al., '84; A majority of the cells that show parvalbumin immunoreacSomogyi et al., '84b; Jones and Hendry, '86), and cell tivity (PV-IR) also have been found to display immunoreacsurface markers (Naegele and Barnstable, '89; Mulligan et tivity for GABA (Celio, '86; Hendry et al., '89; van Brederode al., '89; Mulligan and Hendrickson, '89). Recently, immuno- et al., '89). Most cortical cells displaying immunoreactivity cytochemical studies in cerebral cortex of rat (Celio and for calbindin-D 28k (Cal-IR) also show GABA-IR, but the Heizmann, '81; Feldman and Christakos, '83; Celio, '86; number of GABAergic Cal-IR cells seems to be dependent Kosaka and Heizmann, '891, cat (Stichel et al., '87; Demeule- on animal species, antiserum used, and intensity of staining meester et al., '88; Demeulemeester et al., '891, and monkey Accepted March 30,1990. (Celio et al., '86; DeFelipe et al., '89a,b; Hendry et al., '89; o 1990 WILEY-LISS, INC.

2

J.F.M. VAN BREDERODE ET AL.

(Feldman and Christakos, '83; Celio et al., '86; Stichel et al., '87; Demeulemeester et al., '88; Morrison et al.,'88). In cat and monkey cortex, Cal-IR and PV-IR are present in nonoverlapping subpopulations of GABA cells (Demeulemeester et al., '89; Hendry et al., '89; van Brederode et al., '89), but quantitative information about the relative proportion of PV- and Cal-IR GABA cells in each cortical layer is not available. Furthermore, a detailed description of the various morphological or functional types of GABAergic cells in the striate cortex that show immunoreactivity for Cal or PV is lacking. It is not known why immunoreactivity for Cal and PV is found only in some cortical neurons and not in others, unlike the related and ubiquitous Ca2+-bindingprotein calmodulin (Heizmann, '88). Although both proteins bind Ca2+ions with micromolar dissociation constants and possibly play a role in the buffering of a stimulus-induced increase in intracellular Ca2+(Baimbridge et al., '82; Heizmann, '88), their differential distribution in two, nonoverlapping subpopulations of GABAergic cortical neurons suggests that each protein has a different function. For instance, in the visual system of the zebra finch, PV is present in neurons that contain high levels of cytochrome oxidase activity (Braun et al., '85). In the hippocampus, PV is localized preferentially in nonadapting, fast-spiking interneurons (Kawaguchi et al., '871, while Cal is found mainly in CA1 pyramidal and dentate granule cells (Baimbridge et al., '82) that show lower maintained firing rates in comparison. A relatively high Cal content has been associated with the resistance of certain subclasses of neurons to Ca"mediated neuronal injury due to hypoxia (Wasterlain et al., '88) or glutamate toxicity (Baimbridge and Kao, '88). A further indication that PV is associated with high metabolic activity is found in squirrel monkey visual cortex. Both neurons and neuropil differ in respect to the type of Ca"-binding proteins present; cytochrome oxidase-poor, nongeniculate-input compartments are rich in Cal, while cytochrome oxidase-rich, metabolically active, geniculateinput compartments contain PV (Celio et al., '86). This study utilizes these two Ca"-binding proteins detected immunocytochemically for further quantitative analysis of Macucu monkey striate cortex organization. Laminar distribution, cell size, and morphology of PV-IR and Cal-IR cells are shown and we demonstrate by doublelabeling immunocytochemistry that these are separate populations of smooth stellate cells. We also compare the PV-IR and Cal-IR populations quantitatively to the distribution, size, and morphology of the GABA cell population, and find that most GABA neurons contain one of these Ca"binding proteins and that conversely these proteins are almost entirely confined to GABA neurons. In addition, we examine the relationship between PV-IR and Cal-IR and the pattern of cytochrome oxidase staining in striate cortex to see if these proteins have any consistent relationship to the well-studied functional compartments of Old World monkey visual cortex (Livingstone and Hubel, '82; Hendrickson, '85).

MA'IERIALSANDMETHODS We used striate (primary visual) cortex tissue from both adult male and female Macaca fascicularis and Macaca nemestrina monkeys for this study; no differences were found between the two species. Animals were deeply anes-

thetized with barbiturate and perfused transcardially with 0.5 liter of lactated Ringeddextrose followed by 2-3 liters of 4% paraformaldehyde, 0.1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4; after 45 minutes, 1 liter of 4% paraformaldehyde was perfused and after another 45 minutes the brain was rinsed by sequential perfusion with 0.5 liter each of lactated Ringeridextrose and 20% sucrose in the same phosphate buffer. Small blocks of known orientations were dissected from striate cortex and sunk in 30% sucrose. An alternating series of frozen sections, 20 and 40 km thick, was cut perpendicular or tangential to the pial surface and collected in 0.1 M phosphate buffer. Quantitative measurements (numbers and sizes of immunoreactive cells) were performed on tissue sections from three animals. In two animals, tissue blocks were cut from only areas representing the central visual field; in the other animal staining patterns were compared in sections representing either central or peripheral visual fields. Cell morphology was studied in a total of seven animals including the three animals used for the quantitative studies. We used two different, commercially available polyclonal antisera to GABA, one raised in rabbit (Incstar), and the other one raised in rat (Eugene Tech). The polyclonal antiserum raised in rabbit against chicken intestinal Cal and the mouse monoclonal antibody against carp PV were generous gifts of M. Celio and have been described previously (Celio, '86; Celio et al., '86, '88). Both antisera against GABA were used at 1:1,000 to 1:2,000 dilutions when GABA was localized as the first antigen and at 1 2 5 0 to 1:750 when localized as the second antigen in doublelabeling studies. Both antisera against GABA gave qualitatively and quantitatively similar staining patterns, which differed only in intensity of staining of cell processes. Cal antiserum was diluted 1:3,000 to 1:10,000 when reacted as first antigen and 1:2,000 when reacted as second antigen, or for the study of detailed cell morphology in single-labeling studies. PV antibody was diluted 1:10,000 to 1:20,000 when localized as the first antigen and 1:10,000 when localized as the second antigen.

Immunohistwhemicalprocedures Single-labeling. Free floating sections were incubated at 4°C in primary antiserum diluted in Tris buffered saline (TBS) containing 0.2% Triton and 1%normal goat serum for 24-92 hours. Incubations with the secondary antiserum (goat-anti-rabbit,-rat, or -mouse IgG diluted 1:30) and with the subsequent species-specific peroxidase-antiperoxidase (dilution of 1:100), were carried out at 37°C for 45 minutes each. The peroxidase was visualized with one of three chromogens; carbazole (Biomeda Corp.), which produces a bright red reaction product; 4-chloro-naphthol (4-CI-N), which produces a blue-purple reaction product (Bowker et al., '82); or diaminobenzidine-HC1 (DAB),which produces a brown reaction product. For better control the peroxidase reaction was performed at room temperature after preheating the carbazole or 4-C1-N chromogen solution to 37°C. With this procedure peroxidase visualization was achieved gradually in 10-20 minutes and was slow enough to permit visual inspection of the staining intensity every 2-3 minutes. After rinsing in phosphate buffer, sections stained with the DAB reaction were mounted, cleared, dehydrated, and coverslipped, while carbazole- and 4-C1-N-stained sections were mounted in glycerol and coverslipped. In all cases adjacent perpendicular sections were processed for

PARVALBUMIN A N D CALBINDIN-D 28k IN STRIATE CORTEX

3

cytochrome oxidase to allow precise analysis of laminar borders, and some sections were counterstained with cresyl violet for the same purpose. Double-labeling for GABAIPV, GABAICal, and PVI Cal. The first antigen was stained by the procedure described above with carbazole. After thorough rinsing in TBS, the same sections were then incubated in antiserum against the second antigen and stained with 4-C1-N as the chromogen. Because these prolonged incubations resulted in some degradation of tissue quality, the primary antiserum incubation in both steps was limited to 24 hours. As one control we reversed the order in which the antisera were used in the double-labeling procedure (i.e., PV followed by Cal vs. Cal followed by PV) and also the chromogen order (4-C1-N for the first antigen followed by carbazole for the second antigen vs. carbazole followed by 4-C1-N).A second control consisted of the elimination of the first or second primary antiserum; no antigen carry-over or species cross reactivity was found. Typically, double-labeled sections contained three classes of cells: single-labeled bright red cells from the first antigen, single-labeled clear blue cells from the second antigen, and double-labeled purple cells. The different colored chromogen products were examined both under low power and by high-power Nomarski optics and gave good enough contrast to classify stained cells as single- or double-labeled. Classification was further aided by the observation that the Cal and PV antisera stained both cell bodies and cell processes (see Results), while the rabbit anti GABA antiserum stained mainly cell bodies. Double-labeled cells therefore showed a mixed-color cell body surrounded by single-color cell processes. We also observed that double-labeled cells acquired a “glossy” appearance under transmitted light when compared to more granular single-labeled cells.

derived from stained cells in eight 500 pm columns in four monkeys, while diameters for PV-IR and Cal-IR cells were derived from three 500 pm columns in two of these monkeys. The distributions of the cell diameter of doublelabeled cells within each layer were compared by means of the unpaired Student’s t test, and differences in cell diameter of single-labeled cells between layers by a one-way analysis of variance with Scheffe’s test. Significance was set at P < 0.05. The distribution of PV-IR and Cal-IR neurons was compared for regions of layers 2 and 3 that contained cytochrome oxidase-rich zones, which have been called “dots,” “puffs,” “patches,” or “blobs” (for review see Hendrickson, ’85). Serial tangential and perpendicular 40 p,m sections, spaced 80 pm apart, were incubated for cytochrome oxidase (Wong-Riley, ’79) and the 20 pm thick sections in between were stained for either PV or Cal with DAB as outlined above. Photographs of adjacent sections were aligned, by using blood vessels and other landmarks, to determine the distribution of PV-IR, Cal-IR, and cytochrome oxidase-rich or -poor regions.

Quantitative analysis

Distribution of parValbuminimmunoreactivity

The laminar distributions of GABA-IR, PV-IR, and CalIR neurons were analyzed by using methods described previously (Mulligan et al., ’89). Briefly, all of the labeled neurons in eight nonoverlapping 500 p m wide strips of the entire thickness of striate cortex were traced at a magnification of 350 x with the aid of a camera lucida. Each strip was subdivided into 50 p,m horizontal sectors and the number of immunoreactive cells in each of these sectors was counted. Cell counts from sectors of the same depth were summed and normalized to achieve a sufficiently large population of labeled cells at different depths and to allow for comparisons between antigens. Only stained neurons in which a nucleus was visible were counted. The cross-sectional area of the somata of labeled neurons within the same 500 pm wide columns of striate cortex was measured with the aid of a computer-video microscope system (Curcio and Sloan, ’86). The neurons were first identified with transmitted light, then visualized with Nomarski differential interference optics, and displayed on a black and white video monitor at a magnification of 2,000 X. The cell outline was traced from the video screen onto a graphics tablet at the optical plane showing the greatest cross-sectional area delineated by the perikaryal membrane. For convenience, cell size was expressed as cell diameter; this was derived from computer-calculated cell surface area based on the assumption that the shape of the cell outlines could be approximated by a circle with equivalent surface area. Cell diameters of GABAergic cells were

PV-IR neurons were located in all striate layers, although labeled cells were only rarely seen in layer 1 (Fig. 1A). The numbers and distribution of PV-IR cells were consistent over the range of antibody dilutions tested. PV-IR was found throughout the neuron in nuclei, perikarya, and cell processes. The intensity of immunostaining of PV-IR cell bodies showed little variation between cortical layers. Based on cell morphology the large majority of PV-IR neurons in striate cortex were nonpyramidal, but the shape, number, and size of PV-IR cells varied considerably from layer to layer (Fig. 1A; Table 1). PV-labeled processes and puncta were also present in all cortical layers, but were heaviest in layers 4A and 4C, producing two dense bands of label (Fig. lA, 2A). At high magnification many PV-IR puncta were seen throughout layer 4C (Fig. 2B), which probably represents axon terminals that could be both intrinsic or extrinsic in origin (see below). An almost equally dense band of PV-IR puncta was found in layer 4A, but in contrast to the continuous band of labeling in layer 4C, the pattern in 4A was discontinuous (Fig. lA, Fig. 8). A moderate density of PV-IR neuropil labeling was found in layers 2, 3, and 6 , while density was lowest in layers 1, 4B, and 5 . Sometimes PV-IR puncta could be seen surrounding the cell bodies (Fig. 2, inset) and proximal dendrites of unstained cells. In the white matter underlying layer 6, PV-IR axons formed thick dense bundles, which were confined exclusively to the white matter underlying striate cortex. In sections that contained both

RESULTS In sections of adult macaque striate cortex, immunohistochemical staining with antisera to the two calciumbinding proteins produced consistent but different patterns of specificallylabeled cells and neuropil. No specific staining was detected in control sections when the primary antiserum was omitted. Both antisera labeled heavily the cell bodies and processes of nonpyramidal cells as well as punctate profiles in the neuropil. Sections immunostained for Cal contained some lightly labeled cell bodies that could be pyramidal neurons (see below).

J.F.M. VAN BREDERODE ET AL.

4

Fig. 1. Comparison between tissue sections immunostained for PV-IR (A), Cal-IR (B), and both PV-IR and Cal-IR (C). A Photomicrograph of cell and neuropil labeling for PV-IR in striate cortex. PV-IR cell bodies are found in all layers, but are rare in layer 1. Two prominent bands of PV-IR neuropil labeling can be seen in layers 4Aand 4C. PV-IR neuropil in 4A appears discontinuous; B: Photomicrograph of heavily stained Cal-IR cells, which are found in two bands corresponding to

layers 2-3A and 5-6, respectively. Notice the absence of Cal-IR cells in layer 4C. C: Camera lucida drawing of stained cells in a single tissue section double-labeled for PV-IR (filled circles) and Cal-IR (open circles). Very few cells containing both Ca”-binding proteins (stars) were found. Cortical layers are indicated on the right. Scale bars = 200

striate and prestriate cortex, these PV-IR axon bundles terminated abruptly at the border between the two areas (Fig. 2A). At higher magnification individual PV-IR fibers could be followed upward from the white matter into layer 6, but then were usually lost in the neuropil (Fig. 2C). Similar thick, obliquely and horizontally oriented PV-IR fibers were detected in layers 3 through 6. In layers 2 and 3, PV-IR processes of irregular caliber were mainly vertically oriented and probably represented the stained dendrites of PV-IR cells in these layers (Fig. 2D). Occasionally, vertically oriented PV-IR processes crossed the layer 1-2 border and could be seen running parallel to the pia for long distances in the middle of layer 1 (Figs. lA, 2D). Although an extensive study was not undertaken, it appeared that these PV-IR axon-like fibers in layer 1 were more prevalent in

striate cortex containing the peripheral visual field representation.

@n.

Distribution of calbindinimmunoreactivity The pattern of Cal-IR differed markedly from that observed for PV-IR, and overall, many fewer cells were labeled with Cal. Heavily stained Cal-IR neurons were found in a dense narrow band in layer 2 and in a wider band corresponding to layers 5 and 6 (Fig. 1B). Occasionally, Cal-IR cells in the superficial layers were observed that appeared to send thin axon-like processes to the deep layers of the cortex. In layers 3 through 4B, Cal-IR cell bodies were scattered throughout the neuropil, but almost no heavily stained Cal-IR cell bodies were found in layer 4C. Cal-IR cells were also found in the white matter immediately below

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

5

TABLE 1. Mean Cell Diameter (km) of GABA-, PV- and Cal-IR Neurons in Striate Cortex of Macaque Monkeys' Layer

GABA+~ Range PV+/Cd-3 Range cal+lpv-3

1

2-3.4

3B-4.4

4B

4c

5-6

8.9 (n = 62) 7.0-11.2

10.1 (n = 2431 6.7-15.2 11.6 (n = 37) 9.414.3 9.3 in = 56) 7.W11.5

11.6 (n = 213) 7.1-16.8 11.2 (n = 101) 8.516.1

12.7 (n = 93) 7.618.4 11.5 (n = 58) 7.ai4.4

10.7 in = 215)

-

-

-

11.6 (n = 206) 6.ain.9 11.7 (n = 41) 9.617.1 8.8 in = 20) 7.k10.4

-

7.0-18.5

10.1 (n = 72) 8.3-13.1

'Results areexpressedas mean values of cell diameter of differentcell classes in each layer and the range (minimum value-maximum value)of diameters found in each group. n = number of cells measured in each layer. 'Cell diameter of GABA-IR neurons fmm tissue sections immunostained for GABA only. SDiameterof cells that were PV-IR only (W+/Cal- 1or Cal-IR only CCal+/PV-) determinedfmm single tissue section double-labeled for both antigens.

layer 6. Both the cell number and the density of neuropil labeling increased with increased concentration of antiserum. When Cal antiserum was used at 1:2,000 many lightly stained Cal-IR cells appeared, as a prominent band in upper layer 4Ca at its border with layer 4B. While it was clear from' the pattern of labeled dendrites that the heavily labeled Cal-IR neurons were nonpyramidal, identification of the lightly stained cells was not possible since their dendritic processes were either unlabeled or were very fine and barely visible with light microscopy. For the remainder of this paper the analysis of Cal-IR cells will be limited to the population of heavily staining Cal-IR neurons. Neuropil labeling with Cal consisted of labeled puncta and processes combined with a diffuse but specific labeling that was difficult to resolve light-microscopically. Neuropil labeling was highest in cortical layers 1 through 3, at the 4B/4Ca border, and in layer 5 (Fig. 3A). Dense, small Cal-IR puncta could be resolved at higher magnification in layers 1 and 2, which probably represent axon terminals and dendritic processes of Cal-IR cells in the supragranular layers (Fig. 3B). Neuropil labeling with Cal was virtually absent in layers 4A, lower 4Ca, and 4Cb (Figs. lB, 3A). Thin axon-like Cal-IR horizontal processes were commonly seen in layers 5 and 6 and occasionally, thin axon-like Cal-IR processes were found running parallel to the pial surface for long distances in layer 1.A few thin axon-like Cal-IR processes were also observed in layer 6 and in the underlying white matter at the same level as the PV-IR axon bundles. Vertical bundles of varicose Cal-IR processes, reported by DeFelipe et al. ('89b) to be processes of double bouquet cells in layers 2 and 3, were only infrequently seen in striate cortex, although they were abundant in adjacent prestriate cortex. The complementary staining patterns of PV-IR and Cal-IR cells suggested that they comprised two different nonpyramidal cell populations. In order to test if there was any overlap between these populations, we performed double-labeling experiments using different chromogens to demonstrate PV and Cal in the same section. Comparisons with single-labeled sections showed that double-labeling did not alter the laminar distribution, labeled cell types, or number of either population. As illustrated in Fig. lC, Cal-IR and PV-IR exist in two virtually nonoverlapping neuronal populations, and only rarely was a double-labeled neuron identified.

Quantitative analysisof laminar distribution and cell size Fig. 4A shows a histogram of the average number of PV-IR and heavily labeled Cal-IR cells plotted as a function of distance from the pial surface. Overall, PV-IR cells outnumbered Cal-IR cells by 6 to 1, and the distribution of

the two populations was complementary. The number of heavily labeled Cal-IR cells peaked in upper layer 2 where only a small number of PV-IR cells were found, but from upper layer 2 to layer 4A the number of PV-IR cells gradually increased while the number of Cal-IR cells steadily declined. In layer 4A large numbers of PV-IR cells were found along with only a few Cal-IR cells, while in layer 4Cb no heavily labeled Cal-IR cells were detected and the number of PV-IR cells peaked. To characterize the PV-IR and Cal-IR cells further, their somatic sizes were determined from highly magnified video images of the same 500 )*m wide perpendicular strips used for laminar analysis. Mean cell diameters of Cal-IR cells were significantly smaller than those of PV-IR cells in the supra- and infragranular layers where both cell populations were found (Fig. 4B,C; Table 1).The smallest immunostained neurons in these layers ( < 9 pm in diameter) were exclusively Cal-IR, the largest neurons ( > 11.5 pm) were exclusively PV-IR, and medium-sized cells were either PVor Cal-IR (Fig. 4B,C; Table 1). Size is not an exclusive predictor of Ca2+-bindingprotein content, because in layers 4B and 4C, where Cal-IR cells are virtually absent, many PV-IR cells were found with diameters between 7 and 9 pm (Table 1).

Double labelingof PV-IRand Cal-JR neurons withGABA Several observations from the present work suggested that the calcium-binding proteins co-localizewith GABA, as indicated by others (Celio, '86; Braun et al., '88; Demeulemeester et al., '88; Hendry et al., '89). Both PV-IR and heavily labeled Cal-IR neurons appeared to be exclusively nonpyramidal, and their somatic sizes lay within the range reported for GABAergic cells (Mulligan et al., '89). When the numbers of Cal- and PV-IR neurons in each 50 pm horizontal section were combined (Fig. 4A), this produced a distribution pattern with peaks in layers 2, 4A, and 4C. This combined pattern is strikingly similar to the laminar distribution of GABA-IR cells in monkey striate cortex (compare Fig. 4A to Fig. 5A). These results suggest that a large proportion of GABA-IR cells contains either Cal or PV, but not both, and that, conversely, most PV- and Cal-IR cells are GABAergic. However, this comparison found that the combined total number of cells that stained for PV-IR or Cal-IR was somewhat less than the total number of GABA-IR cells (compare Figs. 4A and 5A), suggesting that not all GABA-IR neurons are PV-IR or Cal-IR. To investigate the extent of co-localization, doublelabeling experiments comparing antiserum to GABA and to each of the Ca2+-bindingproteins were undertaken. G A B 4 PV staining patterns (Fig. 5B) show that about 74% of GABA-IR cells in monkey striate cortex displayed PV-IR.

J.F.M. VAN BREDERODE ET AL.

6

Figure 2

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

7

Fig. 3. Photomicrographs of perpendicular sections of striate cortex immunostained for Cal-IR. A Low-power photomicrograph showing three bands of Cal-IR neuropil corresponding to layers 1 through 3B, the 4Bi4Ca border, and layer 5. Scale bar = 500 pm. B: High-power

photomicrograph of Cal-IR in layers 1 and 2. Heavily stained Cal-IR cells are nonpyramidal with thin, widely spreading dendrites (arrows). Notice the many, small Cal-IR puncta in these layers (arrowheads) and the diffuse background staining. Scale bar = 50 pm.

We estimated that only 4% of PV-IR cells lacked immunoreactivity for GABA in this sample, but the exact number of PV-only cells was difficult to determine due to the very intense staining of some PV-IR cells, which tended to obscure the GABA label. In layer 2, 30% of GABA-IR cells showed PV-IR, and this number gradually increased until in layer 4C almost 100%of GABA-IR cells stained for PV. In layers 5 and 6 the relative number of PV-IR GABA cells dropped to 60%. The GABNCal staining pattern is illustrated in Fig. 5C. Overall only 12% of all GABA-IR neurons were immunoreactive for Cal, and these were located in narrow bands in

layers 2-3A and 5-6. In layer 2 about 40% and in layer 5-6 about 10% of all GABA-IR neurons were Cal-IR. About 6% of Cal-IR neurons did not show GABA-IR; these cells were located mainly in the supragranular layers. We calculated from sections of the striate cortex immunostained for only GABA that on average 398 cells per whole 500 pm wide column are GABAergic, while on average 382 cells/column were classified as GABAergic from double-stained sections. Therefore, double-labeling for GABNPV or GABNCal did not result in more cells per column than in single-labeled sections stained for GABA-IR alone, providing further support for the conclusion that almost all PV-IR and Cal-IR cells are GABAergic. Also, from Figs. 4 and 5 it is clear that the absolute number of cells that stained for GABA, PV, or Cal in different cortical lamina with different combinations of antisera was remarkably consistent. Based on these results, we calculate that 86% of all GABA neurons were either Cal- or PV-IR. The remaining 14% of GABAergic cells that were GABA-only were located mainly in layers 1, 2-3A, and 5-6.

Fig. 2. Photomicrographs of PV-IR in perpendicular sections of striate cortex. A: Low-power photomicrograph showing the striate/ prestriate border (arrowheads in striate cortex). Notice the dark bands of PV neuropil labeling in layers 4A and 4C in striate cortex and their abrupt termination at the border. PV-IR fibers can also be seen in the subcortical white matter (WM)underlying striate but not prestriate cortex. Scale bar = 500 pm; B: High-power photomicrograph showing PV-IR cells and puncta in layers 4C and 5. Notice the increased density of PV-IR puncta in layer 4C compared to layer 5. Scale bar = 50 pm. Inset shows PV-IR puncta surrounding the cell bodies of unstained cells in layer 4Cb. Scale bar = 25 pm. C: Photomicrograph showing PV-IR axons in the white matter (WM) of striate cortex and PV-IR cell bodies and neuropil in layer 6. PV-IR axons can be seen running obliquely between layer 6 and the underlying white matter (arrows). Scale bar = 100 pm. D: Photomicrograph showing PV-IR in layers 1 and 2. Notice the punctate nature of the neuropil and the large PV-IR cell in layer 2 with mainly vertically oriented varicose dendrites, which extend into layer 1 (arrows). Occasionally, horizontally oriented PV-IR axons are seen running parallel to the pia (arrowheads in layer 1). Scale bar = 50 pm.

Cell size of GABAergic neurons in monkey striatecortex If close to 90% of all GABA-IR neurons contain either PV or Cal, then differences in cell size between Cal- and PV-IR neurons (see Fig. 4B and C) should be reflected in inhomogeneities in the cell size distribution for GABA cells. GABA-IR neurons show a characteristic laminar variation in soma size (Fig. 6; Table 1).From layers 1to 4B there is a general trend toward an increase in mean cell diameter, while cells in layer 4C are significantly smaller (P < 0.05). Although mean cell diameter of GABA-IR cells in layers 5-6

J.F.M. VAN BREDERODE ET AL.

8

A

0 Cal

13 PV

1 CalfPV

0

1 2

3 D

E

4A 4B

500

P T

4c

H 1000

-I 0

10

5

::

15

20

5 6

25

NUMBER OF CELLS

B P E R

C E

N T

50

20

Relationship of parvalbuminand calbindinto cytochrome oxidase staining

10

Both tangential and perpendicular serial sections of striate cortex were analyzed. Although the distribution of Cal-IR cells occasionally appeared to be patchy in layers

0

C P E R

C E N

T

~l

5-6 PV Cal

20

10

06

was larger than that in layer 4C ( P < 0.051, there still is a large number of cells with diameters < 10 pm in the deep cortex. The shape of the size distributions varied considerably from layer to layer, with some distributions showing more than one peak and others being skewed, suggesting that inhomogeneities in GABA-IR cells exist not only between layers but also within layers. It is clear from Fig. 6 and Table 1that the cells in layer 1 displayed the smallest variation in soma size, and were smaller on average than cells in all other layers (P < 0.05). In layers 2-3A the distribution appeared to be bimodal, and the average soma size was smaller than in all of the deeper layers (P < 0.05) except layer 4C. The distribution in layers 5-6 was skewed to the left, i.e., toward a large number of small cells, while those in layers 3B-4A, 4B, and 4C were more symmetrical around the mean. GABA-IR neurons > 15 pm were found only in layers 4B, 4Ca, 5 , and 6. A striking similarity can be observed between the shapes of the combined size distributions for Cal- and PV-IR cells (Fig. 4B.C) and those of GABAergic cells in corresponding layers (Fig. 6). By comparing Figs. 4B, 4C, and 6, it can be seen that the bimodal distributions of GABAergic cells in layer 2-3A can be accounted for by combining the larger GABA/PV and smaller GABA/Cal subpopulations. A similar explanation holds for the skewed distributions of GABA cells in layers 5-6. Cell size measurements on tissue sections double-labeled for GABA and Cal-IR confirmed that small-to-medium cells were GABA-IR/Cal-IR, while medium-to-large cells displayed only GABA-IR. The opposite was true for sections double-labeled for GABA and PV (results not shown). In layer 4 the mean cell diameter and range of sizes of PV-IR cells were comparable to that of GABA-IR cells, in agreement with our observations that almost 100%of GABA cells in layer 4 show PV-IR. PV-IRI GABA-IR cells are significantly smaller in layer 4C than those in the adjacent layers immediately above or below (Fig. 5B; Table 1).

8

10

12

14

16

18

CELL DIAMETER (pm)

20

Fig. 4. Diagrams showing the laminar distribution (A) and cell size (B and C) of PV-IR and Cal-IR neurons. A Number of PV-IR and Cal-IR cells as a function of cortical depth. Depth profiles of three classes of cells are shown: cells that were PV-IR only (shaded bars), Cal-IR only (open bars), and cells that were both PV-IR and Cal-IR (solid bars). Each bin represents a 50 x 500 pm field of striate cortex that was part of a horizontal column extending from the pia (depth equals zero) to the border between layer 6 and the white matter (depth equals 1,150 pm). The height of the individual bars represents the average cell count determined from three, nonoverlapping columns. Corresponding cortical layers are indicated on the right. Notice the peaks in layers 2,4A, and 4C, which are the result of adding (or stacking the individual bars as in the graph) the number of PV-IR and Cal-IR cells in each bin. B and C: Histograms of the cell diameters of PV-IR (open bars) and Cal-IR cells (shaded bars) in layers 2-3A (B) and 5-6 (C). Cell area was measured from highly magnified computer-videomicroscope images with the aid of a graphics tablet. Cell size was derived from double-labeled single tissue sections, eliminating errors due to differential shrinkage, and expressed as diameter of a circle with equivalent surface area. Cell counts are expressed as the percentage of the total number of cells that fell into each size category. Both graphs show that PV-IR cells are larger than Cal-IR cells.

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

A

0 GABA c

1 2 3

4A 4B

500

4c

5 6

I000

0

B

5

10

15

0 GABA

25

20

30

4 40

35

0 GABA/PV H PV 1 2

3 4A 4B 4c

5 6

1

0

5

15

10

20

25

30

I 40

35

C GABA/Cal 0

Cal 1 2

3 4A 4B

500

4c I

I

5

6 I

0

5

10

15

20

25

30

NUMBER OF CELLS

35

40

9

2-3A, no clear correlation was found between Cal-IR or PV-IR cell distribution and the localization of cytochrome oxidase (Fig. 7A-D). On the other hand, neuropil labeling for Cal-IR was strongest in layers where cytochrome oxidase neuropil labeling was light, namely in layers 1, 2, 4B/4Ca, and 5 (see Fig. 3A), while the PV-IR neuropil labeling closely corresponded to the cytochrome oxidase distribution, particularly in layers 4A and 4C (Fig. 8). Thus PV-IR neuropil labeling overlapped the geniculocortical terminal zones in 3, 4A, and 4C (Hendrickson et al., ’78; Blasdel and Lund, ’82; Livingstone and Hubel, ’82). PV neuropil labeling in layer 4A in tangential sections was reticular and corresponded to the “honeycomb” pattern of geniculate input (Hendrickson et al., ’78) and cytochrome oxidase activity (Fig. 7A,C). In perpendicular sections PVIR neuropil in layer 4A was patchy and discontinuous, and closely matched the cytochrome oxidase pattern (Fig. 8A,B). In perpendicular sections (Fig. 8A,B) there was a good correspondence between cytochrome oxidase “dots” and PV-IR neuropil in layer 3, but in en face sections this was less evident (Fig. 7A,C). In layers 2 and 3 putative geniculocortical PV-IR neuropil labeling is probably at least partially obscured by the large numbers of labeled dendritic processes and axons of intrinsic PV-IR cells. The coincidence of the high density of PV-IR terminals in the principal thalamocortical recipient layers together with the abundance of PV-IR axon bundles in the subcortical white matter only under striate cortex strongly suggests that a PV-IR projection exists from the dorsal lateral geniculate nucleus (dLGN) to the striate cortex in Old World monkeys.

Morpholo~ of PV-IR and Cal-IR~ l l s Staining with antisera to PV or Cal revealed not only the somata but considerable portions of the proximal dendritic trees of some neurons in striate cortex. Virtually all PV-IR and heavily labeled Cal-IR cells were nonpyramidal as far as could be judged from the shape and orientation of the somata and the arrangement of the proximal dendrites. No spines were observed on stained dendrites, indicating that smooth stellate cells were preferentially labeled by antibodies to PV and Cal. Differences in the somatic size and shape, and in the morphology of the proximal dendritic trees of stained cells indicated that each group was morphologically heterogeneous. In general, PV-IR cells had robust dendrites that originated from the soma as thick trunks and tapered gradually and became varicose. In contrast, Cal-IR cells had fine, smooth dendrites that originated abruptly from the soma and maintained a spindly, nontapering morphology for long distances from the soma. Figs. 9-14 contrast the morphology of cells stained with each

Fig. 5. Laminar distributions of GABA-IR, PV-IR, or Cd-IR neurons, determined from striate cortex immunostained for GABA (A), GABA and PV (B), and GABA and Cal (C). A Average number of GABA-IR cells in 50 ~ 5 0 0pm bins as a function of cortical depth, determined from eight, nonoverlapping columns. Notice the peaks in the GABA depth profile corresponding to layers 2,4A, and 4C. B: Depth profiles of cells that were GABA-IR only (open bars), GABA-IR and PV-IR (shaded bars), or PV-IR only (solid bars). C: Depth profiles of cells that were GABA-IR only (open bars), G B A - I R and Cal-IR (shaded bars), or Cal-IR only (solid bars). Number of cells in each bin in B and C represent the average cell count for each category determined from three, nonoverlapping columns. Corresponding cortical layers are shown on the right in all three graphs.

J.F.M. VAN BREDERODE ET AL.

10

1 40

2-3A

38-4A

30

30

20

20

10

10

0

0

P

E 30 R c 20 E N 10 T

0

4c

48 20

5-6

30

P E

p1

R

c

10

E

6

10

15

20

GABA CELL DIAMETER (pm) Fig. 6. Histograms of cell diameters of GABA-IR cells expressed as a function of laminar location. Cell diameters were determined from highly magnified computer-video-microscope images with the aid of a graphics tablet. The height of each bar represents the percentage of

cells that fell into each size category relative to the total number of stained cells found in a cortical layer. Notice the variations in shape, position of the peak, and width of the distributions between layers.

antiserum. The photomicrographs clearly depict the processes of the labeled cells only in one focal plane. Therefore, camera lucida tracings of some cells are also presented, illustrating the full extent of the labeled processes discernible in single sections. Some PV-IR cells in layers 2, 3A, 5, and 6 were remarkably well labeled (Figs. 9, lo), and in some cases the proximal dendrites of cells in these layers could be traced for several hundred micrometers in single sections cut perpendicular or tangential to the cortical surface (Fig. 9A,B). In layer 2-3A one cell type was consistently observed, and its dendrites appeared to be almost completely revealed (Figs. 9B, 10). It exhibited a round or multipolar soma that gave rise to four to six robust primary dendrites almost all from the apical half of the cell. These generally branched within a short distance of the soma, and the daughter branches adopted a strikingly vertical orientation. In tangential section (parallel to the cortical surface), these heavily labeled cells (Fig. 9C) appeared as round or multipolar somata with only short segments of proximal dendrites visible, confirming that their dendritic trees form narrow vertical cylinders, about 150-200 pm wide and several hundred micrometers long (Figs. 9C, 10).

Large PV-IR cells in layers 5 and 6 (Fig. 12) were multipolar and had long, straight dendrites extending either horizontally or obliquely in perpendicular sections (Fig. 12). These cells also displayed long branched processes in tangential sections (Fig. 9E), indicating that they were multipolar stellate cells (Fig. 9D,E). Throughout all layers, other cell types were apparently labeled with PV antibody, but their dendrites were either obscured by heavy neuropil labeling, or were insufficiently labeled to allow a clear description. The occurrence of single well-labeled examples of different dendritic configurations (e.g., compare Fig. 9A-E) and the variety in soma sizes across the layers suggest the presence of several morphological PV-IR cell types. Although the intensity of Cal-IR was usually less than that for PV-IR, significant portions of the dendritic trees of some Cal-IR cells were labeled, revealing several different cell morphologies (Figs. 10-12). In layers 2 through 4B, heavily labeled Cal-IR cells with round or multipolar somata typically gave rise to three to five spindly primary dendrites that branched at various angles and distances from the somata (Figs. 10, 11A-C). Higher order dendritic branches were generally disposed in a radial fashion around

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

11

Fig. 7. Photomicrographs of adjacent tangential sections through the upper cortex showing a comparison between the pattern of labeling for PV-IR (A) and cytochrome oxidase (C), and between adjacent sections stained for Gal-IR (B)and cytochrome oxidase (D). The same fields of view are shown in A and C and in B and D; asterisks mark identical blood vessels. C shows layer 3 cytochrome oxidase dots in the lower left corner and a band of layer 4A honeycomb cytochrome oxidase staining in the upper right corner (arrowhead). Notice the correspon-

dence between the cytochrome oxidase pattern in layer 4A and the dark band of reticular PV-IR neuropil labeling in that layer (arrowhead in A) and the lack of correlation between the distribution of PV-IR cells and cytochrome oxidase dots in layer 3. In D cytochrome oxidase dots are visible in a section through layer 2-3. No clear correlation is evident between their distribution and that of Cal-IR cells or neuropil in the adjacent section shown in B. Scale bar = 500 pm.

the soma. While some cells displayed a vertical bias, they were clearly different from the PV-IR cells in the same layers (Fig. lo), in that they had smaller somata, much thinner dendrites, and widely spreading processes in both perpendicular (Figs. 10, 11B) and tangential sections (Fig. 11A,C).

In layer 5 some heavily labeled Cal-IR cells in both perpendicular (Fig. 13) and tangential (Figs. 11E, 14) section, had predominantly horizontal orientations suggesting a flattened horizontal form. These cells typically gave rise to three thick proximal dendrites, which branched approximately 10 to 20 km from the soma to form short

12

J.F.M. VAN BREDERODE ET AL. PV/GABA cells as Cal/GABA cells; 3) about 15% of the GABA-IR cells do not contain either one of these Ca2+binding proteins; and 4) when the distribution of these Ca2+-bindingproteins is compared to that of cytochrome oxidase, there is no obvious correlation for either PV-IR or Cal-IR cell bodies, but there is for neuropil staining. The neuropil staining for Cal-IR is highest where cytochrome oxidase staining is lowest, in layers 2, 4B/4Ca, and 5, while PV-IR in the neuropil correlates with both high cytochrome oxidase activity and geniculocortical input.

Distribution of PV-and Cd-IR neurons and co-localizationwithGABA-IR In this study PV-IR neurons were found throughout all layers except 1, while Cal-IR neurons are mainly in layers 2-3 and 5-6. This distribution is in good agreement with earlier observations from cat (Stichel et al., '87; Demeulemeester et al., '881, monkey (Celio et al., '86; Hendry et al., '89; van Brederode et al., '891, and human cerebral cortex (Arai et al., '87; Ichimiya et al., '88; Morrison et al., '88: Bluemcke et al., '89). All of these studies, using differentd antibodies and several species, find remarkably similar cell types and laminar distributions. Even the relative numbers are similar, for in rat somatosensory cortex (Celio and Heizmann, '81; Celio, '86) and in monkey striate cortex (this report) 70-75% of the GABA-IR neurons contain PV, while in cat visual cortex (Demeulemeester et al., '88) and monkey striate cortex (this report) 15-20% contain Cal. Both in cat (Demeulemeester et al., '89) and macaca (Hendry et al., '89; this report), the two CaZ+-binding proteins are in nonoverlapping populations. Minor species differences do exist, however; in adult cat visual cortex both PV (Stichel et al., '87) and Cal (Stichel et al., '87; Demeulemeester et al., '88) label a few small pyramidal neurons, while rat cortex also contains some Cal-IR pyramids (Feldman and Christakos, '83).Heavily labeled pyramids were Fig. 8. Comparison of adjacent perpendicular sections immunostained for PV-IR (A) or reacted for cytochrome oxidase activity (B). not seen in our monkey material, although some of the The same field of view is shown in both sections; arrows indicate tri- lightly labeled Cal-IR neurons from which processes could angles of identical blood vessels. The two prominent bands of PV-IR not be traced could be pyramidal or spiny stellate neurons. neuropil labeling in layers 4A and 4C correspond to the dark bands of This possibility was also mentioned by Celio et al. ('86)for cytochrome oxidase staining in these layers. The staining patterns in squirrel monkey visual cortex, and recently Morrison et al. layer 4A are discontinuous in both A and B. A periodicity in density of PV-IR neuropil labeling of lower layer 3 can be vaguely discerned ('88)have described lightly labeled Cal-IR pyramidal cells in (arrowheads),which corresponds to the location of the base of the cyto- human layers 3 and 5. None of the monkey PV-IR cells resembled pyramidal cells, in agreement with previous chrome oxidase dots in layer 3 (arrowheads in B). Scale bar = 500 Fm. reports in rat (Celio, '86), monkey (Hendry et al., '891, and human (Arai et al., '87; Bluemcke et al., '89). This report is the first quantitative analysis of CaL+tufts that were best seen in tangential sections (Fig. 11E). Other horizontal Cal-IR cells in layer 5 gave rise to long binding proteins and GABA neurons that examines individprocesses that appeared to branch only infrequently (Fig. ual laminae in primary visual cortex. Our finding that 14). In layer 6 Cal-IR cells were smaller, less heavily virtually all PV-IR and heavily labeled Cal-IR neurons are labeled, and displayed round or multipolar somata with also GABA-IR confirms and extends previous work in cat extremely fine, highly branched, and radially disposed (Demeulemeester et al., '88), monkey (Hendry et al., '89), dendrites that were best seen in tangential sections (Figs. and rat (Celio, '86) cortex. We also find that nearly 15% of IID, 14). The diameters of these small circular dendritic the GABA-IR neurons do not contain either Ca2+-binding fields was less than 100 pm, although the finest terminal protein, as measured directly in double-labeled sections or dendrites may not have been stained. by adding up each marker in single-labeled serial sections. Almost all of these cells lie outside of layer 4, and layer 1in particular contained GABA-IR neurons but almost no CalDISCUSSION or PV-IR neurons. This finding indicates that some cells The results of this immunocytochemical study have may have levels of these Caz+-bindingproteins too low to shown: 1) PV- and Cal-IR neurons in monkey striate cortex detect, or that some GABA cells completely lack these are two separate populations, which differ with respect to proteins. Most GABAergic neurons containing the peptides laminar distribution, size of immunostained cell bodies, and cholecystokinin, somatostatin, and neuropeptide Y do not dendritic morphology; 2) almost all PV-IR and Cal-IR cells label for Cal in cat visual cortex (Demeulemeester et a1 , display GABA-IR, although there are six times as many '881, suggesting that many GABA neurons containing

PARVALBTJMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

Fig. 9. Photomicrographs of perpendicular (A, B) and tangential (C-E) sections of striate cortex showing different morphological types of PV-IR cells. A: A large cell with a predominantly horizontal orientation is surrounded by smaller stellate cells close to the layer 1-2 border. B: Large PV-IR cells with vertical, varicose processes (arrows) are common in layers 2 and 3. C: In perpendicular sections through these layers the cells appear as round somas with few lateral processes,

13

emphasizing the vertical nature of their dendritic fields. D: Some cells in layer 5 display a fusiform soma with thick primary dendrites that extend laterally, giving the cell a horizontal bipolar appearance. E: Tangential section through the infragranular layers reveals both small and large (arrowheads) PV-IR cells with long, robust dendrites embedded in a dense meshwork of PV-IR labeled neuropil. Scale bars in A, C, and D = 50 pm; in B = 25 pm; in E = 100 pm.

14

J.F.M. VAN BREDERODE ET AI,.

,

> [r

I

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX peptides may lack Cal or PV. Given the rather different morphological types and laminar distributions of peptidecontaining neurons in rat, monkey, and cat visual cortex (discussed in Hendry et al., '84; Somogyi et al., '84b; Jones and Hendry, '86; Demeulemeester et al., '881, this hypothesis still has to be tested directly in primates. A number of studies in cerebral cortex have shown that the majority of GABAergic neurons contain PV-IR (Celio, '86; Hendry et al., '89), but in other brain regions this proportion is less predictable. For example, in the hippocampus less than 5% of the GABA-IR neurons in stratum radiatum/lacunosum show PV-IR, but 50% of the GABA-IR neurons in stratum pyramidale co-localize with PV-IR (Kosaka et al., '87). All GABA-IR cells in medial septum, but none in lateral septum, are PV-IR (Freund, '89). Similarly, we find that the relative proportion of PV-IR GABA cells varies considerably from layer to layer in the striate cortex, indicating that PV-IR is found only in specific subclasses of GABA neurons. Whereas all PV-IR cells are GABAergic, neuropil labeling is more complicated. For instance, PV is located in GABAnegative retinal axon terminals as well as GABA-IR dendrites and axons in the cat dLGN (Stichel et al., '88). GABA-negative/PV-IR axons occur in the white matter under striate but not prestriate visual cortex (Celio, '86; Bluemcke et al., personal communication; this report), a distribution that fits the cortical labeling pattern seen after dLGN injections (Hendrickson et al., '78). At the EM level, PV-IR synaptic terminals in layer 4 form symmetric contacts (presumably GABA-IR) as well as large asymmetric contacts (Bluemcke et al., '89, 'go), which would fit EM descriptions of dLGN synaptic terminals (reviewed by LeVay, '86). However, the dLGN in the same monkeys described in our studies shows PV-IR in small presumed GABAergicinterneurons (Hendrickson, personal communication),not projection neurons. These findings suggest that some monkey dLGN projection neurons contain PV-IR near the terminal portions of their axons, but not at detectible levels in their cell bodies. This interpretation is supported by the finding that Cal and PV heavily label axons terminating in red nucleus, but not the deep cerebellar nuclear projection neurons that give rise to these axons; however, if colchicine is applied to deep cerebellar nuclei (C. Andreessen and M.R. Celio, personal communication) the cell bodies as well as the axons become labeled. Therefore PV could be synthesized and axonally transported by long-axon projection systems that are not known to be GABA-IR, such as the dLGN projection to the cortex, as well as being a common component throughout many GABA cells. Cell labeling differences may also indicate that some neurons contain Ca2+;binding proteins mainly in

Fig. 10. Camera lucida drawings of well-labeled PV-IR (PARV) and Cal-IR (CALI cells from perpendicular sections of the upper layers of striate cortex. PV-IR cells have large, thick, mainly vertical dendrites; some of these cells resemble chandelier cells (e.g., see Fig. 3, Freund et al., '83).Cal-IR cells, on the other hand, have smaller somata from which fine dendrites radiate in all directions; these cells resemble the smooth stellate cells depicted in Fig. 33 of Lund ('73). In this and the following figures depicting camera lucida drawings, the cells are placed in their relative positions in the layers, but their density is not accurately depicted. Dense labeling of the neuropil prevented the accurate tracing of processes of PV-IR cells in lower layer 3 and in layer 4.Scale bar = 100 km.

15

dendrites and/or cell bodies, adjacent to the majority of inputs onto these cells (Stichel et al., '87). Heavily labeled Cal-IR cell bodies in cerebral cortex invariably are GABA-IR, but the situation is much less clear in other brain regions. Cal is quite widely distributed throughout the rat brain in a variety of neuronal types including dorsal root ganglia, cochlear hair cells, hypothalamic nuclei, hippocampal pyramidal cells, and Purkinje neurons (reviewed by Feldman and Christakos, '83; Enderlin et al., '87). Some of the cells containing Cal are known to be GABAergic, such as Purkinje cells, but many are not. Cal-IR staining intensity also varies widely among neurons of the same region. We find heavily stained neurons in layers 2-3 and 5-6, and many lightly stained cells in layer 4; the latter generally are not likely to be GABA-IR based on their number and laminar distribution (also see Celio et al., '86; Morrison et al., '88). Unlike cerebral cortex, in subcortical nuclei there are a number of neurons that show co-localization for PV and Cal. For instance, in cat dLGN (Demeulemeester et al., '89) 50% of the cells contain both, and the proportion of neurons that co-localizes PV-, Cal-, and GABA-IR varies in each of several zebra finch visual nuclei (Braun et al., '88).

CelltypeslabeledbyParvandCal On the basis of cell size, shape, and dendritic morphology it is clear that PV and Cal label different subpopulations of GABAergic stellate cells in layers 2-3 and 5-6, but without clear indications of the terminal axon arbors we hesitate to assign cell types to PV- and Cal-IR neurons. However, comparisons of soma size, laminhr distribution, and dendritic configuration of labeled cells can be made with cell types identified in previo-us studies with different techniques. The variety in soma size and dendritic configurations of PV-IR cells across the layers of monkey striate cortex point to the existence of many cell types within this population. Given the combination of large soma size, GABA content, and laminar distribution, certainly some classes of basket cells are represented by our PV-IR neurons (Jones, '75; Jones and Hendry, '84; Somogyi and Soltesz, '86; Lund et al., '88). The dendritic patterns of the PV-IR cells in the infragranular layers (e.g., Fig. 12) resemble Golgi impregnated cells of the basket class (types 6-1, 6-2, 6-4, 6-6) described by Lund et al. ('881, and the presence of PV-IR puncta surrounding unlabeled cell bodies (this study; see also Hendry et al., '89) also supports the suggestion that PV labels a population of basket cells. In the supragranular layers many PV-IR cells are vertically oriented, with long processes ascending through layers 3 and 2, and often into layer 1. These cells are thus ideally situated to receive a major input from the prestriate cortex (Van Essen, '84) and from pulvinar (Ogren and Hendrickson, '76). In their dendritic morphology and location these cells strikingly resemble the layer 2 chandelier cell type described in Golgi studies by Freund et al. ('83; see their Fig. 31, and could therefore potentially use feedback information from prestriate cortex to gate the output of pyramidal cells. Although we do not find obvious axoaxonic chandelier-type PV-IR terminals similar to those seen in monkey somatosensory cortex (DeFelipe et al., '89a), Bluemcke et al. ('go), using the same PV antibody as the present study, find at the EM level many small PV-IR axon contacts onto pyramidal cell initial axon segments in macaque striate cortex. Chandelier cell axon terminals in

J.F.M. VAN BREDERODE ET AL.

16

Figure 11

17

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX

5 _.-.-

PARV

6

-.-.-

Fig. 12. Camera lucida tracings of well-labeled PV-IR (PARV) cells from perpendicular sections of the infragranular layers of striate cortex. As in the upper cortical layers, PV-IR cells have large somata and robust dendrites. Although the dendrites could not be traced confidently for long distances due to heavy neuropil labeling, it appears

that PV-IR cells in these layers include cell types with vertical, multipolar, and horizontal dendritic fields. These cells resemble some of the basket cell classes described by Lund et al. ('88) in their Figs. 8, 9, and 17-20. Scale bar = 100 km.

striate cortex may thus have smaller, less obvious axon terminals that are obscured by heavy overall PV-IR neuropi1 labeling. Alternatively, the terminals may contain lower levels of PV than in somatosensory cortex. Cal-IR cells in monkey striate cortex also appear to be morphologically heterogeneous. Although no cortical CalIR cells with a basket cell morphology have been found by Hendry et al. ('89), the dendritic morphology of some of the largest Cal-IR cells in infragranular layers also resembles Golgi descriptions of basket cells, in particular, class 5B-6:l (Lund et al., '88). If a population of relatively small basket cells is indeed Cal-IR, it is puzzling that the neuropil of layers 5-6 conspicuously lacks pericellular Cal-IR puncta. The only layers containing Cal-IR puncta are layers 1 and upper 2, which would not fit the axon distribution of most infragranular basket cells (Lund et al., '88).Other infragranular Cal-IR cells resemble the horizontal stellate cells (Lund, '73, '87).

On the basis of dendritic morphology the supragranular Cal-IR cells resemble several types, including double bouquet cells, when compared to Golgi descriptions (e.g., Fairen et al., '84; Somogyi et al., '84a). Most Cal-IR cells fall into the smooth stellate category, resembling some of those described and illustrated by Lund ('73; see her Fig. 33). Interestingly, comparisons of cells identified in our study and the work of DeFelipe et al. ('89a,b) indicate that the numbers of different cell types can vary quite dramatically from one region of cortex to another. For example, DeFelipe et al. ('89b) describe the occurrence of numerous Cal-IR double-bouquet cells that give rise to bundles of long, varicose processes in monkey sensorimotor cortex. Similarly, we have observed many bundled Cal-IR processes in monkey prestriate cortex, but find them only rarely in striate cortex. This suggests that the number of double bouquet cells varies throughout the cortex, that Cal levels differ in different regions of cortex, or that different Cal antisera do not recognize this cell type to the same extent. In an effort to delineate further the PV- and Cal-IR populations, we measured cell diameters and found that cell size can sometimes predict which protein will be present. When both cell populations are present in the same layer, as in layers 2-3 or 5-6, the smallest cells are Cal-IR, the largest are PV-IR, and medium size cells can be either. However, when mainly PV-IR cells are found, as in layer 4, small, medium, and large cells are all PV-IR. Previous studies of cerebral cortex GABA-IR neurons have shown that cell size can be an indicator of cell type (Hendry and Jones, '81; Houser et al., '83; Hendry et al., '871, predict the presence of cell surface markers (Mulligan et al., '89; Naegele and Barnstable, '89), or mark the relationship to

Fig. 11. Photomicrographs of tangential (A-E) and perpendicular (B) sections of striate cortex, showing the different morphological types of Cal-IR cells. A: Heavily labeled Cal-IR cells in layer 2 give rise to fine, sparsely branching and widely spreading dendrites (arrowheads) that are apparent in both tangential (A) and perpendicular (B) sections. C: Similar cells are seen at lower density in layer 3. D: Small multipolar cells with very fine dendrites (arrows) forming bushy, roughly circular dendritic fields in tangential sections are mainly seen in layer 6. A thicker horizontal axon-like process passes near a horizontal Cal-IR fiber. E: In layer 5 larger Cal-IR somas give rise to a small number of primary dendrites that branch to form tufts (arrowheads) at short distances from the soma. Scale bars in A-C, E = 50 km; in D = 25 km.

18

J.F.M. VAN BREDERODE ET AL. -.-.

CAL

5

-.-.-

6 Fig. 13. Camera lucida tracings of heavily labeled Cal-IR (CAL)cells from perpendicular sections of the infragranular layers of striate cortex. In these layers Cal-IR cells have fine dendrites that adopt either a vertical or horizontal orientation. The cell on the right hears some resemblance to the basket cell type 5B:6-1 described by Lund et al. ('88)in their Fig. 7. Scale bar = 100 pm.

subtypes of geniculocortical projections (Fitzpatrick et al., '87). Within the entire GABA cortical population a large and a small class has been shown to exist, resulting in a bimodal size histogram (Hendry and Jones, '81). This report finds that the population of GABA-IR neurons within each layer of monkey visual cortex differs in average cell size, as well as in the range of cell diameters, suggesting that different classes of GABA cells have unique relationships to particular lamina as well. We also observed that the bimodal or skewed size histograms of GABA cells in layers 2-3 and 5-6 could be accounted for by differences in the size distributions of the PV- and Cal-IR cell populations in these layers. We have previously described a subpopulation of GABA-IR neurons in layers 3, 4B, 5, and 6 of monkey striate cortex that are specifically labeled by the lectin Vicia uitlosa WA,Mulligan et al., '89; Mulligan and Hendrickson, '89). These WA-labeled cells can be distinguished from other GAI3A cells by their large size, which together with their laminar distribution and morphology after intracellular filling suggest that some are basket cells. Since nearly all of the large neurons in these layers are PV-IR, it is not surprising that our preliminary observations find that all WA-labeled neurons are PV-IR, but not all PV-IR are WA-labeled, and that no Cal-IR cells are WA-positive. Recently 90% of the WA-labeled neurons of rat cortex were found to be PV-IR, but no WA-labeled cells were somatostatin- or cholecystokinin-positive (Kosaka and Heizmann, '89).

Correlationwith cytochrome oxidase One of the more exciting recent developments in primate striate cortical neuroanatomy is the finding that the oxidative mitochondrial enzyme cytochrome oxidase has a distribution that parallels both dLGN input and GABA terminal distribution and reflects functional cortical modules (re-

viewed by Hendrickson, '85; Fitzpatrick et al., '87; WongRiley, '89). Because 86%of the GABA-IR neurons are either PV- or Cal-IR, and because others have related PV-IR with high neuronal metabolic activity, we expected to find some correlation between these Ca2+-bindingproteins and cytochrome oxidase, but this was not immediately obvious. Earlier Celio et al. ('86) reported that in the New World squirrel monkey striate cortex Cal-IR neuropil labeling and the number of Cal-IR cells were greater outside of cytochrome oxidase-rich regions. We found some evidence that neuropil staining for Cal-IR is heavy in the cytochrome oxidase-poor layers 4B and 5 in the Old World macaque, but there was no difference in numbers of Cal-IR neurons. In fact, a comparison of slides from both species (Celio, personal communication) showed that many more heavy and light Cal-IR neurons stained in squirrel monkey compared to macaque, and that the overall neuropil labeling also was much heavier in squirrel monkey. Thus there could be a real species difference between New and Old World primates, with both having a band of heavily labeled Cal-IR neurons in layer 2, but New World monkeys having

Fig. 14. Camera lucida drawings of Cal-IR (CAL) cells in tangential sections of the infragranular layers. A marked difference in the morphological types of cells labeled with Cal is evident in this plane of section. Layer 5 Cal-IR cells have larger somas and widely spreading horizontal dendrites that branch infrequently. In contrast, layer 6 Cal-IR cells have smaller somas and very fine dendrites that branch repeatedly to form a dense, bushy radial field. It is likely that the dendritic field is larger than that depicted here, since the fine processes were almost at the limits of resolution of the light microscope and the labeling appeared to fade beyond about 100 +m from the somas. Photomicrographs of these cell types are shown in Fig. 11D,E. Scale bar = 100 pm.

PAFWALBUMIN AND CALBINDIN-D28k IN STRIATE CORTEX 19

_____---_____--- - -_--_

_ _ _ _ _ _ _--_ _ -

-_---__ --_--_-_-_

-_--_ 2 a,

?a

E

20

J.F.M. VAN BREDERODE ET AL.

a more prominent correspondence between Cal-IR and the cytochrome oxidase interdot regions. Despite many similarities in cortical organization, New and Old World monkeys differ in ocular dominance column occurrence (Hendrickson and Wilson, '79; Hendrickson and Tigges, '85) and in serotonin input patterns (Morrison and Foote, '86). PV-IR neuropil showed a clear laminar preference for cytochrome-rich layers 4A and 4C and a somewhat less clear preference for the cytochrome-rich dots of layer 3. A close spatial correlation of these metabolic indicators is seen in layer 4A where the dLGN input has a honeycomb distribution (Hendrickson et al., '78; Blasdel and Lund, '82). Cytochrome oxidase, GAD, and GABA (Hendrickson et al., '81; Fitzpatrick et al., '87) show the same pattern in this layer, and we have found that PV-IR also has a reticulated honeycomb pattern in 4A. It is not possible to determine from our material what the source of this labeling is, but the large number of PV-IR stellate neurons in layers 4A and 4C should make a heavy contribution to the neuropil. The presence of PV-IR axons in the white matter only under striate cortex suggests that some dLGN projection axons are labeled as well. This is supported by our finding that enucleation reduces the intensity of neuropi1 staining for both cytochrome oxidase and PV-IR in layers 4A and 4C of monkey striate cortex (Hendrickson, unpublished observation). EM immunocytochemistry demonstrates asymmetric PV-IR synaptic terminals in layer 4 (Bluemcke et al., 'go), strongly supporting a PV-IR dLGN input; however, many symmetric terminals are also PV-IR, compatible with the large number of (presumably)intrinsic PV-IR/GABA-IR neurons that we find. The outer portion of layer 1 and the top of layer 2 are rich in both GAD-IR terminals and GABAergic cell bodies (Hendrickson et al., '81; Fitzpatrick et al., '87; this report), but these layers are sparse for both PV and cytochrome oxidase. Instead, both layers are rich in Cal-IR cells and terminals, which supports the earlier suggestion (Heizmann, '84; Celio et al., '86) that Cal is not found in metabolically active GABA neurons.

neurons. The ionic mechanism of electrogenesis of fastspiking cells is unknown, however, and it remains to be established whether cortical Cal- and PV-IR neurons differ in their firing properties. A reduction in Cal has been reported in kindling models of epileptogenesis and in epileptic mouse (Baimbridge and Miller, '84; Anderson et al., '87). In contrast, the number of GABAergic PV-IR somata in kindled rat hippocampus shows an increase, while GABA neurons that do not contain PV-IR are reduced by 50% (Kamphuis et al., '89). These results indicate that Cal and PV are found in separate populations of neurons that respond differently to seizure activity. Our demonstration that PV-IR and Cal-IR are found in two, nonoverlapping subpopulations of GABAergic neurons in monkey striate cortex also suggests that the different cell types found in each subpopulation have different mechanisms for maintaining intracellular Ca" homeostasis. A recent intriguing finding is that the NMDA-receptor mediated Ca2+influx is larger than at other excitatory amino acid receptors (Garthwaite and Garthwaite, '871, possibly requiring a larger intracellular Ca2+-buffering capacity. It remains to be seen whether a relationship exists between NMDA-receptormediated processes such as learning, memory, and neuronal plasticity and the presence of PV or Cal in certain neuronal populations.

ACKNOWLEDGMENTS This work was supported by NIH grants EY 01208, EY 04536, EY 07031, and RR 00166 and thevision Core Grant. We thank Andra Erickson, Paul Butler, Janet Stolt, and Kim Allen for expert technical assistance, and Kim Graybeal for help in preparing this manuscript. Antisera against parvalbumin and calbindin-D 28k were generous gifts of Dr. M.R. Celio. We would like to thank Dr. Bluemcke for sharing an early manuscript draft with us (Bluemcke et al., '90) and Dr. M.R. Celio and Dr. K. Braun for helpful comments on an earlier version of our manuscript.

Functionalhplications Although both PV and Cal can bind intracellular Ca2' with high affinity, it is not clear what role they play in Ca"+-dependentbiological functions. It is well known that a rise in intracellular Ca2+concentration can lead to cell injury and cell death, and that the inability to buffer a large Ca2+influx has been implicated in excitatory amino acid toxicity and could be involved in the pathology of degenerative disorders of the nervous system such as Alzheimer's disease (for review see Heizmann, '88). Measurements of intracellular Ca2+concentration in experimentally induced glutamate toxicity in cultured CA1 pyramidal cells have indicated that Cal protects against an increase in Ca2+ (Baimbridge and Kao, '88). In guinea pig cerebral cortex slices, intracellular filling has shown that regular spiking or bursting neurons are pyramidal cells. Fast-spiking neurons are nonpyramidal radial or bitufted stellate cells (McCormick et al., '85), and indirect evidence suggests that many of these cells are both GABA- and PV-IR, but not Cal-IR. Fast-spiking hippocampal interneurons have been found to be PV-IR (Kawaguchi et al., '87), and it has been suggested that PV may act as an intracellular buffer against elevation of intracellular Ca" during fast repetitive firing, thereby blocking the spike frequency adaptation due to the Ca2+activated potassium conductance found in regular spiking

IJTExAmcm Anderson, T.B., J.J. Miller, and K.G. Baimbridge (1987) Specific regional changes in the concentration of parvalbumin immunoreactivity in the Neurosci. Abstr. 13:944. epileptic ( E l ) mouse. SOC. Arai, H., P.C. Emson, C.Q. Mountjoy, L.H. Carassco, and C.W. Heizmann (1987) Loss of parvalhumin-immunoreactive neurons from cortex in Alzheimer-type dementia. Brain Res. 418:164-169. Baimbridge, K.G., and J.J. Miller (1984) Hippocampal calcium-binding protein during commissural kindlinginduced epileptogenesis: Progressive decline and effects of anticonvulsants. Brain Res. 324335-90. Baimbridge, K.G.,and J. Kao (1988) Calbindin D28k protects against glutamate induced neurotoxicity in rat CAl pyramidal neuron cultures. SOC. Neurosci. Abstr. 14:1264. Baimbridge, K.G., J.J. Miller, and C.O. Parkes (1982) Calcium-binding protein distribution in the rat brain. Brain Res. 239:519-525. Blasdel, G.G.,and J.S. Lund (1982) Termination of afferent axons in macaque striate cortex. J. Neurosci. 3:1389-1413. Bluemcke, I., P.R. Hof, J.H. Morrison, and M.R. Celio (1989) The calciumbinding protein pawalbumin in the striate cortex of macaque monkeys and humans. SOC.Neurosci. Abstr. 15:1398. Bluemcke, I., P.R. Hof, J.H. Morrison, and M.R. Celio (1990) Comparison nn the distribution of pawalbumin in the visual cortex of Old World monkeys and humans. J. Comp. Neurol., in press. Bowker, R.M., K.N. Westlund, M.C. Sullivan, and J.D. Coulter (1982) A combined retrograde transport and immunocytochemical staining method for demonstrating the origins of serotonergic projections. J. Histochem. Cytochem. 8305-810.

PARVALBUMIN AND CALBINDIN-D 28k IN STRIATE CORTEX Braun, K., H. Scheich, M. Schachner, and C.W. Heizmann (1985) Distribution of parvalbumin cytochrome oxidase activity and (14C),-deoxyglucose uptake in the brain of the zebra finch. 11. Visual system. Cell Tissue Res. 26Ot117-127. Braun, K., H. Scheich, W. Zuschratter, C.W. Heizmann, C. Matute, and P. Streit (1988) Postnatal development of parvalbumin-, calbindin-, and adult GABA-immunoreactivity in two visual nuclei of zebra finches. Brain Res. 475205-217. Celio, M.R. (1986) Parvalbumin in most gamma-aminobutyric acid containing neurons of the rat cerebral cortex. Science 231t995-997. Celio, M.R., and C.W. Heizmann (1981) Calcium-binding protein parvalbumin as a neuronal marker. Nature 293t300-302. Celio, M.R., L. Scharer, J.H. Morrison, A.W. Norman, and F.E. Bloom (1986) Calbindin immunoreactivity alternates with cytochrome c-oxidase rich zones in some layers of the primate visual cortex. Nature 323:715-717. Celio, M.R., W. Baier, L. Scharer, P.A. de Viragh, and Ch. Gerday (1988) Monoclonal antibodies directed against the calcium binding protein parvalbumin. Cell Calcium 9:81-86. Curcio, C.A., and K.R. Sloan (1986) Computer-assisted morphometry using video-mixed microscopic images and computer graphics. Anat. Rec. 214:329-337. DeFelipe, J., S.H.C. Hendry, and E.G. Jones (1989a) Visualization of chandelier cell axons by parvalbumin immunoreactivity in monkey cerebral cortex. Proc. Natl. Acad. Sci. USA 86:2093-2097. DeFelipe, J., S.H.C. Hendry, and E.G. Jones (198913) Synapses of double bouquet cells in monkey cerebral cortex visualized by calbindin immunoreactivity. Brain Res. 503:49-54. Demeulemeester, H., F. Vandesande, G.A. Orban, C. Brandon, and J.J. Vanderhaeghen (1988) Heterogeneity of GABAergic cells in cat visual cortex. J. Neurosci. 8:988-1000. Demeulemeester, H., F. Vandesande, G.A. Orban, C.W. Heizmann, and R. Pochet (1989) Calbindin D-28k and parvalbumin immunoreactivity is confined to two separate neuronal subpopulations in the cat visual cortex, whereas partial coexistence is shown in the dorsal lateral geniculate nucleus. Neurosci. Lett. 99t6-11. Enderlin, S.,A.W. Norman, and M.R. Celio (1987) Ontogeny of the calciumbinding protein calbindin D-28K in the rat nervous system. Anat. Embryol. 177t15-28. Fairen, A,, J. DeFelipe, and J. Regidor (1984) Nonpyramidal neurons: general account. In A. Peters and E.G. Jones (eds.): Cerebral Cortex, Vol. 1. New York: Plenum Press, pp. 29-245. Feldman, S.C., and S.V. Christakos (1983) Vitamin D-dependent calciumbinding protein in rat brain: biochemical and immunocytochemical characterization. Endocrinology 112290-302. Fitzpatrick, D., J.S. Lund, D.E. Schmechel, and A.C. Towles (1987) Distribution of GABAergic neurons and axon terminals in the macaque striate cortex. J. Comp. Neurol. 264t73-91. Freund, T.F. (1989) GABAergic septohippocampal neurons contain parvalbumin. Brain Res. 478:375-381. Freund, T.F., K.A.C. Martin, A.D. Smith, and P. Somogyi (1983) Glutamate decarboxylase-immunoreactive terminals of Golgi-impregnated axoaxonic cells and of presumed basket cells in synaptic contact with pyramidal neurons of the cat's visual cortex. J. Comp. Neurol. 221:263-278. Garthwaite, G., and J. Garthwaite (1987) Receptor-linked ion channels mediate N-methyl-D-aspartate neurotoxicity in rat cerebellar slices. Neurosci. Lett. 83:241-246. Heizmann, C.W. (1984) Parvalbumin, an intracellular calcium-binding protein; distribution, properties and possible roles in mammalian cells. Experientia 40:910-921. Heizmann, C.W. (1988) Calcium binding proteins of the EF-type. J. Cardiovasc. Pharmacol. 1230-37. Hendrickson, A.E. (1985) Dots, stripes and columns in monkey visual cortex. Trends Neurosci. 8t406410. Hendrickson, A.E., and J.R. Wilson (1979) A difference in (14C)deoxyglucose autoradiography patterns in striate cortex between Macma and Saimiri monkeys following monocular stimulation. Brain Res. 170:353-358. Hendrickson, A.E., and M. Tigges (1985) Enucleation demonstrates ocular dominance columns in Old World, but not in New World monkey striate cortex. Brain Res. 333:340-344. Hendrickson, A.E., J.R. Wilson, and M.P. Ogren (1978) The neuroanatomical organization of pathways between the dorsal lateral geniculate nucleus and visual cortex in Old and New World primates. J. Comp. Neurol. 182123-136.

21

Hendrickson, A.E., S.P. Hunt, and J.Y. Wu (1981) Immunocytochemical localization of glutamic acid decarboxylase in monkey striate cortex. Nature 292r605-607. Hendry, S.H.C., and E.G. Jones (1981) Size and distribution of intrinsic neurons incorporating tritiated GABA in monkey sensory-motor cortex. J. Neurosci. lr390-408. Hendry, S.H.C., E.G. Jones, J. DeFelipe, D. Schmechel, C. Brandon, and P.C. Emson (1984) Neuropeptide-containing neurons of the cerebral cortex are also GABAergic. Proc. Natl. Acad. Sci. USA 81t6526-6530. Hendry, S.H.C., H.D. Schwark, E.G. Jones, and J. Yan (1987) Numbers and proportions of GABA-immunoreactive neurons in different areas of monkey cerebral cortex. J. Neurosci. 7:1503-1519. Hendry, S.H.C., E.G. Jones, P.C. Emson, D.E.M. Lawson, C.W. Heizmann, and P. Streit (1989) Two classes of cortical GABA neurons defined by differential calcium binding protein immunoreactivities. Exp. Brain Res. 76t467-472. Hauser, C.R., S.H.C. Hendry, E.G. Jones, and J.E. Vaughn (1983) Morphological diversity of immunocytochemically identified GABA neurons in the monkey sensory-motor cortex. J. Neurocytol. 12.617-628. Ichimiya, Y., P.C. Emson, C.O. Mountjoy, D.E.M. Lawson, and C.W. Heizmann (1988) Loss of calbindin-28K immunoreactive neurons from the cortex in Alzheimer-type dementia. Brain Res. 475:156-159. Jones, E.G. (1975) Varieties and distribution of nonpyramidal cells in the somatic sensory cortex of the squirrel monkey. J. Comp. Neurol. 16Ot205-268. Jones, E.G., and S.H.C. Hendry (1984) Basket cells. I n A. Peters and E.G. Jones (eds.): Cerebral Cortex, Vol. 1, New York: Plenum Press, pp. 309-336. Jones, E.G., and S.H.C. Hendry (1986) Co-localization of GABA and neuropeptides in neocortical neurons. Trends Neurosci. 9t71-76. Kamphuis, W., E. Huisman, W.J. Wadman, and F.H. Lopes da Silva (1989) Kindling induced changes in parvalbumin immunoreactivity in rat hippocampus and its relation to long-term decrease in GABA-immunoreactivity. Brain Res. 479r23-34. Kawaguchi, Y., H. Katsumaru, T. Kosaka, C.W. Heizmann, and K. Hama (1987) Fast spiking cells in rat hippocampus (CA1 region) contain the calcium binding protein parvalbumin. Brain Res. 41 6:369-374. Kosaka, T., and C.W. Heizmann (1989) Selective staining of a population of parvalbumin-containing neurons in the rat cerebral cortex by lectins with specific affinity for terminal N-acetyl-galactosamine. Brain Res. 483: 15tL163. Kosaka, T., H. Katsumaru, K. Hama, J.-Y. Wu, and C.W. Heizmann (1987) GABAergic neurons containing the Ca2+binding protein parvalbumin in the rat hippocampus and dentate gyrus.Brain Res. 419:119-130. LeVay, S. (1986) Synaptic organization of claustral and geniculate afferents to the visual cortex of the cat. J. Neurosci. 6:3564-3575. Livingstone, M.S., and D.H. Hubel (1982) Thalamic inputs to cytochrome oxidase-rich regions in monkey visual cortex. Proc. Natl. Acad. Sci. USA 79t609a6101. Lund, J.S. (1973) Organization of neurons in the visual cortex, area 17, of the monkey (Macaca mulatta).J. Comp. Neurol. 147t455-496. Lund, J.S. (1987) Local circuit neurons of macaque striate cortex: I. Neurons of laminae 4C and 5A. J. Comp. Neurol. 257:60-92. Lund, J.S., M.J. Hawken, and A.J. Parker (1988) Local circuit neurons of macaque monkey striate cortex: 11. Neurons of laminae 5B and 6. J. Comp. Neurol. 276:l-29. McCormick, D.A., B.W. Connors, J.W. Lighthall, and D.A. Prince (1985) Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. J. Neurophysiol. 54t782-806. Morrison, J.H., and S.L. Foote (1986) Noradrenergic and serotonergic innervation of cortical, thalamic, and tectal visual structures in Old and New World monkeys. J. Comp. Neurol. 243t117-138. Morrison, J.H., K. Cox, P.R. Hof, and M.R. Celio (1988) Neocortical parvalbumin-containing neurons are resistant to degeneration in AlzheNeurosci. Abstr. 14:1085. imer's disease. SOC. Mulligan, K.A., and A.E. Hendrickson (1989) The lectin W A labels a morphologically heterogeneous subpopulation of GABA neurons in the monkey striate cortex. SOC. Neurosci. Abstr. 151397. Mulligan, K.A., J.F.M. van Brederode, and A.E. Hendrickson (1989) The lectin Vicia ui2losa labels a distinct subset of GABAergic cells in macaque visual cortex. Vis. Neurosci. 263-72. Naegele, J.R., and C.J. Barnstable (1989)Molecular determinants of GABAergic local-circuit neurons in the visual cortex. Trends Neurosci. 12t28-34. Ogren, M.P., and A.E. Hendrickson (1976) Pathways between striate cortex

22 and subcortical regions in Macacu rnulattu and Sairniri sciureus: evidence for a reciprocal pulvinar connection. Exp. Neurol. 53:780-800. Somogyi, P., and I. Soltesz (1986) Immunogold demonstration of GABA in synaptic terminals of intracellularly recorded horseradish peroxidasefilled basket cells and clutch cells in the cat’s visual cortex. Neuroscience 19: 1051-1065. Somogyi, P., T.F. Freund, and Z.F. Kisvarday (1984a) Different types of 3H-Gaba accumulating interneurons in the visual cortex of the rat. Characterization by combined autoradiography and Golgi impregnations. Exp. Brain. Res. 54:45-56. Somogyi, P., A.J. Hodgson, A.D. Smith, M.G. Nunzi, A. Gorio, and J.-Y. Wu (1984b) Different populations of GABAergic neurons in t he visual cortex and hippocampus of cat contain somatostatin or cholecystokininimmunoreactive material. J. Neurosci. 42590-2603. Stichel, C.C., W. Singer, C.W. Heizmann, andA.W. Horman (1987) Immunohistochemical localization of calcium-binding proteins, parvalbumin and calbindin-D 28k, in the adult and developing visual cortex of cats: A light and electron microscopic study. J. Comp. Neurol. 262563-577.

J.F.M. VAN BREDERODE ET AL. Stichel, C.C., W. Singer, and C.W. Heizmann (1988) Light and electron microscopic immunocytochemical localization of parvalbumin in the dorsal lateral geniculate nucleus of the cat: Evidence for coexistence with GABA. J. Comp. Neurol. 268;29-37. van Brederode, J.F.M., A.E. Hendrickson, K.A. Mulligan, and M. Celio (1989) Development of calcium-binding proteins in GABA neurons of monkey visual cortex. SOC. Neurosci. Abstr. 15:1335. Van Essen, D.C. (1984) Functional organization of primate visual cortex. In A. Peters and E.G. Jones (eds.): Cerebral Cortex, Vol. 3, Visual Cortex. New York: Plenum Press, pp. 259-329. Wasterlain, G.G., W. Massarweh, A. Sollas, E. Rouk, and R.S. Sloviter (1988) Selective vulnerability of neurons to hypoxia-ischemia: Role of calcium Neurosci. Abstr. 14;1064. binding protein. SOC. Wong-Riley, M.T.T. (1979) Changes in the visual system of monocularly sutured or enucleated cats demonstrable with cytochrome oxidase histochemistry. Brain Res. 171:ll-28. Wong-Riley, M.T.T. (1989) Cytochrome oxidase: an endogenous metabolic marker for neuronal activity. Trends Neurosci. 1294-101.

Calcium-binding proteins as markers for subpopulations of GABAergic neurons in monkey striate cortex.

Recent studies have shown that the presence of immunoreactivity for parvalbumin (PV-IR) and calbindin-D 28k (Cal-IR) can be used as markers for certai...
6MB Sizes 0 Downloads 0 Views