THE JOURNAL OF COMPARATIVE NEUROLOGY 302~197-205(1990)
Parvalbumjn-ContainingGmAergic hterneuromin theRat Neostriatmn RONALD L. COWAN, CHARLES J. WILSON, PIERS C. EMSON, AND CLAUS w. HEIZMANN Department of Anatomy and Neurobiology, University of Tennessee College of Medicine, Memphis, Tennessee 38163 (R.L.C., C.J.W.); MRC Group, Institute of Animal Physiology and Genetics Research, Cambridge, CB2 4AT, United Kingdom (P.C.E.); Department of Pediatrics, Division of Clinical Chemistry, University of Zurich, CH-8032 Zurich, Switzerland (C.W.H.)
ABSTRACT Antibodies to the intracellular calcium binding protein parvalbumin were shown to label specifically a distinct group of neostriatal GABAergic neurons. These neurons corresponded to the intensely staining subclass of neostriatal GABAergic neurons that have previously been shown to be a class of aspiny interneurons in the neostriatum. The parvalbumin neurons were aspiny neurons with varicose dendrites distributed throughout the neostriatum in a pattern identical to the intensely stained GABA neurons, and both populations of neurons showed increased numbers in the lateral part of the neostriatum. Double labeling of single neurons with both the GABA and parvalbumin antisera showed that all parvalbumin neurons were positive for GABA, but some GABA labelled neurons were not immunoreactive for parvalbumin. These parvalbumin-negativeGABAergic neurons were morphologically similar to the spiny projection neurons, which are GABAergic but usually are not so heavily stained. The relationship of the GABA-containing parvalbumin neurons to the striatal mosaic organization was determined by using immunocytochemistry for another calcium binding protein, calbindin D28K, to label the matrix compartment of the striatum. The distribution of parvalbumin-positive neurons relative to the calbindin-positive matrix and calbindin-poor patches was determined by using pairs of adjacent sections stained with the calbindin and parvalbumin antisera. This analysis showed that the somata of the parvalbumin neurons were present in both patch and matrix compartments, and their axons and dendrites crossed the boundaries between compartments. A quantitative analysis of the number of neurons in each compartment revealed that the neurons showed no preferential distribution in either compartment, but instead were present according to the area occupied by that compartment. Approximately 10%of parvalbumin neurons were in the patch compartment, and in these same sections, the patch compartment occupied approximately 10% of the area of those sections. Staining with parvalbumin antibodies can therefore be used to identify a single class of GABAergic aspiny interneurons that is present in both patch and matrix compartments, and whose processes cross the borders between these compartments. Key words: basal ganglia, inhibition, calbindin, calcium binding protein, immunocytochemistry
In addition to the morphological diversity of neurons in the neostriatum as revealed in Golgi studies (e.g., Chang et al., '82), immunocytochemical studies have revealed a n enormous cytochemical diversity in the populations of gamma-amino butyric acid (GABA), enkephalin, substance P, somatostatin, and acetylcholine-containing neurons in the neostriatum (Aronin et al., '84; Beckstead and Kersey, '85; Bolam et al., '85; Graybiel and Chesselet, '84;Oertel and Mugnaini, '84; Oertel et al., '83; Panula et al., '81; Penny et al., '86; Ribak et al., '79). For the most part, the cell types identified in this way have mapped easily onto the 0 1990 WILEY-LISS. TNC
more traditional classes of cells according to their somatodendritic morphology. While the identification of cell types in the neostriatum on the basis of the presence or absence of substances related to neurotransmission seems to require little justification, there is no a priori reason to expect that other molecules with more general functions would be preferentially localized in neurons of particular types. Nonetheless, some such
Accepted August 27, 1990.
198
substances show staining patterns suggestive of a preferential localization in certain cell types. Among these are two calcium binding proteins, calbindin-D28k and parvalbumin. Striatal calbindin was initially discovered by using radioimmunoassay techniques (Baimbridge et al., '82), and was later localized to spiny neostriatal neurons (DiFiglia et al., '89) located specifically in the matrix (acetylcholinesterase-rich, mu-opiate receptor-poor) compartment of the neostriatum (Gerfen et al., '85). Calbindin has been linked to a variety of cellular activities, including a role in kindling induced epilepsy, (Miller and Baimbridge, '831, pathological processes in Huntington's disease (Seto-Ohshima et al., '881, and cellular metabolic activity levels (Braun et al., '85a,b; Celio et al., '86). Parvalbumin, on the other hand, is not localized in the medium spiny projection neurons, but is concentrated in a group of unidentified striatal neurons distributed throughout the neostriatum (Celio and Heizmann, '81;Gerfen et al., '85). These parvalbumin neurons are of similar size and distribution to striatal GABAergic interneurons (Gerfen et al., '851, suggesting that the two neuron populations may overlap or be identical. This is consistent with the localization of parvalbumin in other areas of the central nervous system. Parvalbumin has been demonstrated in GABAergic cells such as cerebellar Purkinje, basket, and stellate cells; hippocampal basket cells (Celio and Heizmann, '82; Kosaka et al., '87); and neocortical interneurons (Celio et al., '86; Celio, '86). In addition, parvalbumin is present in fast twitch muscle fibers, where it is thought to participate in postcontraction relaxation (Celio and Heizmann, '82; Pechere et al., '771, and in fast spiking hippocampal neurons (Kawaguchi et al., '871, where its calcium binding capacities have been postulated to play a role in conferring rapid firing properties on these neurons. In this report we show that parvalbumin is preferentially localized in a small population of GABAergic interneurons in the neostriatum that have been recognized previously on the basis of their high concentration of GABA and GABAsynthesizing enzymes, and that correspond to a classical type of neostriatal interneuron on the basis of their somatodendritic morphology. In addition, we show that this population of neurons is present in both the calbindin-poor patches and in the calbindin-rich matrix compartment of the neostriatum, and that their dendrites freely cross the boundary that separates these two compartments of the neostriatal neuropil.
R.L. COWAN ET AL.
methods were employed. In some animals the brains were sectioned on a rotary microtome at a thickness of 2-4 pm after embedding in polyethylene glycol (MW1000). In this procedure, blocks of tissue containing the neostriatum were dehydrated in alcohol, infiltrated in polyethylene glycol at 40"C, and allowed to cool in a desiccator at 4°C. Adjacent pairs of sections were cut on a metal knife at room temperature and collected into buffer. The polyethylene glycol embedment was removed from the sections by repeated washing in buffer. The immunocytochemical procedure was then applied to the free-floating semi-thin sections. In the other procedure, adjacent 40-100 pm sections were cut from unembedded tissue on a vibratome, collected in buffer, and processed free-floating. The problem of identification of neurons in both of the section pairs was simpler in the thinner sections, in which nearly all neurons present in one of the pairs could also be identified in the other. The thicker sections cut with a vibratome were more durable, however; and their thickness made it easier to verify landmarks in adjacent sections. As a result, most of the analysis was performed on the vibratome sectioned material, in which matching faces of the sections were identified and only neurons whose somata were sectioned so as to be present at the surface of both sections could be counted in the analysis. Comparison of the distribution of neurons in the neostriatal patches and matrix was likewise performed on adjacent sections arranged so that matching faces of vibratome sections could be identified and compared. Rabbit anti-GABA antiserum was purchased from Immunonuclear Corporation. Rabbit anti-chicken calbindin and sheep anti-rabbit parvalbumin polyclonal antisera were raised by P.C. Emson at Cambridge and goat anti-rat parvalbumin was raised by C.W. Heizmann in Zurich. Standard procedures were used for immunocytochemical staining and were the same with all three antisera. They were 1)treatment for 4 hours at room temperature in 4% normal serum (normal serum corresponded to the species of the secondary antibody) in 0.01 M potassium phosphate buffered saline (KPBS), pH 7.4, containing 0.5% Triton-X, 2) overnight incubation a t 4°C in primary antisera (dilutions of 1:1,000-1:2,000) in 4% normal serum in KPBS containing 0.5% Triton-X; 3) washing in KPBS, followed by 1hour room temperature incubation in biotinylated secondary antisera (dilution of 1:lOOor 1:200) in KPBS containing 0.5% Triton-X. Goat anti-rabbit secondary was used to bind the GABA and calbindin antibodies, and rabbit anti-goat secondary was used to bind the parvalbumin antibody; 4) washing in KPBS, followed by 30 minute incubation in EXPERWIENTALPROCEDURES avidin-biotin-peroxidase solution in KPBS with 0.5% TriMale adult (200-500 g) Long-Evans or Sprague-Dawley ton-X, and 5) Washing in KPBS, followed by staining with rats were used in all experiments. The brains were fixed by diaminobenzidine. intracardial perfusion of aldehydes after the animals were Negative controls were obtained by substituting primary deeply anesthetized with sodium pentobarbital (75 mgkg antisera with normal serum, and then processing the tissue i.p.1. The fixative consisted of either sequential 4%formal- identically to tissue containing primary antisera. Preabsorpdehyde in 0.1 M acetate buffer (pH 6.5) followed by 4% tion of the antisera to calbindin or parvalbumin with their formaldehyde and 0.1% glutaraldehyde in 0.05 M borate corresponding high pressure liquid chromatography-puribuffer (pH 9.5) according to Berod et al. ('81) or with a fied antigens M) was used as a control for nonspecific simpler fixative consisting of 4%formaldehyde and 0.1-3% binding of these antisera to the tissue sections. In all cases glutaraldehyde in 0.15 M sodium phosphate buffer (pH 7.4). the characteristic staining patterns of the antisera were No consistent differences were observed in labeling between absent in the control sections. tissue prepared with the different buffers. Sections were pretreated for 30 minutes with 1% sodium borohydride before further processing for immunocytochemistry (ElRESULTS dred et al., '83). ParValbuminneurons For double labeling of cells with GABA and parvalbumin, The parvalbumin antisera labeled medium-sized striatal individual neurons were identified in adjacent sections processed with the two antisera. Two different sectioning neurons having numerous dendritic processes (Fig. 1).
PARVALBUMIN NEURONS Staining was generally of equal intensity in the nucleus, cytoplasm, and dendrites, and the nucleolus appeared as a clear zone. The neurons had a broad size range, and were both polygonal and fusiform. Using the criterion of Chang et al. ('82), we classified the neurons into small, medium, or large on the basis of cross-sectional area measurements. A majority (83%)of the parvalbumin neurons were mediumsized (cross-sectional area 100-300 km'); 15%were small (cross-sectional area < 100 km2), and a small percentage (2%)were large, having cross-sectional areas greater than 300 km'. Dendrites were darkly stained, with very fine cell processes plainly visible. Dendritic spines, if present, would have been clearly visible. The dendrites were all aspiny throughout their length, and no somatic spines were observed. Most neurons had varicose dendritic processes (Figure 1, arrows), and many such processes were seen in isolation in the neuropil. A fine plexus of axons and axonal varicosities were present throughout the neostriatum.
Spatial distributionand somatic sizes To determine whether parvalbumin neurons could be a subset of the intensely staining GAJ3A-containingneurons in the neostriatum, we compared the spatial distributions and the somatic sizes of these two cell types in sections cut from the same brains. GABA antisera worked best on tissue fixed with the high glutaraldehyde fixatives. Under these conditions, many darkly labeled neurons were seen throughout the neostriatum. The GABA neurons were darkly stained in their nuclei and cytoplasm, while little dendritic staining was evident. Staining in the neuropil was much greater in the GABA sections than in the sections stained for parvalbumin. The distribution of the GABA neurons was similar to that of the
199
parvalbumin neurons. An example showing a plot of all intensely stained GABA neurons and all parvalbumin immunoreactive neurons in adjacent sections is shown in Figure 2. The similarity in their distribution as seen in that figure was consistent in all cases. The strong tendency for more parvalbumin-stained neurons to be located in the lateral half of the neostriatum was also shared by the intensely staining GABA-positive cells. Although not greatly dissimilar in numbers, there was a consistent difference between the cell populations, with GABA neurons being slightly more numerous (e.g. Fig. 2). To examine the size distribution of the neurons, we measured the cross-sectional area of 200 neurons of each type selected randomly from 16 sections from two animals. To avoid problems with penetration of reagents used in the immunocytochemical procedures, we analysed only darkly stained neurons that intersected a single surface of the section, and whose nucleoli were in focus at the cut surface. Sections used in the analysis were separated by several hundred micrometers, insuring that cells would not be counted more than once. The distance between prominent tissue landmarks measured in sections stained with the parvalbumin and GABA antisera were the same, indicating that there was no differential tissue shrinkage that could complicate comparison of cell sizes. No attempt was made to correct for the shrinkage common to the two groups of sections. The cross-sectional areas of the somata were measured by using an on-line video digitizer and image analysis program (Image, obtained from Wayne Rasband, NIH). A comparison of the cross-sectional areas of GABA neurons and parvalbumin neurons revealed that parvalbumin-containing cells fell within the size distribution of GABA-positive neurons. The somatic size distributions of
Fig. 1. Photomicrograph of a pawalbumin-containing striatal interneuron. Note the dendritic varicosities (arrows) and the absence of somatic or dendritic spines.
R.L. COWAN ET AL.
200
Fig. 2. Plots of adjacent sagittal neostriatal sections showing the similar distributions of pawalbumin-positive neurons (left) and intensely GABA-positive neurons (right).The intensely GABA-positive
both intensely GABA-positive and parvalbumin-positive neurons are shown in Figure 3. The cross-sectional areas of the parvalbumin-containingneurons ranged from 65 to 483 pm2.The distribution obtained for GABA neurons covered this same range, but also included a group of smaller neurons, ranging from 40 to 379 pm'. While both distributions had about 2% large neurons, the GABA group had 50% small neurons (cross-sectional area < l o 0 pm2, see above) compared to the 15%small neurons in the parvalbumin group. The presence of the smaller neurons in the GABA sample resulted in a difference in the average cross-sectional area measurements for each group. The mean cross-sectional area for the GABA neurons was 117 60 pm2 (n = 200), while that of the parvalbumin neurons was larger at 148 57 pm2 (n = 200). An unpaired t-test (df = 398) revealed that the cross-sectional areas of the two groups were significantly different (P < 0.01). This difference and the distributions shown in Figure 3 demonstrate that the parvalbumin cells are completely contained within the distribution of GABA cells, but cannot account for all GABA neurons.
*
*
ColocaziZationof GABA and parvalbumin Parvalbumin and GABA immunoreactivity were tested on single cells in adjacent sections. Alternate sections were stained with a different antiserum, and the sections were placed in register by using blood vessels as fiducial marks. Neurons whose somata were cut through at the faces of the sections were identified in both sections of a pair and examined for the presence of immunoreactivity to GABA and parvalbumin. The fact that there were more intensely GABA-positive neurons than parvalbumin-positive cells ruled out the possibility of an exclusive one-to-one correspondence of the two cell populations. The remaining possibility, that the parvalbumin neurons were a subset of GABApositive cells, was tested by examining the GABA reactivity of parvalbumin-containing neurons. All the parvalbumin neurons tested in this manner were also GABAergic. Exam-
neurons are slightly more numerous than the parvalbumin neurons, but they share a common distribution within the nucleus. AC, anterior commissure; IC, internal capsule; ACp, anterior commissure posterior.
j0 1
UY
5L
40
GABA (n=200) Q
PARVALBUMIN (n=200)
2 30 z
:20
cL-$
Q)
n
El 10
z
0
45 65 85 105 125 145 165 185 205 225 245 >245
Cross Sectional Area (pd)
Fig. 3. The distribution of the cross-sectional areas of neurons stained using antisera to GABA (dark shading) and to parvalbumin (light shading) neurons. The distribution of the GABA neurons contains a population of smaller cells, as well as a group of neurons corresponding to the distribution observed for parvalbumin-positive neurons. The size distribution of the GABA-positivebut parvalbuminnegative cell group corresponds to that of the spiny projection neurons.
ples of two pairs of parvalbumin-containingGABA neurons are shown in Figure 4A-D. As expected, there were some intensely GABA-positive neurons in the same sections that did not also contain parvalbumin (not shown).
Pawalbuminneuronsin the neostriatal mosaic Adjacent sections were used to examine the distribution of parvalbumin neurons in relation to the neostriatal mosaic, as indicated by the distribution of calbindin-labeled spiny neurons. One section from a pair of adjacent sections was processed to visualize calbindin and the other half was processed to visualize parvalbumin. The boundaries of the calbindin-positive patches were most clear in the caudal and
PARVALBUMIN NEURONS
201
Fig. 4. Photomicrographs of adjacent sections stained usingantihodies to GABA (left,A,C) and parvalbumin (right, B,D).The focal level of the microscope is set to the opposing surfaces of the sections to maximize the correspondence between the two sections. A and B are adjacent and C and D are adjacent. Arrows in all sections point to blood vessels used as landmarks for alignment of the sections. The parallel
diagonal lines in the sections labelled for GABA are artifacts of vihratome sectioningpresent at the surface of the section. These are not present in the parvalbumin sections because of the minimal neuropil staining in these sections. Parvalhumin neurons in B correspond to GABA-stained neurons in A, and parvalhumin-positive cells in D are the same as those stained for GABA in C.
ventral parts of the neostriatum near the globus pallidus, because the staining for calbindin is strongest in this part of the neostriatum. Therefore, the analysis of the mosaic organization of parvalbumin neurons was restricted to these regions, where nearly every spiny neuron in the matrix compartment is positive for calbindin, while calbindin staining in the neurons of the patches is weak, and sometimes undetectable. An example of the calbindin staining pattern as seen at low magnification is shown in Figure 5. Previous work has established that the patch boundaries seen in this way correspond to the mosaic organization observed with other methods, and that the calbindin-poor patches correspond to the striosomes (Gerfen et al., '85). Ten plots of the locations of parvalbumin neurons and patch-matrix boundaries were made from pairs of wellstained adjacent sections from three animals. The boundaries of the calbindin-poor patches were drawn with the aid of a drawing tube and a 4 x objective. Parvalbumin neurons from the corresponding adjacent section were then superimposed on the plot of the calbindin-stained section. The ratio of patch area to the combined patchimatrix area was
calculated to determine the percentage of area occupied by the patches. The ratio of parvalbumin neurons in patches to the total number of parvalbumin neurons in the combined patchimatrix area was calculated and compared to the percentage of area occupied by the patches. An example illustrating the appearance of calbindin and parvalbumin staining on adjacent sections is shown in Figure 6. Parvalbumin neurons and their axons and dendrites were present in both the calbindin-poor patches and in the matrix compartment. In addition, parvalbumin immunoreactive axons and dendrites frequently crossed the compartmental boundaries. Plots of the entire usable portion of pairs of sections placed in register as in Figure 6 suggested that the neurons were not preferentially associated with either compartment. Examples from these plots are shown in Figure 7A-C. Quantitative analysis of 10 sections from 3 different cases indicated that parvalbumin neurons are distributed in the patchimatrix compartments in proportion to the area occupied by the respective compartments. Calbindin-poor patches occupied 9.8% of the total striatal area in the sections, while patch parvalbumin
R.L. COWAN ET AL.
202
Fig. 5. Photomicrograph of calbindin staining in a parasagittal section of the rat neostriatum, as used for the study of the compartmental localization of pawalbumin-positive neurons. The more heavily stained regions compose the matrix, while the calbindin-poor areas are patches. GP, globus pallidus.
neurons composed 10.2% of the total number of parvalbumin neurons in the same regions.
DISCUSSION Identity of the parvalbuminneurons In recent years it has become increasingly evident that the neostriatum contains two different classes of GABAcontaining neurons. While the most common neuron, the spiny projection cell, is easily recognized by its characteristic and well-known morphological features, the identity and functional properties of GABAergic interneurons are much less well understood. In immunocytochemical studies, the GABAergic interneurons have been recognized by their intense staining. Somatic GABA immunoreactivity of spiny neurons is very low in the absence of special treatments such as colchicine pretreatment. The population of intensely GABA-positiveneurons observed under these conditions has been shown to include members of a class of interneurons recognized by their morphological features when impregnated with the Golgi method, their cytological features in electron micrographs, and their failure to be stained by retrograde axonal transport from the substantia
nigra and globus pallidus (Bolam et al., '83a,b; Chang et al., '82; Gerfen et al., '85). It is not known whether these cells represent a single neuronal population or are heterogeneous, and their position within the established circuitry of the neostriatum is unclear. The present results indicate that at least a large proportion of these cells are members of a relatively homogeneous population of GABAergic interneurons that can be recognized by their staining with antiserum directed against parvalbumin. Not all intensely GABA positive neurons could be stained with antibodies to parvalbumin. The parvalbumin-negative GABA neurons were somewhat smaller than the parvalbumin-positive ones, falling into the somatic size range associated with the spiny neurons. The division of GABA-immunoreactive neurons into intensely staining and lightly staining categories is somewhat arbitrary, and one interpretation of these results is that some relatively darkly stained spiny neurons were included in our sample of intensely GABA-positive cells. Alternatively, there may be a third class of GABA-positivecells. The parvalbumin neurons in our study are the cells described as GABA interneurons in the combined cytochemical and Golgi studies of Bolam et al. ('83a,b) and the GAD
PARVALBUMIN NEURONS
203
Fig. 6. Photomicrographs of adjacent sections, one stained using the antiserum to calbindin (top),and the other stained for parvalbumin (bottom). Focus has been adjusted to the opposing surfaces of the two sections to maximize the correspondence between them. Blood vessels used as landmarks to align the sections are indicated by the arrows. The
patch compartment visualized by the calbindin staining is outlined in the adjacent parvalbumin section. Parvalbumin neurons can be observed in both the patch and matrix compartments, and parvalbuminpositive processes cross the patchimatrix boundaries.
immunoreactive type 2 cell of Kubota et al. ('87). Like the cell described by those authors, the parvalbumin neurons were aspiny and had varicose dendrites. These features of these neurons would indicate that they w e the same neurons categorized by Chang et al. ('82) as Type V medium
neurons. Some parvalbumin neurons are slightly too large to fit the criteria of Chang et al. for medium-sized neurons. This is in agreement with studies using GABA immunocytochemistry that show the intensely staining interneuron to have larger somata than spiny neurons, although not as
204
R.L. COWAN ET AL. neurons (Bolam et al., '83a,b). The parvalbumin-containing GABA neuron likewise is not the same as the somatostatincontaining striatal interneuron (Chesselet and Graybiel, '861, because the latter does not stain with antisera to GABA (Chesselet and Robbins, '83). These findings indicate that the medium-sized aspiny striatal interneuron population consists of at least three neurochemically distinct subgroups.
Mosaic organizationof parvdbumin cells The present findings show that the somata of the parvalbumin-containing GABA interneurons are distributed in the patch and matrix compartments without preference; that is, they are distributed according to the area occupied by each compartment. We have also observed that the axons and dendrites of parvalbumin-containingneurons can cross the boundaries between the patch (striosome) and matrix Compartments. Inputs to neostriatum from specific areas of cortex and from certain brainstem structures are normally segregated into zones corresponding to patch and matrix compartments (e.g., Graybiel et al., '79). For example, prelimbic cortex and a subdivision of nigral dopaminergic fibers project to patches, while somatosensory, motor, visual cortex, and other dopaminergic and non-dopaminergic fibers project primarily to matrix (Donoghue and Herkenham, '86; Gerfen, '84, '85). Within this pattern there is further specialization of projections based on laminar origin (Gerfen, '90). The dendritic and local axonal fields of the spiny neurons likewise respect the boundaries of the mosaic (Kawaguchi et al., '89; Penny et al., '88), while most interneurons appear to send their dendrites across the boundaries (Penny et al., '88). Parvalbumin-containing GABAergic interneurons may now be added to the list of interneuron types that may mediate an associational interconnection between the two compartments.
The possible functionalrole of the parValbumin cell
Fig. 7. Representative plots of the distribution of parvalbumin neurons in sagittal section pairs stained as in Figure 6. The distribution of parvalbumin neurons is shown in relation to calbindin delineated patch/matrix compartments. The parvalbumin neurons (black dots) are not uniformly distributed but they are present both in the calbindinpoor patches (unshaded) and in the matrix (shaded). ac, anterior commissure; GP, globus pallidus; ic, internal capsule.
large as the largest neostriatal neurons, which are probably aspiny cholinergic neurons (Oertel and Mugnaini, '84; Penny et al., '86). This conclusion also agrees with the results of electron microscopic studies of parvalbumin neurons in the neostriatum (Kita et al., '90). Striatal interneurons are neurochemically diverse; and GABA, substance P, and somatostatin have previously been identified in medium aspiny neurons. The parvalbumincontaining GABA interneuron described here is not related to the substance P-containing interneurons, since they are morphologically distinct from the GABA accumulating
Calcium-binding proteins are involved in a rich diversity of homeostatic and functional properties within the central nervous system (Carafoli, '87; McBurney and Neering, '87). Parvalbumin in fast twitch muscle fibers is thought to provide them with the capacity to relax rapidly by removing calcium from the calmodulin associated contractile protein (Heizmann, '84; Pechere et al., '77). In the mammalian brain, parvalbumin is generally associated with those GABAcontaining neurons that characteristically fire at very high rates. For example, the neurons of the nucleus reticularis thalami are all intensely parvalbumin positive, as are a large number of the projection neurons of the globus pallidus. Fast spiking nonpyramidal neurons of the hippocampus contain parvalbumin, while other GABAergic nonpyramidal neurons in that structure exhibit firing patterns more like those of the pyramidal cells (Kawaguchi et al., '87). It has been suggested that parvalbumin may convey rapid spiking properties to these neurons by, for example, blocking a calcium mediated potassium after hyperpolarization. Carafoli ('871, on the other hand, has suggested that parvalbumin functions not as a calcium buffer, but as a calcium activated second messenger, functioning directly as a transducer of the calcium intracellular signal. In either case, if the correlation between fast firing interneurons and parvalbumin can be extended to the neostriatum, it suggests the existence of a fast firing neostriatal GABAergic interneuron, comparable to the
PARVALBUMIN NEURONS basket cells of the hippocampus and cerebral cortex. This neuron type can interconnect cells across the boundaries of the neostriatal mosaic.
ACKNOWIEDGMEWI'S This work was supported by NIH grant NS20743 and NIH RCDA NS01078 (to C.J.W.), NATO collaborative research grant 0517/85 (C.J.W., P.C.E) and Swiss National Science Foundation grant 3.139.0.88 (to C.W.H.). R. Cowan was supported by NIMH fellowship MH09722.
LnERATuREcm Aronin, N., M. DiFigilia, G.A. Graveland, W.J. Schwartz, and J.Y. Wu (1984) Localization of immunoreactive enkephalins in GABA synthesizing neurons in the rat neostriatum. Brain Res. 300:376-380. Baimbridge, K.G., J.J. Miller, and C.O. Parkes (1982) Calcium-binding protein distribution in the rat brain. Brain Res. 239:519-525. Beckstead, R.M., and K.S. Kersey (1985) Immunohistochemical demonstration of differential substance P, met-enkephalin, and glutamic-aciddecarboxylase-containing cell body and axon distributions in the corpus striatum of the cat. J. Comp. Neurol. 232:481498. Berod, A,, B.K. Hartman. and J.F. Pujol (1981) Importance of fixation in immunohistochemistry: Use of formaldehyde at variable pH for localization of tyrosine hydroxylase. J. Histochem. Cytochem. 29:844-850. Bolam, J.P., D.J. Clarke, A.D. Smith, and P. Somogyi (1983a) A type of aspiny neuron in the rat neostriatum accumulates L3Hl y-aminobutyric acid: combination of Golgi-staining, autoradiography, and electron microscopy. J. Comp. Neurol. 213:121-134. Bolam, J.P., P. Somogyi, H. Takagi, I. Fodor, and A.D. Smith (1983b) Localization of substance P-like immunoreactivity in neurons and nerve terminals in the neostriatum of the rat: A correlated light and electron microscopic study. J. Neurocytol. 12:325-344. Bolam, J.P., Powell, J.F., Wu, J.-Y., and Smith, A.D. (1985) Glutamate decarboxylase-immunoreactive structures in the rat neostriatum: A correlated light and electron microscopic study including a combination of golgi impregnation with immunocytochemistry. J. Comp. Neurol. 237:l-20. Braun, K., H. Scheich, M. Schachner, and C.W. Heizmann (1985a) Distribution of parvalbumin, cytochrome oxidase activity and '"C-2-deoxyglucose uptake in the brain of the zebra finch I. Auditory and vocal motor systems. Cell Tissue Res. 240t101-115. Braun, K., H. Scheich, M. Schachner, and C.W. Heizmann (198513)Distribution of parvalbumin, cytochrome oxidase activity and "C-2-deoxyglucose uptake in the brain of the zebra finch 11. Visual system. Cell Tissue Res. 240:117-127. Carafoli, E. (1987) Intracellular calcium homeostasis. Annu. Rev. Biochem. 56:395433. Celio, M. (1986)Parvalbumin in most gamma-aminohutyric acid-containing neurons of the rat cerebral cortex. Science 231:995-997. Celio, M.R., and C.W. Heizmann (1981) Calcium-binding protein parvalbumin as a neuronal marker. Nature 293:300-302. Celio, M.R., and C.W. Heizmann (1982) Calcium-binding protein parvalbumin is associated with fast contracting muscle fibres. Nature 297504506. Celio, M.R., L. Scharer, J.H. Morrison, A.W. Normanand, 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. Chang, H.T., C.J. Wilson, and S.T. Kitai (1982) A golgi study of rat neostriatal neurons: Light microscopic analysis. J. Comp. Neurol. 208: 107-126. Chesselet, M.F., and E. Robbins (1983) NADPH-Diaphorase positive neurons in the striatum contain the messenger RNA (mRNA) for somatostatin but not for glutamic acid decarboxylase (GAD): An in situ hybridization study. Soc. Neurosci. Abs. 13%?9. Chesselet, M.F., and A.M. Graybiel (1986) Striatal neurons expressing somatostatin-like immunoreactivity: Evidence for a peptidergic interneuronal system in the cat. Neuroscience 17547-571. DiFigilia, M., S. Christakos, and N. Aronin (1989) Ultrastructural localization of immunoreactive calbindin-D28K in the rat and monkey basal
205 ganglia, including subcellular distribution with colloidal gold labeling. J. Comp. Neurol. 2791653-665. Donoghue, J.P., and M. Herkenham (19861 Neostriatal projections from individual cortical fields conform to histochemically distinct compartments in the rat. Brain Res. 365397-403. Eldred, W.D., C. Zucker, H.J. Karten, and S.Yazulla (1983) Comparison of futation and penetration enhancement techniques for use in ultrastructural immunocytochemistry. J. Histochem. Cytochem. 31:285-292. Gerfen, C.R. (1984) The neostriatal mosaic: Compartmentalization of corticostiatal input and striatonigral output systems. Nature 31 I :461464. Gerfen, C.R. (1985) The neostriatal mosaic. I. Compartmental organization of projections from the striatum to the substantia nigra in the rat. J. Comp. Neurol. 236:454-476. Gerfen, C.R. (1990) The neostriatal mosaic: Striatal patch-matrix organization is related to cortical lamination. Science 246:385-388. Gerfen, C.R., K.G. Baimbridge, and J.J. Miller (1985) The neostriatal mosaic: Compartmental distribution of calcium-binding protein and parvalbumin in the basal ganglia of the rat and monkey. Proc. Natl. Acad. Sci. U.S.A. 82t8780-8784. Graybiel, A.M., and M.-F. Chesselet (1984) Compartmental distribution of striatal cell bodies expressing [Metlenkephalin-like immunoreactivity. Proc. Natl. Acad. Sci. U.S.A. 81:7980-7984. Graybiel, A.M., C.W. Ragsdale, Jr., and S.M. Edley (1979) Compartments in the striatum of the cat observed by retrograde cell labeling. Exp. Brain Res. 34:189-195. Heizmann, C.W. (1984) Parvalbumin, a n intracellular calcium-binding protein: Distribution, properties and possible roles in mammalian cells. Experientia 40:910-921. Kawaguchi, Y., H. Katsumaru, T. Kosaka, C.W. Heizmann, and K. Hama (1987) Fast spiking cells in rat hippocampus (CAl region) contain the calcium-binding protein parvalbumin. Brain Res. 41 6:369-374. Kawaguchi, Y., C.J. Wilson, and P.C. Emson (1989) Intracellular recording of identified neostriatal patch and matrix spiny cells in a slice preparation preserving cortical inputs. J. Neurophysiol. 62: 1052-1068. Kita, H., T. Kosaka, and C.W. Heizman (1990) Parvalbumin-immunoreactive neurons in the rat neostriatum: A light and electron microscopic study. Brain Res. (in press). 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. Kubota, Y., S.Inagaki, S. Shimada, S. Kito, and J.-Y. Wu (1987) Glutamate decarboxylase-like immunoreactive neurons in the rat caudate putamen. Brain. Res. Bull. 18:687-697. McBurney, R.N., and I.R. Neering (1987) Neuronal calcium homeostasis. TINS 10:164-169. Miller, J.J.,and K.G. Baimbridge (1983) Biochemical and immunohistochemical correlates of kindling-induced epilepsy: Role of calcium binding protein. Brain Res. 278:322-326. Oertel, W.H., and E. Mugnaini (19841 Immunocytochemical studies of GABAergic neurons in rat basal ganglia and their relations to other neuronal systems. Neurosci. Lett. 47.233-238. Oertel, W.H., G. Riethmuller, E. Mugnaini, D.E. Schmechel, A. Weindl, C. Gramsch, and A. Herze (1983) Opiod peptide-like immunoreactivity localized in GABAergic neurons of rat neostriatum and central amygdaloid nucleus. Life Sci. 33[Suppl. 11:73-76. Panula, P., J.Y. Wu, and P. Emson (1981) Ultrastructure of GABA-neurons in cultures of rat neostriatum. Brain Res. 219:202-207. Pechere, J.-F., J. Derancourt, and J. Haiech (1977) The participation of parvalbumins in the activation-relaxation cycle of vertebrate fast skeletal muscle. FEBS Lett. 75:lll-114. Penny, G.R., S.Afsharpour, and S.T. Kitai (1986) The glutamate decarboxylase-, leucine enkephalin-, methionine enkephalin- and substance P-immunoreactive neurons in the neostriatum of the rat and cat: Evidence for partial population overlap. Neuroscience 17: 1011-1045. Penny, G.R., C.J. Wilson, and S.T. Kitai (1988) Relationship of the axonal and dendritic geometry of spiny projection neurons to the compartmental organization of the neostriatum. J. Comp. Neurol. 269:275-289. Ribak, C.E., J.E. Vaughn, and E. Roberts (1979) The GABA neurons and their axon terminals in the rat corpus striatum as demonstrated by GAD immunocytochemistry. J. Comp. Neurol. 187:261-284. Seto-Ohshima, A., P.C. Emson, E. Lawson, C.Q. Mountjoy, and L.H. Carasco (19881 Loss of matrix calcium-binding protein-containing neurons in Huntington's disease. Lancet 1:1252-1255.
Parvalbumjn-ContainingGmAergic hterneuromin theRat Neostriatmn RONALD L. COWAN, CHARLES J. WILSON, PIERS C. EMSON, AND CLAUS w. HEIZMANN Department of Anatomy and Neurobiology, University of Tennessee College of Medicine, Memphis, Tennessee 38163 (R.L.C., C.J.W.); MRC Group, Institute of Animal Physiology and Genetics Research, Cambridge, CB2 4AT, United Kingdom (P.C.E.); Department of Pediatrics, Division of Clinical Chemistry, University of Zurich, CH-8032 Zurich, Switzerland (C.W.H.)
ABSTRACT Antibodies to the intracellular calcium binding protein parvalbumin were shown to label specifically a distinct group of neostriatal GABAergic neurons. These neurons corresponded to the intensely staining subclass of neostriatal GABAergic neurons that have previously been shown to be a class of aspiny interneurons in the neostriatum. The parvalbumin neurons were aspiny neurons with varicose dendrites distributed throughout the neostriatum in a pattern identical to the intensely stained GABA neurons, and both populations of neurons showed increased numbers in the lateral part of the neostriatum. Double labeling of single neurons with both the GABA and parvalbumin antisera showed that all parvalbumin neurons were positive for GABA, but some GABA labelled neurons were not immunoreactive for parvalbumin. These parvalbumin-negativeGABAergic neurons were morphologically similar to the spiny projection neurons, which are GABAergic but usually are not so heavily stained. The relationship of the GABA-containing parvalbumin neurons to the striatal mosaic organization was determined by using immunocytochemistry for another calcium binding protein, calbindin D28K, to label the matrix compartment of the striatum. The distribution of parvalbumin-positive neurons relative to the calbindin-positive matrix and calbindin-poor patches was determined by using pairs of adjacent sections stained with the calbindin and parvalbumin antisera. This analysis showed that the somata of the parvalbumin neurons were present in both patch and matrix compartments, and their axons and dendrites crossed the boundaries between compartments. A quantitative analysis of the number of neurons in each compartment revealed that the neurons showed no preferential distribution in either compartment, but instead were present according to the area occupied by that compartment. Approximately 10%of parvalbumin neurons were in the patch compartment, and in these same sections, the patch compartment occupied approximately 10% of the area of those sections. Staining with parvalbumin antibodies can therefore be used to identify a single class of GABAergic aspiny interneurons that is present in both patch and matrix compartments, and whose processes cross the borders between these compartments. Key words: basal ganglia, inhibition, calbindin, calcium binding protein, immunocytochemistry
In addition to the morphological diversity of neurons in the neostriatum as revealed in Golgi studies (e.g., Chang et al., '82), immunocytochemical studies have revealed a n enormous cytochemical diversity in the populations of gamma-amino butyric acid (GABA), enkephalin, substance P, somatostatin, and acetylcholine-containing neurons in the neostriatum (Aronin et al., '84; Beckstead and Kersey, '85; Bolam et al., '85; Graybiel and Chesselet, '84;Oertel and Mugnaini, '84; Oertel et al., '83; Panula et al., '81; Penny et al., '86; Ribak et al., '79). For the most part, the cell types identified in this way have mapped easily onto the 0 1990 WILEY-LISS. TNC
more traditional classes of cells according to their somatodendritic morphology. While the identification of cell types in the neostriatum on the basis of the presence or absence of substances related to neurotransmission seems to require little justification, there is no a priori reason to expect that other molecules with more general functions would be preferentially localized in neurons of particular types. Nonetheless, some such
Accepted August 27, 1990.
198
substances show staining patterns suggestive of a preferential localization in certain cell types. Among these are two calcium binding proteins, calbindin-D28k and parvalbumin. Striatal calbindin was initially discovered by using radioimmunoassay techniques (Baimbridge et al., '82), and was later localized to spiny neostriatal neurons (DiFiglia et al., '89) located specifically in the matrix (acetylcholinesterase-rich, mu-opiate receptor-poor) compartment of the neostriatum (Gerfen et al., '85). Calbindin has been linked to a variety of cellular activities, including a role in kindling induced epilepsy, (Miller and Baimbridge, '831, pathological processes in Huntington's disease (Seto-Ohshima et al., '881, and cellular metabolic activity levels (Braun et al., '85a,b; Celio et al., '86). Parvalbumin, on the other hand, is not localized in the medium spiny projection neurons, but is concentrated in a group of unidentified striatal neurons distributed throughout the neostriatum (Celio and Heizmann, '81;Gerfen et al., '85). These parvalbumin neurons are of similar size and distribution to striatal GABAergic interneurons (Gerfen et al., '851, suggesting that the two neuron populations may overlap or be identical. This is consistent with the localization of parvalbumin in other areas of the central nervous system. Parvalbumin has been demonstrated in GABAergic cells such as cerebellar Purkinje, basket, and stellate cells; hippocampal basket cells (Celio and Heizmann, '82; Kosaka et al., '87); and neocortical interneurons (Celio et al., '86; Celio, '86). In addition, parvalbumin is present in fast twitch muscle fibers, where it is thought to participate in postcontraction relaxation (Celio and Heizmann, '82; Pechere et al., '771, and in fast spiking hippocampal neurons (Kawaguchi et al., '871, where its calcium binding capacities have been postulated to play a role in conferring rapid firing properties on these neurons. In this report we show that parvalbumin is preferentially localized in a small population of GABAergic interneurons in the neostriatum that have been recognized previously on the basis of their high concentration of GABA and GABAsynthesizing enzymes, and that correspond to a classical type of neostriatal interneuron on the basis of their somatodendritic morphology. In addition, we show that this population of neurons is present in both the calbindin-poor patches and in the calbindin-rich matrix compartment of the neostriatum, and that their dendrites freely cross the boundary that separates these two compartments of the neostriatal neuropil.
R.L. COWAN ET AL.
methods were employed. In some animals the brains were sectioned on a rotary microtome at a thickness of 2-4 pm after embedding in polyethylene glycol (MW1000). In this procedure, blocks of tissue containing the neostriatum were dehydrated in alcohol, infiltrated in polyethylene glycol at 40"C, and allowed to cool in a desiccator at 4°C. Adjacent pairs of sections were cut on a metal knife at room temperature and collected into buffer. The polyethylene glycol embedment was removed from the sections by repeated washing in buffer. The immunocytochemical procedure was then applied to the free-floating semi-thin sections. In the other procedure, adjacent 40-100 pm sections were cut from unembedded tissue on a vibratome, collected in buffer, and processed free-floating. The problem of identification of neurons in both of the section pairs was simpler in the thinner sections, in which nearly all neurons present in one of the pairs could also be identified in the other. The thicker sections cut with a vibratome were more durable, however; and their thickness made it easier to verify landmarks in adjacent sections. As a result, most of the analysis was performed on the vibratome sectioned material, in which matching faces of the sections were identified and only neurons whose somata were sectioned so as to be present at the surface of both sections could be counted in the analysis. Comparison of the distribution of neurons in the neostriatal patches and matrix was likewise performed on adjacent sections arranged so that matching faces of vibratome sections could be identified and compared. Rabbit anti-GABA antiserum was purchased from Immunonuclear Corporation. Rabbit anti-chicken calbindin and sheep anti-rabbit parvalbumin polyclonal antisera were raised by P.C. Emson at Cambridge and goat anti-rat parvalbumin was raised by C.W. Heizmann in Zurich. Standard procedures were used for immunocytochemical staining and were the same with all three antisera. They were 1)treatment for 4 hours at room temperature in 4% normal serum (normal serum corresponded to the species of the secondary antibody) in 0.01 M potassium phosphate buffered saline (KPBS), pH 7.4, containing 0.5% Triton-X, 2) overnight incubation a t 4°C in primary antisera (dilutions of 1:1,000-1:2,000) in 4% normal serum in KPBS containing 0.5% Triton-X; 3) washing in KPBS, followed by 1hour room temperature incubation in biotinylated secondary antisera (dilution of 1:lOOor 1:200) in KPBS containing 0.5% Triton-X. Goat anti-rabbit secondary was used to bind the GABA and calbindin antibodies, and rabbit anti-goat secondary was used to bind the parvalbumin antibody; 4) washing in KPBS, followed by 30 minute incubation in EXPERWIENTALPROCEDURES avidin-biotin-peroxidase solution in KPBS with 0.5% TriMale adult (200-500 g) Long-Evans or Sprague-Dawley ton-X, and 5) Washing in KPBS, followed by staining with rats were used in all experiments. The brains were fixed by diaminobenzidine. intracardial perfusion of aldehydes after the animals were Negative controls were obtained by substituting primary deeply anesthetized with sodium pentobarbital (75 mgkg antisera with normal serum, and then processing the tissue i.p.1. The fixative consisted of either sequential 4%formal- identically to tissue containing primary antisera. Preabsorpdehyde in 0.1 M acetate buffer (pH 6.5) followed by 4% tion of the antisera to calbindin or parvalbumin with their formaldehyde and 0.1% glutaraldehyde in 0.05 M borate corresponding high pressure liquid chromatography-puribuffer (pH 9.5) according to Berod et al. ('81) or with a fied antigens M) was used as a control for nonspecific simpler fixative consisting of 4%formaldehyde and 0.1-3% binding of these antisera to the tissue sections. In all cases glutaraldehyde in 0.15 M sodium phosphate buffer (pH 7.4). the characteristic staining patterns of the antisera were No consistent differences were observed in labeling between absent in the control sections. tissue prepared with the different buffers. Sections were pretreated for 30 minutes with 1% sodium borohydride before further processing for immunocytochemistry (ElRESULTS dred et al., '83). ParValbuminneurons For double labeling of cells with GABA and parvalbumin, The parvalbumin antisera labeled medium-sized striatal individual neurons were identified in adjacent sections processed with the two antisera. Two different sectioning neurons having numerous dendritic processes (Fig. 1).
PARVALBUMIN NEURONS Staining was generally of equal intensity in the nucleus, cytoplasm, and dendrites, and the nucleolus appeared as a clear zone. The neurons had a broad size range, and were both polygonal and fusiform. Using the criterion of Chang et al. ('82), we classified the neurons into small, medium, or large on the basis of cross-sectional area measurements. A majority (83%)of the parvalbumin neurons were mediumsized (cross-sectional area 100-300 km'); 15%were small (cross-sectional area < 100 km2), and a small percentage (2%)were large, having cross-sectional areas greater than 300 km'. Dendrites were darkly stained, with very fine cell processes plainly visible. Dendritic spines, if present, would have been clearly visible. The dendrites were all aspiny throughout their length, and no somatic spines were observed. Most neurons had varicose dendritic processes (Figure 1, arrows), and many such processes were seen in isolation in the neuropil. A fine plexus of axons and axonal varicosities were present throughout the neostriatum.
Spatial distributionand somatic sizes To determine whether parvalbumin neurons could be a subset of the intensely staining GAJ3A-containingneurons in the neostriatum, we compared the spatial distributions and the somatic sizes of these two cell types in sections cut from the same brains. GABA antisera worked best on tissue fixed with the high glutaraldehyde fixatives. Under these conditions, many darkly labeled neurons were seen throughout the neostriatum. The GABA neurons were darkly stained in their nuclei and cytoplasm, while little dendritic staining was evident. Staining in the neuropil was much greater in the GABA sections than in the sections stained for parvalbumin. The distribution of the GABA neurons was similar to that of the
199
parvalbumin neurons. An example showing a plot of all intensely stained GABA neurons and all parvalbumin immunoreactive neurons in adjacent sections is shown in Figure 2. The similarity in their distribution as seen in that figure was consistent in all cases. The strong tendency for more parvalbumin-stained neurons to be located in the lateral half of the neostriatum was also shared by the intensely staining GABA-positive cells. Although not greatly dissimilar in numbers, there was a consistent difference between the cell populations, with GABA neurons being slightly more numerous (e.g. Fig. 2). To examine the size distribution of the neurons, we measured the cross-sectional area of 200 neurons of each type selected randomly from 16 sections from two animals. To avoid problems with penetration of reagents used in the immunocytochemical procedures, we analysed only darkly stained neurons that intersected a single surface of the section, and whose nucleoli were in focus at the cut surface. Sections used in the analysis were separated by several hundred micrometers, insuring that cells would not be counted more than once. The distance between prominent tissue landmarks measured in sections stained with the parvalbumin and GABA antisera were the same, indicating that there was no differential tissue shrinkage that could complicate comparison of cell sizes. No attempt was made to correct for the shrinkage common to the two groups of sections. The cross-sectional areas of the somata were measured by using an on-line video digitizer and image analysis program (Image, obtained from Wayne Rasband, NIH). A comparison of the cross-sectional areas of GABA neurons and parvalbumin neurons revealed that parvalbumin-containing cells fell within the size distribution of GABA-positive neurons. The somatic size distributions of
Fig. 1. Photomicrograph of a pawalbumin-containing striatal interneuron. Note the dendritic varicosities (arrows) and the absence of somatic or dendritic spines.
R.L. COWAN ET AL.
200
Fig. 2. Plots of adjacent sagittal neostriatal sections showing the similar distributions of pawalbumin-positive neurons (left) and intensely GABA-positive neurons (right).The intensely GABA-positive
both intensely GABA-positive and parvalbumin-positive neurons are shown in Figure 3. The cross-sectional areas of the parvalbumin-containingneurons ranged from 65 to 483 pm2.The distribution obtained for GABA neurons covered this same range, but also included a group of smaller neurons, ranging from 40 to 379 pm'. While both distributions had about 2% large neurons, the GABA group had 50% small neurons (cross-sectional area < l o 0 pm2, see above) compared to the 15%small neurons in the parvalbumin group. The presence of the smaller neurons in the GABA sample resulted in a difference in the average cross-sectional area measurements for each group. The mean cross-sectional area for the GABA neurons was 117 60 pm2 (n = 200), while that of the parvalbumin neurons was larger at 148 57 pm2 (n = 200). An unpaired t-test (df = 398) revealed that the cross-sectional areas of the two groups were significantly different (P < 0.01). This difference and the distributions shown in Figure 3 demonstrate that the parvalbumin cells are completely contained within the distribution of GABA cells, but cannot account for all GABA neurons.
*
*
ColocaziZationof GABA and parvalbumin Parvalbumin and GABA immunoreactivity were tested on single cells in adjacent sections. Alternate sections were stained with a different antiserum, and the sections were placed in register by using blood vessels as fiducial marks. Neurons whose somata were cut through at the faces of the sections were identified in both sections of a pair and examined for the presence of immunoreactivity to GABA and parvalbumin. The fact that there were more intensely GABA-positive neurons than parvalbumin-positive cells ruled out the possibility of an exclusive one-to-one correspondence of the two cell populations. The remaining possibility, that the parvalbumin neurons were a subset of GABApositive cells, was tested by examining the GABA reactivity of parvalbumin-containing neurons. All the parvalbumin neurons tested in this manner were also GABAergic. Exam-
neurons are slightly more numerous than the parvalbumin neurons, but they share a common distribution within the nucleus. AC, anterior commissure; IC, internal capsule; ACp, anterior commissure posterior.
j0 1
UY
5L
40
GABA (n=200) Q
PARVALBUMIN (n=200)
2 30 z
:20
cL-$
Q)
n
El 10
z
0
45 65 85 105 125 145 165 185 205 225 245 >245
Cross Sectional Area (pd)
Fig. 3. The distribution of the cross-sectional areas of neurons stained using antisera to GABA (dark shading) and to parvalbumin (light shading) neurons. The distribution of the GABA neurons contains a population of smaller cells, as well as a group of neurons corresponding to the distribution observed for parvalbumin-positive neurons. The size distribution of the GABA-positivebut parvalbuminnegative cell group corresponds to that of the spiny projection neurons.
ples of two pairs of parvalbumin-containingGABA neurons are shown in Figure 4A-D. As expected, there were some intensely GABA-positive neurons in the same sections that did not also contain parvalbumin (not shown).
Pawalbuminneuronsin the neostriatal mosaic Adjacent sections were used to examine the distribution of parvalbumin neurons in relation to the neostriatal mosaic, as indicated by the distribution of calbindin-labeled spiny neurons. One section from a pair of adjacent sections was processed to visualize calbindin and the other half was processed to visualize parvalbumin. The boundaries of the calbindin-positive patches were most clear in the caudal and
PARVALBUMIN NEURONS
201
Fig. 4. Photomicrographs of adjacent sections stained usingantihodies to GABA (left,A,C) and parvalbumin (right, B,D).The focal level of the microscope is set to the opposing surfaces of the sections to maximize the correspondence between the two sections. A and B are adjacent and C and D are adjacent. Arrows in all sections point to blood vessels used as landmarks for alignment of the sections. The parallel
diagonal lines in the sections labelled for GABA are artifacts of vihratome sectioningpresent at the surface of the section. These are not present in the parvalbumin sections because of the minimal neuropil staining in these sections. Parvalhumin neurons in B correspond to GABA-stained neurons in A, and parvalhumin-positive cells in D are the same as those stained for GABA in C.
ventral parts of the neostriatum near the globus pallidus, because the staining for calbindin is strongest in this part of the neostriatum. Therefore, the analysis of the mosaic organization of parvalbumin neurons was restricted to these regions, where nearly every spiny neuron in the matrix compartment is positive for calbindin, while calbindin staining in the neurons of the patches is weak, and sometimes undetectable. An example of the calbindin staining pattern as seen at low magnification is shown in Figure 5. Previous work has established that the patch boundaries seen in this way correspond to the mosaic organization observed with other methods, and that the calbindin-poor patches correspond to the striosomes (Gerfen et al., '85). Ten plots of the locations of parvalbumin neurons and patch-matrix boundaries were made from pairs of wellstained adjacent sections from three animals. The boundaries of the calbindin-poor patches were drawn with the aid of a drawing tube and a 4 x objective. Parvalbumin neurons from the corresponding adjacent section were then superimposed on the plot of the calbindin-stained section. The ratio of patch area to the combined patchimatrix area was
calculated to determine the percentage of area occupied by the patches. The ratio of parvalbumin neurons in patches to the total number of parvalbumin neurons in the combined patchimatrix area was calculated and compared to the percentage of area occupied by the patches. An example illustrating the appearance of calbindin and parvalbumin staining on adjacent sections is shown in Figure 6. Parvalbumin neurons and their axons and dendrites were present in both the calbindin-poor patches and in the matrix compartment. In addition, parvalbumin immunoreactive axons and dendrites frequently crossed the compartmental boundaries. Plots of the entire usable portion of pairs of sections placed in register as in Figure 6 suggested that the neurons were not preferentially associated with either compartment. Examples from these plots are shown in Figure 7A-C. Quantitative analysis of 10 sections from 3 different cases indicated that parvalbumin neurons are distributed in the patchimatrix compartments in proportion to the area occupied by the respective compartments. Calbindin-poor patches occupied 9.8% of the total striatal area in the sections, while patch parvalbumin
R.L. COWAN ET AL.
202
Fig. 5. Photomicrograph of calbindin staining in a parasagittal section of the rat neostriatum, as used for the study of the compartmental localization of pawalbumin-positive neurons. The more heavily stained regions compose the matrix, while the calbindin-poor areas are patches. GP, globus pallidus.
neurons composed 10.2% of the total number of parvalbumin neurons in the same regions.
DISCUSSION Identity of the parvalbuminneurons In recent years it has become increasingly evident that the neostriatum contains two different classes of GABAcontaining neurons. While the most common neuron, the spiny projection cell, is easily recognized by its characteristic and well-known morphological features, the identity and functional properties of GABAergic interneurons are much less well understood. In immunocytochemical studies, the GABAergic interneurons have been recognized by their intense staining. Somatic GABA immunoreactivity of spiny neurons is very low in the absence of special treatments such as colchicine pretreatment. The population of intensely GABA-positiveneurons observed under these conditions has been shown to include members of a class of interneurons recognized by their morphological features when impregnated with the Golgi method, their cytological features in electron micrographs, and their failure to be stained by retrograde axonal transport from the substantia
nigra and globus pallidus (Bolam et al., '83a,b; Chang et al., '82; Gerfen et al., '85). It is not known whether these cells represent a single neuronal population or are heterogeneous, and their position within the established circuitry of the neostriatum is unclear. The present results indicate that at least a large proportion of these cells are members of a relatively homogeneous population of GABAergic interneurons that can be recognized by their staining with antiserum directed against parvalbumin. Not all intensely GABA positive neurons could be stained with antibodies to parvalbumin. The parvalbumin-negative GABA neurons were somewhat smaller than the parvalbumin-positive ones, falling into the somatic size range associated with the spiny neurons. The division of GABA-immunoreactive neurons into intensely staining and lightly staining categories is somewhat arbitrary, and one interpretation of these results is that some relatively darkly stained spiny neurons were included in our sample of intensely GABA-positive cells. Alternatively, there may be a third class of GABA-positivecells. The parvalbumin neurons in our study are the cells described as GABA interneurons in the combined cytochemical and Golgi studies of Bolam et al. ('83a,b) and the GAD
PARVALBUMIN NEURONS
203
Fig. 6. Photomicrographs of adjacent sections, one stained using the antiserum to calbindin (top),and the other stained for parvalbumin (bottom). Focus has been adjusted to the opposing surfaces of the two sections to maximize the correspondence between them. Blood vessels used as landmarks to align the sections are indicated by the arrows. The
patch compartment visualized by the calbindin staining is outlined in the adjacent parvalbumin section. Parvalbumin neurons can be observed in both the patch and matrix compartments, and parvalbuminpositive processes cross the patchimatrix boundaries.
immunoreactive type 2 cell of Kubota et al. ('87). Like the cell described by those authors, the parvalbumin neurons were aspiny and had varicose dendrites. These features of these neurons would indicate that they w e the same neurons categorized by Chang et al. ('82) as Type V medium
neurons. Some parvalbumin neurons are slightly too large to fit the criteria of Chang et al. for medium-sized neurons. This is in agreement with studies using GABA immunocytochemistry that show the intensely staining interneuron to have larger somata than spiny neurons, although not as
204
R.L. COWAN ET AL. neurons (Bolam et al., '83a,b). The parvalbumin-containing GABA neuron likewise is not the same as the somatostatincontaining striatal interneuron (Chesselet and Graybiel, '861, because the latter does not stain with antisera to GABA (Chesselet and Robbins, '83). These findings indicate that the medium-sized aspiny striatal interneuron population consists of at least three neurochemically distinct subgroups.
Mosaic organizationof parvdbumin cells The present findings show that the somata of the parvalbumin-containing GABA interneurons are distributed in the patch and matrix compartments without preference; that is, they are distributed according to the area occupied by each compartment. We have also observed that the axons and dendrites of parvalbumin-containingneurons can cross the boundaries between the patch (striosome) and matrix Compartments. Inputs to neostriatum from specific areas of cortex and from certain brainstem structures are normally segregated into zones corresponding to patch and matrix compartments (e.g., Graybiel et al., '79). For example, prelimbic cortex and a subdivision of nigral dopaminergic fibers project to patches, while somatosensory, motor, visual cortex, and other dopaminergic and non-dopaminergic fibers project primarily to matrix (Donoghue and Herkenham, '86; Gerfen, '84, '85). Within this pattern there is further specialization of projections based on laminar origin (Gerfen, '90). The dendritic and local axonal fields of the spiny neurons likewise respect the boundaries of the mosaic (Kawaguchi et al., '89; Penny et al., '88), while most interneurons appear to send their dendrites across the boundaries (Penny et al., '88). Parvalbumin-containing GABAergic interneurons may now be added to the list of interneuron types that may mediate an associational interconnection between the two compartments.
The possible functionalrole of the parValbumin cell
Fig. 7. Representative plots of the distribution of parvalbumin neurons in sagittal section pairs stained as in Figure 6. The distribution of parvalbumin neurons is shown in relation to calbindin delineated patch/matrix compartments. The parvalbumin neurons (black dots) are not uniformly distributed but they are present both in the calbindinpoor patches (unshaded) and in the matrix (shaded). ac, anterior commissure; GP, globus pallidus; ic, internal capsule.
large as the largest neostriatal neurons, which are probably aspiny cholinergic neurons (Oertel and Mugnaini, '84; Penny et al., '86). This conclusion also agrees with the results of electron microscopic studies of parvalbumin neurons in the neostriatum (Kita et al., '90). Striatal interneurons are neurochemically diverse; and GABA, substance P, and somatostatin have previously been identified in medium aspiny neurons. The parvalbumincontaining GABA interneuron described here is not related to the substance P-containing interneurons, since they are morphologically distinct from the GABA accumulating
Calcium-binding proteins are involved in a rich diversity of homeostatic and functional properties within the central nervous system (Carafoli, '87; McBurney and Neering, '87). Parvalbumin in fast twitch muscle fibers is thought to provide them with the capacity to relax rapidly by removing calcium from the calmodulin associated contractile protein (Heizmann, '84; Pechere et al., '77). In the mammalian brain, parvalbumin is generally associated with those GABAcontaining neurons that characteristically fire at very high rates. For example, the neurons of the nucleus reticularis thalami are all intensely parvalbumin positive, as are a large number of the projection neurons of the globus pallidus. Fast spiking nonpyramidal neurons of the hippocampus contain parvalbumin, while other GABAergic nonpyramidal neurons in that structure exhibit firing patterns more like those of the pyramidal cells (Kawaguchi et al., '87). It has been suggested that parvalbumin may convey rapid spiking properties to these neurons by, for example, blocking a calcium mediated potassium after hyperpolarization. Carafoli ('871, on the other hand, has suggested that parvalbumin functions not as a calcium buffer, but as a calcium activated second messenger, functioning directly as a transducer of the calcium intracellular signal. In either case, if the correlation between fast firing interneurons and parvalbumin can be extended to the neostriatum, it suggests the existence of a fast firing neostriatal GABAergic interneuron, comparable to the
PARVALBUMIN NEURONS basket cells of the hippocampus and cerebral cortex. This neuron type can interconnect cells across the boundaries of the neostriatal mosaic.
ACKNOWIEDGMEWI'S This work was supported by NIH grant NS20743 and NIH RCDA NS01078 (to C.J.W.), NATO collaborative research grant 0517/85 (C.J.W., P.C.E) and Swiss National Science Foundation grant 3.139.0.88 (to C.W.H.). R. Cowan was supported by NIMH fellowship MH09722.
LnERATuREcm Aronin, N., M. DiFigilia, G.A. Graveland, W.J. Schwartz, and J.Y. Wu (1984) Localization of immunoreactive enkephalins in GABA synthesizing neurons in the rat neostriatum. Brain Res. 300:376-380. Baimbridge, K.G., J.J. Miller, and C.O. Parkes (1982) Calcium-binding protein distribution in the rat brain. Brain Res. 239:519-525. Beckstead, R.M., and K.S. Kersey (1985) Immunohistochemical demonstration of differential substance P, met-enkephalin, and glutamic-aciddecarboxylase-containing cell body and axon distributions in the corpus striatum of the cat. J. Comp. Neurol. 232:481498. Berod, A,, B.K. Hartman. and J.F. Pujol (1981) Importance of fixation in immunohistochemistry: Use of formaldehyde at variable pH for localization of tyrosine hydroxylase. J. Histochem. Cytochem. 29:844-850. Bolam, J.P., D.J. Clarke, A.D. Smith, and P. Somogyi (1983a) A type of aspiny neuron in the rat neostriatum accumulates L3Hl y-aminobutyric acid: combination of Golgi-staining, autoradiography, and electron microscopy. J. Comp. Neurol. 213:121-134. Bolam, J.P., P. Somogyi, H. Takagi, I. Fodor, and A.D. Smith (1983b) Localization of substance P-like immunoreactivity in neurons and nerve terminals in the neostriatum of the rat: A correlated light and electron microscopic study. J. Neurocytol. 12:325-344. Bolam, J.P., Powell, J.F., Wu, J.-Y., and Smith, A.D. (1985) Glutamate decarboxylase-immunoreactive structures in the rat neostriatum: A correlated light and electron microscopic study including a combination of golgi impregnation with immunocytochemistry. J. Comp. Neurol. 237:l-20. Braun, K., H. Scheich, M. Schachner, and C.W. Heizmann (1985a) Distribution of parvalbumin, cytochrome oxidase activity and '"C-2-deoxyglucose uptake in the brain of the zebra finch I. Auditory and vocal motor systems. Cell Tissue Res. 240t101-115. Braun, K., H. Scheich, M. Schachner, and C.W. Heizmann (198513)Distribution of parvalbumin, cytochrome oxidase activity and "C-2-deoxyglucose uptake in the brain of the zebra finch 11. Visual system. Cell Tissue Res. 240:117-127. Carafoli, E. (1987) Intracellular calcium homeostasis. Annu. Rev. Biochem. 56:395433. Celio, M. (1986)Parvalbumin in most gamma-aminohutyric acid-containing neurons of the rat cerebral cortex. Science 231:995-997. Celio, M.R., and C.W. Heizmann (1981) Calcium-binding protein parvalbumin as a neuronal marker. Nature 293:300-302. Celio, M.R., and C.W. Heizmann (1982) Calcium-binding protein parvalbumin is associated with fast contracting muscle fibres. Nature 297504506. Celio, M.R., L. Scharer, J.H. Morrison, A.W. Normanand, 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. Chang, H.T., C.J. Wilson, and S.T. Kitai (1982) A golgi study of rat neostriatal neurons: Light microscopic analysis. J. Comp. Neurol. 208: 107-126. Chesselet, M.F., and E. Robbins (1983) NADPH-Diaphorase positive neurons in the striatum contain the messenger RNA (mRNA) for somatostatin but not for glutamic acid decarboxylase (GAD): An in situ hybridization study. Soc. Neurosci. Abs. 13%?9. Chesselet, M.F., and A.M. Graybiel (1986) Striatal neurons expressing somatostatin-like immunoreactivity: Evidence for a peptidergic interneuronal system in the cat. Neuroscience 17547-571. DiFigilia, M., S. Christakos, and N. Aronin (1989) Ultrastructural localization of immunoreactive calbindin-D28K in the rat and monkey basal
205 ganglia, including subcellular distribution with colloidal gold labeling. J. Comp. Neurol. 2791653-665. Donoghue, J.P., and M. Herkenham (19861 Neostriatal projections from individual cortical fields conform to histochemically distinct compartments in the rat. Brain Res. 365397-403. Eldred, W.D., C. Zucker, H.J. Karten, and S.Yazulla (1983) Comparison of futation and penetration enhancement techniques for use in ultrastructural immunocytochemistry. J. Histochem. Cytochem. 31:285-292. Gerfen, C.R. (1984) The neostriatal mosaic: Compartmentalization of corticostiatal input and striatonigral output systems. Nature 31 I :461464. Gerfen, C.R. (1985) The neostriatal mosaic. I. Compartmental organization of projections from the striatum to the substantia nigra in the rat. J. Comp. Neurol. 236:454-476. Gerfen, C.R. (1990) The neostriatal mosaic: Striatal patch-matrix organization is related to cortical lamination. Science 246:385-388. Gerfen, C.R., K.G. Baimbridge, and J.J. Miller (1985) The neostriatal mosaic: Compartmental distribution of calcium-binding protein and parvalbumin in the basal ganglia of the rat and monkey. Proc. Natl. Acad. Sci. U.S.A. 82t8780-8784. Graybiel, A.M., and M.-F. Chesselet (1984) Compartmental distribution of striatal cell bodies expressing [Metlenkephalin-like immunoreactivity. Proc. Natl. Acad. Sci. U.S.A. 81:7980-7984. Graybiel, A.M., C.W. Ragsdale, Jr., and S.M. Edley (1979) Compartments in the striatum of the cat observed by retrograde cell labeling. Exp. Brain Res. 34:189-195. Heizmann, C.W. (1984) Parvalbumin, a n intracellular calcium-binding protein: Distribution, properties and possible roles in mammalian cells. Experientia 40:910-921. Kawaguchi, Y., H. Katsumaru, T. Kosaka, C.W. Heizmann, and K. Hama (1987) Fast spiking cells in rat hippocampus (CAl region) contain the calcium-binding protein parvalbumin. Brain Res. 41 6:369-374. Kawaguchi, Y., C.J. Wilson, and P.C. Emson (1989) Intracellular recording of identified neostriatal patch and matrix spiny cells in a slice preparation preserving cortical inputs. J. Neurophysiol. 62: 1052-1068. Kita, H., T. Kosaka, and C.W. Heizman (1990) Parvalbumin-immunoreactive neurons in the rat neostriatum: A light and electron microscopic study. Brain Res. (in press). 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. Kubota, Y., S.Inagaki, S. Shimada, S. Kito, and J.-Y. Wu (1987) Glutamate decarboxylase-like immunoreactive neurons in the rat caudate putamen. Brain. Res. Bull. 18:687-697. McBurney, R.N., and I.R. Neering (1987) Neuronal calcium homeostasis. TINS 10:164-169. Miller, J.J.,and K.G. Baimbridge (1983) Biochemical and immunohistochemical correlates of kindling-induced epilepsy: Role of calcium binding protein. Brain Res. 278:322-326. Oertel, W.H., and E. Mugnaini (19841 Immunocytochemical studies of GABAergic neurons in rat basal ganglia and their relations to other neuronal systems. Neurosci. Lett. 47.233-238. Oertel, W.H., G. Riethmuller, E. Mugnaini, D.E. Schmechel, A. Weindl, C. Gramsch, and A. Herze (1983) Opiod peptide-like immunoreactivity localized in GABAergic neurons of rat neostriatum and central amygdaloid nucleus. Life Sci. 33[Suppl. 11:73-76. Panula, P., J.Y. Wu, and P. Emson (1981) Ultrastructure of GABA-neurons in cultures of rat neostriatum. Brain Res. 219:202-207. Pechere, J.-F., J. Derancourt, and J. Haiech (1977) The participation of parvalbumins in the activation-relaxation cycle of vertebrate fast skeletal muscle. FEBS Lett. 75:lll-114. Penny, G.R., S.Afsharpour, and S.T. Kitai (1986) The glutamate decarboxylase-, leucine enkephalin-, methionine enkephalin- and substance P-immunoreactive neurons in the neostriatum of the rat and cat: Evidence for partial population overlap. Neuroscience 17: 1011-1045. Penny, G.R., C.J. Wilson, and S.T. Kitai (1988) Relationship of the axonal and dendritic geometry of spiny projection neurons to the compartmental organization of the neostriatum. J. Comp. Neurol. 269:275-289. Ribak, C.E., J.E. Vaughn, and E. Roberts (1979) The GABA neurons and their axon terminals in the rat corpus striatum as demonstrated by GAD immunocytochemistry. J. Comp. Neurol. 187:261-284. Seto-Ohshima, A., P.C. Emson, E. Lawson, C.Q. Mountjoy, and L.H. Carasco (19881 Loss of matrix calcium-binding protein-containing neurons in Huntington's disease. Lancet 1:1252-1255.
