GABAergic interneurons in the somatosensory thalamus of the guinea-pig: A light and ultrastructural immunocytochemical investigation

Neuroscience

Vol. 59,

Pergamon

0306-4522(93)E0057-W

Printed in

Great

NO. 4, pp. 961-973, 1994 Ekvier Science Ltd Copyright 0 1994IBRO

Britain. All rights reserved 0306-4522/94-$6.00 + 0.00

GABAERGIC INTERNEURONS IN THE SOMATOSENSORY THALAMUS OF THE GUINEA-PIG: A LIGHT AND ULTRASTRUCTURAL IMMUNOCYTOCHEMICAL INVESTIGATION R. SPREAFICO,*~ C. FRASSONI,* P. ARCELLI$ and S. DE BIASI~

lDipartimento Neurofisiologia, Istituto Nazionale Neurologico C. Besta, via Celoria 11,20133 Milano, Italy SDipartimento Fisiologia e Biochimica Generali, Sezione di Istologia e Anatomia Umana, Universitl di Milano, Italy A&met-This work was performed to confirm previous data reporting the presence of GABAergic interneurons in the ventrobasal complex of guinea-pig, and to investigate the intrinsic organization of this nucleus compared to that of thalamic nuclei lacking interneurons. Immunocytochemical experiments were performed on the thalamus of adult guinea-pigs perfused with mixed aldehydes using an anti-GABA serum. At light microscopy, the immunoreaction on floating Vibratome sections showed that GABAergic neurons are present only in the reticular and lateral geniculate nuclei and in the ventrobasal complex. Quantitative evaluation of their number indicated that they are 20 and 15% of the total neuronal population in lateral geniculate nucleus and ventrobasal complex, respectively, while they are less than 1% in ventrolateral nucleus. At the ultrastructural level, the postembedding immunogold procedure showed the presence, in the ventrobasal complex, of GABA-labeled profiles involved in complex synaptic arrangements similar to those found in carnivores and primates. Conversely. GABA-labeled terminals in thalamic nuclei devoid of interneurons formed exclusively axo-dendritic or axo-somatic contacts, like in rats and mice. The present data suggest that GABAergic neurons in the ventrobasal complex of guinea-pigs give rise to functionally important rearrangements of its intrinsic synaptic organization and that they represent the morphological basis for an intrinsic modulatory mechanism that is absent in other thalamic nuclei lacking inhibitory interneurons. The phylogenetic implications of these findings are also discussed in comparison to other animal species.

Only few neuroanatomical studies have been performed on guinea-pig brain since this animal species is not used in research as much as mouse, rat, cat or monkey.L0~‘7~22~23~M~4’ Physiological in vitro studies have been performed in different brain regions of guinea-pigs, while in oivo experiments are rarely In the last few years, interest in reported. 9~‘5~16~37~38~50 guinea-pigs has risen, partly due to the new whole brain in vitro technique proposed by Llinas and co-workers.8*3’ This technique provides new electrophysiological and pharmacological approaches to the study of different areas of the CNS, including deep brain regions, as it bypasses the limitations of the previous brain slices methods and some problems of the in viuo studies. The interest of neurobiologists in tTo whom correspondence should be addressed F, terminals with flattened vesicles; ICC, immunocytochemistry; IR, immunoreactive; LG, lateral geniculate nucleus; LGV, large granular vesicle; LR, large terminals with round vesicles; NGS, normal goat serum; PBS, phosphate buffered saline; PSD, presynaptic dendrite; PV, paraventricular nucleus; Rt, reticular nucleus; SR, small terminal with round vesicles; TCR, thalamocortical relay nucleus; VB, ventrobasal complex; VL, ventrolateral nucleus.

Abbreuiarions:

961

the brain of guinea-pigs has been scarce so far, presumably because this species belongs to the family of caviomorphs, generally classified as a suborder of rodents. Therefore, most of the physiological results obtained from guinea-pig are thought to be comparable to those from the most common and less expensive myomorph rodents (rat and mouse). A recent immunocytochemical report by Asanuma’ showed, however, that in the ventrobasal complex (VB) of the thalamus of guinea-pigs, 10% of the neurons are GABAergic, thus suggesting a substantial difference from rat and mouse VB, where GABA-positive cells are less than 1%.2,7,‘2,29 If, as suggested by Caja1,4*5the increment of interneurons is a sign of species evolution, the increased number of GABAergic cells in VB of guinea-pig (but differences in other brain regions could also be expected) may indicate that this animal evolved differently from the myomorph rodents. In line with this hypothesis are the recent data from molecular biology suggesting that the guinea-pig belongs to a different taxonomic order and developed independently from myomorph and sciuromorph rodents.” This hypothesis would be further corroborated by the demonstration that the presence of

R. SPREAFICOet al.

962

GABAergic neurons although fewer than those found in carnivores and primates, modifies the intrinsic synaptic organization of VB in guineapigs compared to that in VB of rats. We performed light- and electron-microscopic immunocytochemical experiments using a specific anti-GABA serum on guinea-pig thalamus not only to confirm the lightmicroscopical data obtained by Asanuma’ but also to investigate the intrinsic synaptic organization of VB by comparing it to other thalamic nuclei that lack interneurons. Preliminary results have been presented in abstract form.45 EXPERIMENTAL

PROCEDURES

Five adult guinea-pigs (Hartley, Charles River), under deep anesthesia (chloral hydrate 4%; I ml/g body weight), were intracardially perfused with a solution of 2.5% glutaraldehyde and 0.5% paraformaldehyde in phosphate buffer. Brains were cut, using an Oxford Vibratome, in SO-pm-thick coronal (three animals) or horizontal (two animals) sections. Three thalami (one coronal and two horizontal) were processed only for light microscopic investigation. For the immunocytochemical procedure serial, alternate, sections through the thalamus were preincubated for 1 h in phosphate buffered saline (PBS) containing 10% normal goat serum (NGS), 0.2% Triton X-100 and 0.1% NaN,. After rinsing in PBS, sections were incubated for 12 h in a rabbit polyclonal anti-GABA serum (Sigma) diluted 1: 10,000 in PBS-NGS-Triton. After multiple rinses in PBS, sections were incubated for 1h in biotinylated goat anti-rabbit IgG (Vector) diluted 1: 200 in PBS-NGS-Triton, rinsed in PBS and incubated for 1 h in avidin-biotin-peroxidase complex (Vector) diluted 1: 100 in PBS. After rinsing in PBS and 0.1 M Tris-HCl buffer (pH 7.6) sections were reacted in a fresh solution of 0.05% diaminobenzidine tetrahydrochloride (Sigma) and 0.002% H,O, in Tris-HCl for 5510 min, washed in Tris-HCl buffer and then mounted on gelatin coated slides, dehydrated and coverslipped. Sections adjacent to those processed for immunocytochemistry (ICC) were counterstained with Thionin (1%) and used for cytoarchitectonic control. The distribution of GABAimmunoreactive (IR) neurons in the thalamus was charted using an x-y plotting computerized system connected to the microscope stage by means of transducers. The cytoarchitectonic boundaries of thalamic nuclei were identified on adjacent counterstained sections and drawn on the charting by means of a projection microscope. A quantitative evaluation of the number of GABA-IR neurons was performed on sampled section from VB, lateral geniculate (LG), and ventrolateral (VL) nuclei in two animals, by counting only labeled and unlabeled neurons with the nucleolus in the plane of the section under Nomarski optic at 40 x In the remaining two animals, Vibratome sections were osmicated, dehydrated in ethanols and embedded in EponSpurr resin. Blocks of about 1 mm2 were trimmed from VB, VL and paraventricular (PV) nuclei of the thalamus and glued to epoxy blanks. For collection of quantitative data, semithin sections (0.5-l pm thick) from three blocks, sampled from different areas of VB and VL, were cut on an ultramicrotome and collected on gelatinized slides. After removal of the plastic, sections were reacted for GABA ICC, using the same protocol described for Vibratome sections, and subsequently counterstained with Toluidine Blue (0.1%). Areas of both GABA-IR and non-IR neurons were calculated, by means of an appropriate imaging computer program, on 444 VB and 232 VL cells showing the nucleolus in the plane of the section. For electron microscopy, thin sections were cut from VB, VL and PV nuclei and either collected on copper

grids and counterstained with uranyl acetate and lead citrate, or collected on nickel grids and processed with a postembedding immunogold labeling procedure as previously described. 6,7 Briefly, after a 15-min immersion in saturated sodium metaperiodate, grids were incubated at room temperature with 10% NGS (30min) and then with the anti-GABA serum (1: 10,000; 1 h). After extensive rinses, grids were incubated in a solution of goat anti-rabbit immunoglobulins coupled to 15-nm gold particles (Biocell) diluted 1: 30 (1 h) and then counterstained with uranyl acetate and lead citrate. Thin sections were observed and photographed with a Jeol T8 or Zeiss 902 electron microscope. Immunocytochemical controls

Specificity of the immunolabeling was evaluated by blocking experiments in which the anti-GABA serum was absorbed with GABA, both in free and conjugated form. Method specificity was controlled by the application of rabbit non-immune serum as well as by the processing of a series of sections omitting various stages from the regular staining sequence. Preabsorption of the antiserum with GABA (free or conjugated) abolished labeling, as did substitution of primary antiserum with pre-immune serum. RESULTS

Light microscopy

The pattern of distribution of GABA-IR elements was similar in all animals. Within the thalamus GABA-IR neurons were not homogeneously distributed among the different nuclei. The main concentration of GABA-IR cells was found within the reticular thalamic nucleus (Rt). Although no counting of immunolabeled cells was performed, in sections processed for GABA ICC and counterstained with thionin all the neurons within Rt were labeled (Figs lA, B, 2). These cells were small or medium sized, with a round or fusiform perikaryon. Within Rt, numerous IR puncta, interpreted either as terminals (frequently close to IR cell bodies) or as sectioned fibers, were also present. At the medial border of Rt, GABA-IR fibers crossing the internal medullary lamina toward the lateral thalamic nuclei were also observed (Fig. IB). Within the dorsal thalamic nuclei, the main concentration of GABA-IR neurons was found in LG, where they represented 20% of the total population. These cells had a small diameter and round perikaryon, and were randomly distributed throughout the nucleus (Fig. 2C, D). In VB, small, round-shaped GABA-IR neurons were also present (Fig. IA-C). Although sometimes clustered, they were distributed throughout the two subdivisions, ventroposterolateral and ventroposteromedial of the nucleus (Fig. 2B-D). In Vibratome sections, GABA-IR cells accounted for 15-16% of the total population. The analysis of semithin, immunoreacted, sections revealed that 14.3% of VB neurons were GABA-IR and had a mean area of between 86.5 and 143.2 (f32.2) pm2, ranging 236.2 pm*. In contrast, the non-IR cells (87.7%) interpreted as projecting neurons, had a mean area of 267.1 (k67.4) pm*, ranging between 113.1 and 449.3 pm2. The mean area of the total neuronal

Interneurons in the Eminea-pig thalamus

Fig. 1. (A) Low power photomicrograph from a Thionin-stained horizontal section through the dorsal thalamus of guinea-pig. Scale bar = 300 pm. (B) Low power photomicrograph from a horizontal thalamic section immunoreacted for GABA visualization. Labeled neurons are present only in VB and Rt but not in VL. Note in addition bundles of GABA-IR fibers within VB and VL. Scale bar = 100pm. (C, D) High-power photomicrographs of VB (C) and VL (D) after GABA immunolabehng. In VB GABA-IR neurons and puncta are visible (C) while in VL (D) only immunoreactive puncta are present. Scale bars=40pm.

963

R.

SPREAFXO

Abbreviations

AD AM AV CL CM Hb LD LP MD mt

anterodorsal nucleus anteromedial nucleus anteroventral nucleus centrolateral nucleus centromedial nucleus habenular nucleus laterodorsal nucleus posterolateral nucleus mediodorsal nucleus mammillothalamic tract

et al.

used in Figure 2

PC PO Pt Re Rh

paracentral nucleus posterior group paratenial nucleus reuniens nucleus rhomboid nucleus ventromedial nucleus ventroposterolateral nucleus ventroposteromedial nucleus zona incerta

VM VPL VPM ZI

cells

m

nonOABA

ms (urn*)

ceils m

OiABA c&a

1

Fig. 3. Histograms showing the frequency distribution of cell soma areas (pm*) of GABA-IR and non-GABA-IR neurons in samples from VB and VL.

was 252.4 (k75.6) pm2. The histogram of the distribution of cell areas of GABA-IR and nonIR neurons demonstrated that the cell population of VB comprises two different groups, with an overlap for cells with areas around 200 grn2 (Fig. 3). GABAIR was very intense in the neuropil of VB. At high magnification numerous immunoreactive fibers and small- and medium-sized puncta were observed (Fig. 1C). These elements were particularly evident in semithin sections, where IR puncta were found in close contact with both GABA-IR and non-IR neurons and their dendrites (Fig. 4A). population

Fig. 2. Drawings from sampled serial coronal sections through the thalamus of guinea-pig processed for GABA immunolabeling. Dots indicate the distribution GABA positive neurons in different thalamic nuclei.

Interneurons in the guinea-pig thahmus

965

Fig. 4. Light-micrographs of semithin sections of guinea-pig thalamus immunostained with the antiGABA serum. (A) VB complex. Small GABA-IR neurons (arrows) are evident among large, unlabeled neurons (asterisks). The neuropil contains coarse GABA-IR fibers and several GABA-IR puncta, some of which are adjacent to cell bodies. Inset: Higher magnification of the boxed area showing a GABA-IR neuron (arrow) close to an unlabeled neuron (asterisk) surrounded by GABA-IR puncta (arrowheads). (B) VL nucleus. All the neurons are unlabeled (asterisks) and the neuropil contains GABA-IR puncta and fibers. Scale bars = 5 @cm(A, B); 1 pm (inset).

The reconstruction of serial Vibratome sections plotted throughout the rostrocaudal extent of the dorsal thalamus demonstrated that, in contrast to VB and LG nuclei, only few scattered GABA-IR

neurons, approximately l%, were present in VL and in the remaining thalamic nuclei (Figs 13, 2A-D). Also the neuropil of these nuclei was different from that observed in VB and LG. At high power, only

Fig. 5. Electron micrographs of guinea-pig thalamus. (A-E) VB complex. (A) Synaptic glomerulus formed by an LR terminal that synapses (arrows) on a dendrite (D) and on its protrusion (P), and also on an adjacent PSD bouton (asterisk); arrowheads point to puncta adherentia. G, glial ensheathment. An adjacent SR terminal contacts a small dendrite. (B) Triadic arrangement formed by an LR terminal that is presynaptic to a dendrite (D) and to a PSD bouton (asterisk), which in turn synapses on the same dendrite. The direction of the synapses is indicated by the arrows. G, glial ensheathment. (C) An F terminal synapses (arrows) on dendrites, also contacted by SR terminals. (D) An SR terminal contacts a PSD bouton (asterisk). (E) Serial synapses (arrows) formed by an SR terminal presynaptic to a PSD bouton (asterisk) that in turn contacts a small dendrite (D). (F) VL nucleus. An LR terminal contacts (arrow) a dendritic protrusion (P), and an F terminal synapses (arrowheads) on two dendrites. G, glial ensheathment. An adjacent SR terminal contacts a small dendrite. (G) PV nucleus. Varicosity (V) rich in large granular vesicles. Scale bars = 1 pm. Fig. 6. Electron micrographs of guinea-pig VB complex after immunogold labeling with anti&ABA serum. (A) An unlabeled LR terminal contacts (arrow) a GABA-IR PSD bouton (asterisk). An adjacent GABA-IR F terminal contacts a dendrite. SR terminals are unlabeled. (B) Serial synapses formed by a GABA-IR PSD bouton (2), which in turn synapses on an unlabeled dendritic protrusion (P). P is also post synaptic to an unlabeled LR terminal. The direction of the synapses is indicated by the arrows. G, glial ensheathment. (C) An unlabeled SR terminal contacts a GABA-IR PSD bouton (asterisk). An adjacent GABA-IR F terminal (F) synapses on a dendrite, Scale bars = 0.5 flrn.

966

R. SPREAFICOet al.

Interneurons

in the guinea-pig thalamus

Fig. 6

967

968

R.

SPREAlWO

et uf.

Fig. 7. Electron micrographs of guinea-pig thalamus after immunogold labeling with anti-GABA serum. (A, B) VB complex. (A) A GABA-TR F t~~ina~ contacts (arrow) a GABA-IR proximal dendrite. (B) GABA-IR F terminals contact (arrows) unlabeled, distal, dendrites. Adjacent SR terminals are unia~~ed. (C) VL r~4eus. A GABA-IR F terminal contacts (arrow) an unlabeled dendrite, an adjacent SK terminal is unla~l~. M, GABA-IR my~iinat~ fiber. (D) PV nucleus, Two CABA-IR F terminals contact (arrows) the same unlabeled dendrite. Adjacent SR terminals are uniabeled. Scale bars = 0.5 pm.

969

Intemeurons in the guinea-pig thalamus small immunoreactive puncta were present throughout the nuclei, without any appreciable difference among the different nuclei (Figs lD, 4B). A marked difference in neuropil staining was observed at the border between VB and VL, particularly in horizontal sections. Differences between VB and VL were also observed in semithin sections. In VL unlabeled neurons were frequently clustered in groups of seven to 10 neurons, while in VB they were randomly distributed. The analysis of semithin sections confirmed that GABA-IR cells in VL are less than 1% and showed differences in the neuropil staining. The mean area of the total neuronal population in VL was 251.8 (k60.1) pm’, and the distribution of the areas of the total neuronal population (GABA-IR and non-IR cells) in VL completely overlapped that in VB (Figs 3, 4). Electron microscopy Synaptic organization. In the neuropil of VB, four types of vesicle-containing profiles could be distinguished (Fig. SA-E). (1) Large terminals with round clear vesicles (LR terminals), occasional large granular vesicles (LGV), and containing many mitochondria (Fig. 5A). They were invaginated by one or more dendritic protrusions and made multiple, asymmetrical (Gray type 1) synaptic contacts. Non-synaptic filamentous contacts (puncta adherentia) were also frequently found between LR terminals and their postsynaptic targets (Fig. 5A). The LR terminals preferentially contacted proximal dendrites or their protrusions and were surrounded by an electronlucent astroglial sheath. Some of the LR terminals had a more complex arrangement as they also contacted, in the plane of the section, one or more vesicle-containing profiles interpreted as boutons of presynaptic dendrites (PSD, see below) also surrounded by the glial sheath (Fig. 5A). Triadic arrangements were also present, with the LR terminal being presynaptic to both the dendrite and the PSD, and the PSD being in turn presynaptic to the same dendrite (Fig. 5B). LR terminals were never postsynaptic to other structures and were located in the extraglomerular neuropil. (2) Small terminals with densely packed round clear vesicles (SR terminals) and few mitochondria (Fig. 5A, C). They established a single, asymmetric, synaptic contact preferentially with distal dendrites, but they were also involved in serial synaptic arrangements with PSD (Fig. 5D, E). The SR terminals were never postsynaptic to other structures. (3) Terminals heterogeneous in size, with pleomorphic or flattened clear vesicles (F terminals) and containing several large mitochondria (Fig. 5C). They made symmetrical synaptic contacts, often multiple, with dendrites of different caliber and also with cell bodies. (4) Medium size profiles with a pale cytoplasm containing pleomorphic vesicles, loosely packed, and scarce mitochondria (Fig. 5A, B, D, E). They established a single, symmetric, synaptic contact with dendrites, also contacted by LR terminals (Fig. 5B),

and occasionally ,with other similar profiles. In turn, they were contacted by LR (Fig. 5A, B) and SR (Fig. 5D, E) terminals. For their morphological characteristics, and in analogy to similar profiles identified in VB of other animal species’8~‘9*28*36 these profiles were interpreted as PSD boutons of local circuit neurons. VL and PV showed a much simpler organization, as their neuropil lacked synaptic glomeruli (Fig. 5F, G). Terminals similar to the LR and SR types described in VB could be detected, but they were never involved in complex synaptic arrangements (Fig. 5F). F terminals were also present, but no PSDs were found (Fig. 5F). LGVs were more numerous than in VB, and were present in several LR and F terminals, and also in profiles heterogeneous in size and shape lacking overt synaptic specializations (Fig. 5G). GABA immunogold labeling. The ultrastructural preservation of material prepared for postembedding immunolabeling was of similar quality to that of tissue processed with our standard procedure for electron microscopy, thus allowing the detection of all the above profiles (Figs 6, 7). GABA labeling was visualized by the presence of 15-nm gold particles over several neuronal profiles. Because the background labeling was very low in our preparations, it was not necessary to make statistical analysis of the distribution of gold particles for the identification of labeled profiles. Gold particles were scattered in the cytoplasm of labeled profiles and often concentrated over mitochondria, as already demonstrated in previous reports on other species.6*7s” In the neuropil of VB, GABA labeling was found in F terminals and PSD boutons, whereas LR and SR terminals were always unlabeled. GABA-IR F terminals contacted labeled and unlabeled dendrites of different size (Fig. 7A, B), and labeled and unlabeled cell bodies. GABA-IR PSD boutons received asymmetric synapses from LR (Fig. 6A) and SR (Fig. 6C) terminals and contacted unlabeled dendrites and other labeled PSD boutons (Fig. 6B). In the neuropil of VL (Fig 7C) and PV (Fig. 7D), GABA labeling was only found in F terminals involved in simple axo-dendritic and axo-somatic contacts with unlabeled profiles. LR and SR terminals were always unlabeled with profiles rich in LGV. In all the thalamic nuclei examined GABA labeling was also found in myelinated (Fig. 7B) and unmyelinated fibers.

DISCUSSION

Morphology and synaptic organization of GABAergic neurons

The present data confirm and extend the lightmicroscopic observation by Asanuma.’ At the light microscopic level, GABAergic cells were 15% of the total neuronal population in VB, whereas they were

970

R. SPREAFICO et al.

virtually absent (< 1%) in all the other dorsal thalamic nuclei (with the exception of LG, where approximately 20% of the cells were GABA-IR). In the guinea-pig, like in most mammals, all Rt neurons were GABA-IR.46 Since no differences were found in the mean areas of the total neuronal population between VB and VL, the presence of a population of small GABA-IR neurons in VB, but not in VL, raises the question of the origin of these neurons. Previous studies in the LG of cat49 suggested that interneurons and relay cells are born at the same time during neurogenesis. Although no conclusive data are available on the rest of the thalamus, the existence of two populations of neurons with similar morphology, but including one GABA-IR in VB but not in VL, would suggest that both groups of cells have the same origin and that only later, during migration or when they have reached their definitive location, they express their functional role and neurotransmitter content. The ultrastructural investigation showed remarkable differences in the synaptic organization between VB and the other nuclei examined (VL, PV). The distinguishing feature of VB is the presence of large aggregations of vesicle-containing profiles engaged in complex synaptic arrangements and showing a structural organization similar to that of the synaptic glomeruli in the dorsal thalamus of several mammals.‘&20,36Synaptic glomeruli are never detected in VL and PV of guinea-pigs; these contain only simple axodendritic or axosomatic synapses, like the dorsal thalamic nuclei of rat (with the exception of LG). The synaptic glomeruli of guinea-pig VB consist of: (i) LR terminals, morphologically similar to terminals identified as ascending (lemniscal, spinothalamic or spinocervicothalamic) excitatory afferents in different animal species+‘.‘6 and (ii) vesicle-containing profiles resembling the PSD boutons described in thalamic glomeruli of other mammals,24~3s and considered as interneurons.27,28~36 of GABAergic appendages Glomerular PSD boutons are post-synaptic to LR terminals and to other PSD boutons and contact conventional dendrites, often forming triadic arrangements with LR terminals. In the extraglomerular neuropil PSD boutons are also contacted by SR terminals, some of which were shown to be descending cortical afferents in other species.7~‘9~25~3b In the present investigation we demonstrate, by means of GABA immunogold labeling, that PSD boutons in guinea-pig VB are GABAergic, like those found in the same nucleus of cat and monkey,*’ and therefore they play an inhibitory role. Their origin from GABAergic interneurons is supported by the observation that they are absent in thalamic nuclei of guinea-pig lacking GABAergic cells. All the thalamic nuclei examined contain several GABA-IR F terminals that establish synaptic contacts with dendrites and cell bodies. They are similar to terminals that in other mammals are thought to derive mostly from the thalamic Rt nucleus,‘s2’ which

is known to exert a powerful inhibitory control over the dorsal thalamus.20~28,40 Functional

implications

Although GABAergic interneurons within the guinea-pig VB are less numerous than those observed in cat and monkey, ‘3,20,32,39.43 they are sufficient to rearrange the intrinsic organization of the nucleus by the formation of complex synaptic arrangements. It is therefore reasonable to assume that the physiological properties of VB in guinea-pig should be different from those in the adjacent VL and PV, and also from those in VB of rat and mouse,2,29,43as these nuclei lack interneurons. The two thalamic GABAergic systems (Rt and interneurons) are thought to play different roles. Rt receives a major input from collaterals of thalamocortical relay (TCR) cell axons and sends back, through its axons, an inhibitory input on to TCR neurons. Therefore, it takes part in a negative feedback system that controls the transmission of information from thalamus to cortex and may also be responsible for selective attention.2’,47.48In contrast, interneurons have been implicated in mechanisms that change the quality, rather than the quantity, of information transmitted to the cortex.42 A large proportion of the contacts established by interneurons is, in fact, via vesicle-filled dendritic appendages called PSD boutons. The individual PSD boutons can act as separate synaptic units, independent of the other activities of the parent neuron, and they probably release the synaptic transmitter in a graded fashion as a result of local synaptic input. 35 PSD boutons are ideally placed to change the character of the synaptic input mediated by ascending afferents, since they are postsynaptic to the TCR neuron dendrite. This arrangement (synaptic triad) is thought to result in a feed-forward inhibition, as the initial excitatory signal conveyed by the ascending afferent is rapidly converted to an inhibitory one through the GABAergic PSD bouton. Unfortunately only few electrophysiological data are available on the guinea-pig thalamus, mainly from in vitro studies.‘S~16~26~50 The relative low percentage of VB interneurons and their small diameter may explain the difficulties in recording from these cells, and the consequent homogeneity of the data reported in studies with intracellular recordings. Recent data from in viuo extracellular recordings in guinea-pig VB showed a small percentage of inhibitory responses after cortical or peripheral receptive field stimulation.37 The same study demonstrated that most of the recorded relay neurons showed a long lasting after burst inhibition. The authors suggested that regulation of somatosensory input and response pattern in VB could be mediated by a distinct modulatory corticofugal control. Although this hypothesis cannot be ruled out, our results suggest that, within VB, GABAergic interneurons represent the morphological basis for an intrinsic modulatory mechanism.

971

Interneurons in the guinea-pig thalamus Phylogenetic

considerations

The presence of complex synaptic arrangements of GABAergic terminals from both Rt and interneurons demonstrates a high degree of functional development in guinea-pig VB, similar to that in carnivores and primates. On the other hand, the simple synaptic organization of VL and PV is similar to that observed throughout the rat dorsal thalamus’ with the exception of LG. The typical synaptic organization of carnivores or primates and of rats are therefore simultaneously present in the guinea-pig dorsal thalamus, making it a good model for anatomical and physiological studies. The other two thalamic nuclei that contain GABAergic cells, i.e. Rt and LG, seem to have similar synaptic organization in all mammals. These two nuclei are highly conserved during phylogenesis and contain GABAergic neurons also in animals, such as the opossum, retaining characteristics of the early stages of mammalian evolution.33,46 It is generally believed that the guinea-pig belongs to the order of rodents, but a recent report, based on data from molecular biology, questioned this classification.” The analysis of amino acid sequences from species in different phylogenetic orders suggests that the guinea-

pig represents a separate evolutionary lineage from myomorph rodents (rat and mouse). The present study, showing a difference in the intrinsic organization of VB, supports the idea that the guinea-pig should be classified rat and mouse.

in a different

order from that of

This hypothesis would be in line with the Cajal proposa1415that the number of local circuit neurons increases in the course of evolution of mammalian

brain. This idea has been supported by other authors’4*” and more recently discussed by Penny et a1.33 and Spreafico et aI.& on the basis of immunocytochemical data These latter works, in agreement with the present results, provided evidence that the number and distribution, as well as the size of GABAergic neurons change in different species and that size differences are related to phylogeny. Since GABAergic neurons are recognized as an important element for the regulation of intrinsic mechanisms within a neuronal structure, the increased number of local circuit neurons within the guinea-pig VB should be regarded as a piece of mosaic representing the great behavioral flexibility of advanced mamma1s.33

CONCLUSIONS

The present data suggest that GABAergic neurons in the ventrobasal complex of guinea-pigs give rise to functionally important rearrangements of its intrinsic synaptic organization and that they represent the morphological basis for an intrinsic modulatory mechanism that is absent in other thalamic nuclei lacking inhibitory interneurons. Moreover the presence of GABAergic interneurons makes the ventrobasal complex of guinea-pigs similar to that of lagomorphs (rabbit), carnivores (cat) and primates (monkey). Acknowledgements-The authors are grateful to Ms. M. Denegri for typing the manuscript. This work was partially supported by NIH grant NS27827 (SDB), MURST 40% and by the H. de Jur Foundation.

REFERENCES

1. Asanuma C. (1991) The distribution of local circuit neurons and GABAergic boutons within the guinea-pig thalamus. Sot. Neurosci. Abstr. 17, 625. 2. Barbaresi P., Spreafico R., Frassoni C. and Rustioni A. (1986) GABAergic neurons are present in the dorsal column nuclei but not in the ventroposterior complex of rats. Brain Res. 382, 305-326. 3. Blomqvist A., Flink R. Westman J. and Wiberg M. (1985) Synaptic terminals in the ventroposterolateral nucleus of the thalamus from neurons in the dorsal column and lateral cervical nuclei: an electron microscopic study in the cat. J. Neurocytol. 14, 869686. 4. Cajal S. R. (1909) Histologic du Sist&me Nerveux de I’homme et des VPrtebrPs. Maloine, Paris. 5. Cajal S. R. (1966) Studies on the diencephalon (transl. Ramon-Moliner E.). Charles C. Thomas, Springfield. 6. De Biasi S., Frassoni C. and Spreafico R. (1986) GABA immunoreactivity in the thalamic reticular nucleus of the rat: a light and electron microscopic study. Bruin Res. 399, 143-147. 7. De Biasi S., Frassoni C. and Spmafico R. (1988) The intrinsic organization of the ventroposterolateral nucleus of the rat: a double labeling ultrastructural investigation with gamma-amino butyric acid immunogold staining and lectin conjugated horseradish peroxidase. Somatosensory Res. 5, 187-203. 8. de Curtis M., Par6 D. and Llinis R. (1991) The electrophysiology of the olfactory-hippccampal circuit in the isolated and perfused adult mammalian brain in oitro. Hippocampus 1, 341-354. 9. Dingledine R. (1984) Brain slices. Plenum Press, New York. IO. Giuffrida R., Aicardi G., Sapienza S. and Rapisarda C. (1989) Studio topografico e immunocitochimico delle proiezioni corticotalamiche e corticorubrali nella cavia. Proc. XLI Congr. Naz. Sot. It. Fisiol. 176. 11. Graur D., Hide W. A. and Li W.-H. (1991) Is the guinea-pig a rodent? Nature 351, 649652. 12. Houser C. R., Vaughn J. E., Barber R. P. and Roberts E. (1980) GABA neurons are the major cell type of the nucleus reticularis thalami. Brain Res. 200, 341-354. 13. Hunt C. A., Pang D. Z. and Jones E. G. (1991) Distribution and density of GABA cells in intralaminar and adjacent nuclei of monkey thalamus. Neuroscience 43, 185-196.

972

R. SPREAFICO et al.

14. Jacobson M. (1975) Development and evolution of type II neurons: conjectures a century after Golgi. In Golgi Centennial symposium Proceedings (ed. Santini M.), pp. 147-151. Raven Press, New York. 15. Johansen H. and LlinPs R. (1984) Blectrophysioiogical properties of guinea-pig thalamic neurons: an in vitro study. J. Physiol. 349, 205-226. 16. Johansen H. and Llinks R. (1984) Ionic basis for the el~trores~n~veness and oscillatory properties of guinea-pig thalamic neurons in vitro. J. Physiol. 349, 221-241, 17. Jones D. G., Dittmer M. M. and Reading L. C. (1974) Synaptogenesis in guinea-pig cerebral cortex: a glutaraldehydePTA study. Brain Res. 70, 245-259. 18. Jones E. G. and Powell T. P. S. (1969) Electron microscopy of synaptic glomeruli in the thalamic relay nuclei of the thalamus. Proc. R. Sec. Lond. B. 172, 153-171. 19. Jones E. G. and Powell T. P. S. (1969) An electron microscopic study of the mode of termination of cortico-thalamic fibres within the sensory relay nuclei of the thahunus. Proc. R. SOC. Load. B. 172, 173-185. 20. Jones E. G. (198.5) The Thawed. Plenum Press, New York. 21. Jones E. G. (1988) Modern views of cellular thalamic mechanisms. In Cellular Thaiamic Mechanisms (eds Bentivogho M. and Spreafico R.), pp. l-22. Elsevier, New York. 22. Kolston J., Gsen K. K., Hackney C. M., Ottersen 0. P. and Storm-Mathisen J. (1992) An atlas of glycine- and GABA-like immunoreactivity and colocalization in the cochlear nuclear complex of the guinea-pig. Anat. Embryol. 186,443~65. 23. Kumoi K., Saito N. and Tanaka C. (1987) Immunohistochemical localization of g~ma-aminobutyric acid- and asp~tate~n~ining neurons in the guinea-pig vestibular nuclei. Brain Res. 416, 22-33. 24. Lieberman A. R. and Webster K. E (1972) Presynaptic dendrites and a distinctive class of synaptic vesicle in the rat dorsal lateral geniculate nucleus. Bruin Res. 42, 196-200. 25. McAllister J. P. and Wells J. (1981) The structural organization of the ventral posterolateral nucleus in the rat. J. camp. Neurol. 197, 271-301. 26. McCormick D. and Wang Z. (1989) Serotonin and noradrenaline excite GABAergic neurones of the guinea-pig and cat nucleus reticularis thalami. J. Neurophysiol. 442, 235-255. 27. Ohara P. T., Lieberman A. R., Hunt S. P. and Wu J.-Y. (1983) Neural elements cont~ning glutamic acid decarboxylase (GAD) in the dorsal lateral genicuiate nucleus of the rat: immunohist~hemi~l studies by light and electron microscopy. Neuroscience 8, 189-211.

28. Ohara P. T., Chazal G. and Ralston H. J. III (1989) Ultrastructural analysis of GABA-immunoreactive elements in the monkey thalamic ventrobasal complex. J. camp. Neurol. 283, 541-558. 29. Ottersen 0. P. and Storm-Mathisen J. (1984) GABAtontaining neurons in the thalamus and pretectum of the rodent. Anat. Embryol. 170, 197-207. 30. Ottersen 0. P. and Storm-Math&en J. (1985) Different neuronal localization of aspartate-like and glutamate-like immuno~ctiviti~ in the ~p~rnpus of rat, guinea-pig and senegalese baboon (papio papio), with a note on the dist~bution of ~a-~nobutyrate. Neuroscience 16, 589-606. 31. Pare D., de Curtis M. and Llinas R. (1992) Role of the hippocampal-entorhinal loop in temporal lobe epilepsy: extra- and intracellular study in the isolated guinea-pig brain. J. Neurosci. 12, 1867-1881. 32. Penny G. R., Fitzpatrick D., Schmechel D. E. and Diamond I. T. (1983) Glutamic acid decarboxylase-immunoreactive neurons and horseradish peroxidase-labeled projection neurons in the ventral posterior nucleus of the cat and Galago senegtalensis. J. Neurosci. 3, 1868-1887. 33. Penny G. R., Conley M., Schmechel D, E. and Diamond I. T. (1984) The distribution of glutamic acid decarboxylase immuno~ti~ty in the ~en~phalon of the opossum and rabbit. J. camp. Neurol. 223, 38-56. 34. Rakic P. (1976) Local Circuit Neurons. MIT Press, Cambridge. 35. Ralston H. J. III (1971) Evidence for presynaptic dendrites and a proposal for their mechanism of action. Nature 230, 585-587.

36. Ralston H. J. III, Ohara P. T., Ralston D. D. and Chaza.l G. (1988) The neuronal and synaptic organization of the cat and primate somatosensory thalamus. In CeBrdar Thulamic Mechanisms (eds Bentivoglio M. and Spreafico R.), pp. 127-141. Elsevier, New York. neurons relaying 37. Rapisarda C., Palmeri A. and Sapienxa S. (1992) Cortical modulation of thal~o~orti~l exteroceptive info~ation: a ~~ostimulation study in guinea-pig. ExpI Bruin Res. 88, 140-150. 38. Richards C. N. and Sercombe R. (1970) Calcium, magnesium and the electrical activity of guinea-pig olfactory cortex in vitro. J. Physiol. 211, 511-581. 39. Rinvik E., Ottersen 0. P. and Storm-Math&n J. (1987) Gamma-aminobutyrate-like immunoreactivity in the thalamus of the cat. Neuroscience 21, 781-805. 40. Rustioni A., Battaglia G., De Biasi S. and Giuffrida R. (1988) Neuromediators in somatosensory thalamus: an immunocytochemical overview. In Cellular Thaiamic Mechantsms (eds Bentivoglio M. and Spreatico R.), pp. 271-296. Elsevier, New York. 41. Schofield B. R. and Cant N. B. (1991) Organization of the superior olivary complex in guinea-pig. I. Cyto~chi~ture, cytochrome oxidase histochemistry, and dendritic morphology. J. camp. Nemo/. 314, 645-670. 42. Sherman S. M. and Koch C. (1986) The control of retinogeniculate transmission in the mammalian lateral geniculate nucleus. Expl Brain Res. 63, l-20. 43. Spreafmo R., Schmechel D. E., Ellis L. C. and Rustioni A. (1983) Cortical relay neurons and intemeurons in the nucleus ventralis posterolateralis of cats: a horseradish peroxidase, electron-microscopic, Golgi and immunocytochemical study. Neuroscience 9,491-509. neurons in the 44. Sprea&o R., De Biasi S., Battaglia G. and Rustioni A. (1992) GABA- and ~u~a~~n~~i~ tfialamus of rats and cats: an immun~~hemi~l study. In N~rotr~mitters in Epilepsy (eds Avanzini G., Bugelyx Fariello R. and Heinemann U.), pp. 107-l 1.5. Epilepsy Res. suppl. 8. Ekevier, Amsterdam. 45. Spreafico R., Frassoni C., de Curtis M. and De Biasi S. (1992) Ultrastructural investigation on the dorsal thalamus of guinea-pig. Sot. Neurosci. Abstr. lg, 1018. 46. Spreafico R., Frassoni C., Regondi M. C., Arcelli P. and De Biasi S. (1993) Interneurons in the mammalian thalamus: a marker of species? In Thakmic Networks@ Relay and Modulution (eds Minciacchi D., Molinari M., Ma&i 0. and Jones E. G.), pp. 17-28. Pergamon Press, Oxford.

Interneurons in the guinea-pig thalamus

913

47. Steriade M. and Deschenes M. (1984) The thalamus as a neuronal oscillator. Bruin Rex Rev. 8, 163. 48. Steriade M. and Llinas R. R. (1988) The functional states of the thalamus and the associated neuronal interplay. Physiol. Rev. 63, 649-142. 49. Weber A. J., Kalil R. E. and Hickey T. L. (1986) Genesis of interneurons in the dorsal lateral geniculate nucleus of the cat. J. camp. Neural. 252, 385-391. 50. Yamamoto C. and Sawada S. (1981) The amount of transmitter release from optic nerve terminal in thin slice from the lateral geniculate nucleus in vitro. In Electrophysiology of isolated mammalian CNS preparations (eds Karkut

G. A. and Wheal H. V.), pp. 233-256. Academic Press, New York. (Accepted 4 October 1993)