THE JOURNAL OF COMPARATIVE NEUROLOGY 297~582-593(1990)

Thalamic Midline Cell Populations Projecting to the Nucleus Accumbens, Amygdala, and Hippocampus in the Rat HONG-SEN SU AND MARINA BENTIVOGLIO Institute of Anatomy, University of Verona, Italy

ABSTRACT The organization of the thalamic midline efferents to the amygdaloid complex, hippocampal formation, and nucleus accumbens was investigated in the rat by means of multiple retrograde fluorescent tracing. The present findings indicate that these connections derive from separate cell populations of the thalamic midline, with a low degree of divergent collateralization upon more than one of the targets examined. The neural populations projecting to the amygdala, hippocampus, or accumbens are highly intermingled throughout the thalamic midline, but display some topographical prevalence. Midline thalamo-hippocampal cells are concentrated in the nucleus reuniens; thalamoaccumbens neurons prevail in the ventral portion of the paraventricular nucleus, and in the central medial nucleus. Thalamo-amygdaloid cells display a topographical prevalence in the rostra1 third of the thalamic midline and are concentrated in the dorsal part of the paraventricular nucleus and in the medial part of the nucleus reuniens. Both dorsally in the paraventricular nucleus and ventrally in the nucleus reuniens, thalamo-amygdaloid cells are located closer to the ependymal lining than the neurons projecting to the hippocampus or nucleus accumbens. Further, thalamo-amygdaloid cells, especially in the paraventricular nucleus, extend their dendritic processes in the vicinity of the ependymal lining, where they arborize profusely. These features indicate a close topographical relationship of neurons projecting to the amygdala with ependymal cells. The fairly discrete origin of midline outputs to the amygdala, hippocampus, and accumbens indicates that the flow of information is conveyed through separate channels from the thalamic midline to limbic and limbic-related targets. Together with the literature on the limbic afferents to the thalamus, these findings emphasize the relationships between the thalamus and the limbic system subserved by parallel input-output routes. However, because of the overlap of the projection cell populations, the thalamic midline may represent a locus of interaction among neurons connected with different parts of the limbic system. The functional implications of these findings are discussed in relation to the “nonspecific” thalamic system, as well as to the circuits involved in memory formation. Key words: thalamus, limbic system, basal ganglia,parallel channels, memory processing

The thalamic output to the limbic system has received much less attention than the thalamo-neocortical connection. Early reports based on anterograde degeneration tracing (Nauta and Whitlock, ’54) indicated that thalamic projections to limbic cortical fields originated from medial “nonspecific” nuclear territories. More recently, axonal tracing studies on the subcortical afferents to amygdala and hippocampus have mentioned projections of the thalamic midline upon these structures in different mammalian species (Segal, ’77; Beckstead, ’78; Herkenham, ’78; Veening, ’78; Baisden and Hoover, ’79; Ottersen and Ben Ari, ’79; Wyss et al., ’79; Amaral and Cowan, ’80; Mehler, ’80; Aggleton et al., ’80; Riley and Moore, ’81; Russchen, ’82; o 1 9 9 0 WILEY-LISS, INC.

Yanagihara et al., ’87),indicating that the midline nuclei are a rich source of thalamo-limbic connections. The thalamic midline also projects to the striatum. Midline efferents are densely distributed upon the nucleus accumbens (Groenewegen et al., ’80; Newman and Winans, ’80; Beckstead, ’84; Kelly and Stinus, ’84; Jayaraman, ’85; Phillipson and Griffiths, ’85; Berendse et al., ’88), which is Accepted February 19,1990, Address reprint requests to Dr. M. Bentivoglio, Institute of Anatomy, Medical Faculty, Strada Le Grazie, 37134 Verona, Italy. H.-S. Su was on leave from the Department of Anatomy, Hunan Medical University, Changsha,People’s Republic of China.

MIDLINE THALAMOLIMBIC PROJECTIONS the part of the striatum receiving dense afferents from limbic structures, and in particular the amygdala and hippocampus, rather than from the neocortex (Carman et al., '63; De Olmos and Ingram, '72; Swanson and Cowan, '77; Kelley and Domesick, '82; Groenewegen et al., '87). These data suggest that limbic and limbic-related structures receive innervation from the thalamic midline. However, detailed studies of the organization of these connections are still lacking. The present investigation was aimed at verifying, in the rat, the interrelationships among thalamic midline cells projecting upon the amygdala, hippocampal formation, and nucleus accumbens, as well as the eventual occurrence of thalamic branched projections to these targets. This study was based on the simultaneous visualization of two or three retrograde tracers.

MATERIALS AND METHODS Fourteen albino rats were used in the experiments. Multiple tracing experiments were performed on 8 of these animals; in the other 6 animals single tracing experiments were devoted to analyzing the morphological aspects of labeled cells. The tracers, fast blue, diamidino yellow dihydrochloride, true blue (FB, DY, TB; Illing) and fluoro-gold (FG; Fluorochrome, Inc.) were used for double or triple retrograde labeling. The tracers were dissolved or suspended in distilled water (2% FB, 2% DY, 3% TB, 2% FG) and injected stereotaxically with Hamilton microsyringes, under barbiturate anesthesia (Nembutal 40 m g h g i.p.). Injections were made into the nucleus accumbens, the amygdaloid complex, and the hippocampal formation in different combinations (see Table 1for the injected volumes). In order to obtain a complete retrograde filling of thalamic cells and to analyze their morphology, single injections of FG were performed in 4 animals (VR41, D29, VR46, VR51) and a longer survival time than in the previous cases was adopted (see Table 1). In these animals colchicine (80 pg/20 jd) was injected stereotaxically in the lateral cerebral ventricle under deep anesthesia 24 hours before sacrifice, since this pretreatment had been found to enhance the fluorescence and the dendritic filling of FG labeled neurons

A bbrevia tiom

Ac AM BL CeM CP DY FB FG fr IAM

L LHb MD MHb mt PT PV PVA PVP Re Rh sm TB

nucleus accumbens anteromedial n. basolateral amygdaloid n. central medial n. caudoputamen diamidino yellow fast blue fluoro-gold fasciculus retroflexus interanteromedial n. lateral amygdaloid n. lateral habenular n. mediodorsal n. medial habenular n. mammillothalamic tract parataenial n. paraventricular n. anterior paraventricular n. posterior paraventricular n. nucleus reuniens rhomboid nucleus stria medullaris true blue

583 TABLE 1. Experimental Parametera Inj. in the Accumbens

vR36 VR41 D29

2%DY0.2~1 2%DY 0 . 1 ~ 1 2% D Y 0 . 3 4 2% F G 0 . 2 4 2%DY0,15d 3%TB0.15d 2% DY 0.2 rl 2% FG 0.1 crl 2%FG0.15d 2%FG0.3~1

Amygdala 2% F B 0 . 3 d 2% TB0.1111 2% FG0.2d 2%TB0.2d 3%TBO.I5d 2XFGO.15d 3%TB0.25111 2%DY 0.24

Inj.in the Hippocampus

2%DY0.3rl 2%FG02pl 2%DY0.25~1 2%DY0.35rl 3% TBO.25d

2% FG 0.2 rrl

vR46 VR51 D14 D22

Inj. in the

2%FG 0.45PI

20%HRp0.2pl 20% HRP 0.2 rl

survival Time (Days) 5 5 4 4 3 4 4 4 13 9 I 11 2 2

(H.S. Su, unpublished observations). Finally, horseradish peroxidase (HRP, Boehringer, 20% aqueous solution) was injected under deep anesthesia in 2 cases (D14, D22), with the aim of pursuing an ultrastructural investigation of thalamo-amygdaloid cells. The latter cases were used in the present study for the topographical and morphological analysis of the cell population projecting to the amygdala. After the survival times indicated in Table 1, the animals injected with fluorescent tracers were deeply anesthetized and perfused through the ascending aorta with saline followed by 10% buffered formalin (pH 7.2). The brains were removed, soaked in 30% buffered sucrose, and cut on a freezing microtome into 40-hm-thick serial sections in the frontal plane. Every third section was mounted on slides from distilled water, air dried, and coverslipped with Entellan without dehydration. Adjacent sections were counterstained with cresyl violet. The unstained sections were studied with a Leitz Ploempack fluorescence microscope using filter mirror systems A and D (360 nm and 390 nm excitation wavelength, respectively). The tracers FB or TB, DY, and FG can be visualized simultaneously under the same light excitation wavelengths and can be easily distinguished on the basis of their different colour and cell labeling features: T B and FB result in blue labeling of the cytoplasm, DY in yellow-green labeling of the nucleus, and FG in silver granular cytoplasmic labeling. The 2 cases injected with HRP were transcardially perfused under deep anesthesia with saline followed by 2.5% glutaraldehyde and 1% paraformaldehyde in phosphate buffer (pH 7.2). The brains were immediately cut on a Vibratome into coronal sections. Every third section, 50 pm thick, was processed for HRP visualization using tetramethylbenzidine as a chromogen (Mesulam, '78); adjacent sections were counterstained with cresyl violet. The extent of the injection areas and the distribution of retrogradely labeled cells in the thalamic midline nuclei were charted with an X-Y plotter attached to the microscope stage by means of transducers. Cytoarchitectonic controls of the location of the injection areas and distribution of labeled cells were made on the counterstained sections; the boundaries of the thalamic nuclear territories, as seen after counterstaining, were superimposed upon the charts of labeled cells with the aid of a projection microscope.

RESULTS In the cytoarchitectonic organization of the midline nuclear group, we distinguished a dorsal and a ventral subdivi-

504

sion, following Jones ('85) and Faull and Mehler ('85). The dorsal component included the paraventricular (PV) nucleus, and at anterior levels the parataenial (PT) nucleus, lateral to PV, and the interanteromedial (IAM) nucleus, ventral to PV. The reuniens (Re) and rhomboid (Rh) nuclei were recognized as the main structures of the ventral component. The dorsal and ventral sectors of the midline nuclear group appeared divided by the fibers of the internal medullary lamina, in which the central medial (CeM) nucleus is embedded. On the basis of its connectivity and its relationships with the internal medullary fibers, CeM can be considered part of the anterior intralaminar system (see Macchi and Bentivoglio, '86). However, due to its location a t the thalamic midline, CeM is considered here with the midline nuclear group. In the dorsal portion of the midline, a distinct border between PV and the intermediodorsal (IMD) nucleus, as well as between the latter and CeM, was often difficult to determine; thus IMD and PV are not considered here as separate nuclear entities. Deliberately large volumes of dye were injected in each of the targeted areas, including as much of the target structure as possible, in order to optimize the number of retrogradely labeled neurons. Drawings of representative examples of injection sites in experiments VR35 and VR22 are shown (Figs. 1, 2) for purposes of illustrating the location and extent of injection areas, revealing that the hippocampal formation, amygdaloid complex, and nucleus accumbens can be extensively injected without likely overlap. As shown in Figures 1and 2, the tracer injections involved mainly the ventral portion of the hippocampal formation and did not involve its rostra1 sectors contiguous with the amygdala, nor its most caudal-dorsal region. As a consequence, the hippocampus-projecting cells are mostly underestimated in the present study. In one experiment (case VR43), the injections in the hippocampal formation were also targeted to its dorsal region and involved the entorhinal cortex. Injections in the amygdaloid complex were centered in the basal nuclear group and involved the adjacent amyg daloid nuclear structures; in some cases the injected dye diffused ventrally into the cortical amygdaloid nuclei. The injections into the nucleus accumbens involved both its medial and lateral portions. Thus the amygdala and accumbens injections may have involved contiguous regions, but the relative accuracy of dye distribution principally within targeted structures (Figs. 1, 2) minimizes the likelihood of significant diffusion beyond the limits of the target. The spread of tracers along the needle tracks was minimal in all cases. All of the tracer injections consistently resulted in retrograde labeling of neurons in the thalamic midline nuclei. Labeled cells were located ipsilaterally to the tracer injections. A few neurons were also scattered on the contralateral side in the same structures labeled ipsilaterally. Contralatera1 labeled cells were observed especially after injections into the nucleus accumbens and amygdaloid complex. For the sake of clarity, the overall distribution of the labeled cell populations in the midline nuclei are outlined here before describing their interrelationships.

Distributionof the labeled cell populations The location of labeled cells is shown in the representative experiments VR35 and VR22 (Figs. 1,2) and is summarized in Figure 5. The cells labeled from the amygdala were located throughout the rostrocaudal extent of the midline nuclei, with an

H.-S. SU AND M. BENTIVOGLIO anterior prevalence and a medial distribution. Rostrally, neurons labeled from the amygdala were found throughout the anterior paraventricular (PVA) nucleus and prevailed in the dorsal portion of this structure. Labeled neurons were also observed in PT, especially in its dorsomedial portion, in IAM, and in the medial part of Re. Proceeding posteriorly, the thalamo-amygdaloid cell population appeared concentrated in two sectors of the midline: the most dorsal part of PV, including its posterior pole (PVP), and the most medial and ventral part of Re. Amygdala-projecting cells were also seen throughout the anteroposterior extent of CeM and Rh. Thus a medial location was a characteristic and consistent feature of the cells retrogradely labeled from amygdaloid injections. Further, both dorsally in PV and ventrally in Re, thalamo-amygdaloid cells were consistently located in proximity to the ependymal lining, closer to ependymal cells than the other 2 labeled neuronal populations. The neurons labeled from the injections in the hippocampal formation were concentrated in Re, throughout its anteroposterior extent, and predominated in the lateral part of the nucleus. Thalamo-hippocampal neurons were also scattered in PT, with a ventrolateral prevalence, as well as in PV and Rh. Injections in the nucleus accumbens resulted in dense filling of thalamic midline structures. The labeled neurons were concentrated in PV, where they prevailed in the ventral sector of the nucleus throughout its extent, as well as in CeM. Cells labeled from the nucleus accumbens were also seen in PT, Rh, and in the medial portion of Re.

Interrelations among labeled cell populations and cell morphology As shown in cases VR35 and VR22 (Figs. 1, 2), the vast majority of thalamic midline cells were single labeled in multiple tracing experiments. Double-labeled cells represented less than 5 % of the labeled neurons and occurred in the overlap area of 2 labeled cell populations. Thus a few cells double labeled from injections in the amygdala and accumbens were seen in PV (Fig. 2). Double-labeled neurons were detected only occasionally in PT, IAM, Rh, and CeM. Cells double labeled from both accumbens and hippocampus, or from both amygdala and hippocampus, were observed in Re. Double-labeled neurons were more numerous in Re in case VR43 than in the other ones. In the latter experiment, in which the FG injection involved dorsal and ventral portions of the hippocampal formation and the entorhinal cortex, about 10% of the Re cell population labeled from the FG injection was double labeled from the injection in the nucleus accumbens, and about 4% was double labeled from the injection in the amygdala. No triple labeled neurons were observed. Paraventricular nucleus. The 3 labeled cell populations were tightly intermingled in PV; however, the PV neurons labeled from either the amygdala or hippocampal formation or nucleus accumbens displayed quantitative as well as topographic differences (Figs. 1,2). The cells labeled from the nucleus accumbens or the amygdala were distributed throughout the anteroposterior extent of PV and were very numerous. The neurons labeled from the hippocampal injections were mainly seen in the anterior part of PV and were less numerous in PV than in the other 2 cell populations. These findings were consistently verified in all cases, including VR43, in which the

'. .:..E./--.\ . *

an

S N O I L 3 3 P 0 8 6 318EUI?O&UV?VHL 3NI?(IIW

989

H.-S. SU AND M. BENTIVOGLIO

586

Fig. 2. Schematic representation of the injection sites and distribution of thalamic midline retrogradely labeled cells after tracer injections in the nucleus accumbens and amygdaloid complex in the doublelabeling experiment VR22. See legend to Figure 1for further explanations.

located dorsally and appeared mainly round, with a bipolar dendritic arborization (Fig. 3B).In the most medial portion of PV,the long axis of labeled cells was oriented vertically, parallel to the midline. A peculiar orientation of cell bodies and dendrites was evident in the most dorsal part of PV, which, as stated above, contained mainly cells labeled from the amygdala. In the most dorsal portion of PV,especially in PVA, the long axis of thalamo-amygdaloid neurons was oriented parallel to the border between the ependyma and the cerebrospinal fluid (CSF) (Figs. 3A, 4A) and their dendrites ran toward the ependyma. These dendritic branches arborized profusely and formed a dense plexus close to the ependymal cells (Fig. 4); small dendritic protrusions were detected occasionally very close to the ependymal surface (Fig. 4B). Parataenial nucleus. As shown in cases VR35 and VR22 (Figs. 1, 2), 3 single-labeled cell populations were intermingled in PT. The neurons retrogradely labeled from

the nucleus accumbens were mainly scattered in the ventrolateral portion of PT, whereas those labeled from the amygdala prevailed in the dorsomedial portion of the nucleus. Topographic segregation of the cells labeled from the hippocampus was not evident in PT. The somata of PT labeled cells were slightly larger (14--26 pm the longest diameter) than those observed in PV, and

Fig. 3. Photomicrographs of thalamic midline cells, retrogradely labeled from injections of the tracer fluoro-gold into the amygdaloid complex (A and D), or nucleus accumbens (B and C). A. Dorsal portion of the paraventricular nucleus; note the density of thalamo-amygdaloid neurons close to the ependymal lining (indicated by the interrupted line). B. Ventral part of the paraventricular nucleus; note the bipolar organization of dendrites of labeled neurons. Retrogradely labeled cells of the rhomboid nucleus are shown in C, and of the parataenial nucleus in D. Calibration bars: 20 pm in A, C, D; 10 pm in B.

Figure 3

H.-S. SU AND M. BENTIVOGLIO

588

they were multipolar or round. With a few arborizing branches, 3-5 dendrites originated radially from the cell bodies (Fig. 3D) displaying occasional bulbous protrusions and dilatations. Interanteromedial, rhomboid, and central medial nuclei. The 3 labeled cell populations were highly intermingled in IAM and Rh. However, in Rh the cells labeled from the accumbens were consistently more numerous than those labeled from the amygdala or hippocampus. The neurons labeled from the accumbens were mainly located in the anterior portion of Rh, whereas those labeled from the hippocampus were predominantly posterior. Labeled IAM or Rh cells were small- or medium-size (Fig. 3C) and their longest diameter ranged from 1 2 Fm to 20 pm. In the dorsal part of Rh, the labeled dendrites were mainly directed transversely, whereas they displayed a vertical orientation in the ventral portion of the nucleus. The neurons labeled from the accumbens were densely grouped in CeM, which also contained some cells labeled from the amygdala. The labeled CeM cells displayed mainly a fusiform shape and were oriented transversely, along the fibers of the internal medullary lamina. Nucleus reuniens. The 3 different labeled cell populations were detected throughout the anteroposterior extent of Re; the neurons labeled from the hippocampus in Re were more numerous than those labeled from amygdala or accumhens. A medial prevalence of the 2 latter cell populations and a lateral prevalence of that labeled from the hippocampus were consistently observed, but in case VR43 the neurons labeled from the hippocampal injection were also numerous in the medial portion of Re. Since in the latter case the hippocampal injection also involved the entorhinal cortex, this finding suggests that the medial part of Re could give rise to entorhinal afferents. Labeled neurons were mainly small-size (12-15 Fm), ovoid-shape, with a rather profuse dendritic arborization oriented both transversely and vertically.

DISCUSSION The midline mosaic of thalamolimbic populations The present findings indicate that the cells projecting to the nucleus accumbens, amygdaloid complex, and hippocampal formation are densely grouped in the thalamic midline. Although the different cell populations are highly intermingled, the present data clearly indicate that the neurons projecting to the amygdaloid complex are located mainly in the dorsal part of P V and medial portion of Re, those projecting to the hippocampal formation predominate in Re, whereas the midline efferents to the nucleus accumbens originate mainly from the ventral portion of PV, CeM, and Rh (Fig. 5). A topographical organization of thalamic midline projecting neurons has not been emphasized in previous single tracing studies, but was observed in this study with the simultaneous labeling of more than one cell population. Dense projections of the midline thalamus, and in particular of the Re nucleus, upon the entorhinal cortex have been described in several mammalian species, including primates (Segal, ’77; Herkenham, ’78; Room and Groenewegen, ’86; Insausti et al., ’87; Yanagihara et al., ‘87; Wouterlood et al., ’89). In the present study the entorhinal cortex was not examined as a separate target of midline efferents, but it was involved in the largest injections of the hippocampal formation, which extended into the parahippocampal region.

Midline cells are not the only source of thalamic input to the areas investigated in the present study. Afferents to the hippocampal formation originate from the anterior nuclei and also from the lateral dorsal nucleus in the rat (Wyss et al., ’79), cat (Irle and Markowitsch, ’82; Yanagihara et al., ’87), and monkey (Amaral and Cowan, ’80). The thalamic territory projecting to the amygdala also includes, in the posterior diencephalon, cells located close to the fasciculus retroflexus in the rat (Veening, ’78; Ottersen and Ben Ari, ’79), cat (Ottersen and Ben Ari, ’79; Russchen, ’82), and monkey (Aggleton et al., ’80; Mehler, ’80); amygdaloid projections from the medial pulvinar were reported in primates (Jones and Burton, ’76; Aggleton et al., ’80). Further, thalamic projections to the nucleus accumbens originate also from the posterior intralaminar complex (mainly the medial part of the parafascicular nucleus) in rodents (Newman and Winans, ’80; Phillipson and Griffiths, ’85) and the cat (Groenewegen et al., ’80; Beckstead, ’84; Jayaraman, ’85). However, the midline appears as the only thalamic “locus” in which the neurons projecting t o amygdala, hippocampus, and accumbens are located in close proximity to one another. In contrast, thalamic midline efferents are also distributed upon the neocortex; according to data obtained in the monkey (Friedman et al., ’87, ’89), midline fibers can exert their influence upon a variety of neocortical association fields through termination in layer I. Further, midline projections upon the cingulate cortex have been described in the rat (Thompson and Robertson, ’87),cat (Robertson and Kaitz, ’81; Musil and Olson, ’88a), and monkey (Vogt et al., ’79, ’87), as well as the prefrontal cortex in the cat (Niimi et al., ’81; Martinez-Moreno et al., ’87), where they innervate preferentially the ventromedial sector (Musil and Olson, ’88b). Therefore, in the thalamic midline, neural processes of different cell populations may interact, and they may receive common afferent information and convey it to different targets. The organization of the “non~pecific”thalamus, observed in Golgi preparations, supports the assumption of a possible structural interaction among midline cells. The Scheibels (’67) described in thalamic midline nuclei (e.g., in Re) an “open” or continuous dendritic field, characterized by the confluence of neuronal processes, in contrast with the apparent “closure” established by the dendritic arborization of adjacent thalamic nuclei, such as the mediodorsal (MD) nucleus. Furthermore, cells with radiating dendritic processes have been reported to predominate in the lie nucleus of the rat (Baisden and Hoover, ’79).

The midline versus the intralaminar connections On the basis of the finding that midline and intralaminar neurons were the main thalamic source of input to the

Fig. 4. The photomicrographs illustrate the structural relationships of paraventricular neurons, retrogradely labeled from injections of the tracer fluoro-gold into the amygdala, with the ependymal lining (outlined by dots). A. Note the orientation of the most dorsal labeled neuron parallel to the ependyma, and its dendritic arborizations extending toward the ependymal lining. B.Note the dendritic processes of labeled cells arborizing toward the ependyma. Short arrows point to distal dendritic ramifications, and long arrows to dendritic shafts in both A and B. Calibration bars in A and B: 20 pm.

Figure 4

H.-S. SU AND M. BENTIVOGLIO

590

monkey anterior cingulate cortex, Vogt et al. (’87) proposed that “midline and intralaminar nuclei can be classified as part of the limbic thalamus.” The present data on heavy limbic and limbic-related projections of midline structures, including the intralaminar CeM, support this proposal. However, the general pattern of organization of intralaminar and midline circuits displays similarities and differences that deserve comment. Both midline and intralaminar efferents are represented by multiple output channels, which directly influence wide sectors of the cortex and basal ganglia. However, the organization of intralaminar and midline connections indicates that they are mainly part of different functional systems. The lateral wing of the anterior intralaminar group, which includes the central lateral and paracentral nuclei, projects densely upon the neocortex and the neostriatum, and receives main inputs from the deep cerebellar nuclei and the spinothalamic tract (see Macchi and Bentivoglio, ’86). Previous studies in the cat noted that the anterior intralaminar nuclei convey spinal and cerebellar inputs to the motor cortex, cerebellar input to the anterior parietal association cortex, and that they may convey spinal and cerebellar inputs directly upon the head of the caudate through the thalamostriatal connection (Bentivoglio et al., ’88). The input-output organization of the midline nuclei is still poorly known. In the rat, thalamic midline nuclei receive afferents from brainstem structures involved in visceral and autonomic regulation, and derive their main input from a variety of hypothalamic nuclei (Herkenham, ’78; Cornwall and Phillipson, ’88), such that it has been stated that “the overall picture of these structures is of a periventricular system connected with accumbens projection neurons” (Cornwall and Phillipson, ’88). The present data emphasize that information is also conveyed from the thalamic midline upon amygdala and hippocampus. All of these data, taken together with the findings on the midline cortical projections, suggest that the midline thalamus acts as an interface between the limbic system, limbicrelated portions of the basal ganglia, and the association cortices, whereas the more lateral anterior intralaminar sector could subserve mechanisms of sensorimotor integration through connections with the neocortex and neostriatum. The cortical and subcortical efferents of the intralaminar nuclei are characterized by a low degree of divergent collateralization (Bentivoglio et al., ’81; Macchi et al., ’84). The present evidence of a limited proportion of midline branched cells further supports the view that widely distributed (so called nonspecific) thalamic projections derive from different neuronal subpopulations defined by their different targets.

Parallel channels in the exchange of information between thalamus and limbic system It has been shown in axonal transport tracing studies that the projections from hippocampus and amygdala to the thalamus diverge in two different terminal territories. In the monkey and the rat, the hippocampo-thalamic pathway reaches mainly the anterior nuclei and sparsely the midline (Swanson and Cowan, ’77; Meibach and Siegel, ’77; Aggleton et al., ’86). In contrast, the amygdalo-thalamic pathway terminates densely in the medial portion of MD in the rat (Krettek and Price, ’77) and terminates in the monkey in the MD magnocellular portion and sparsely in the midline

nuclei (Price and Amaral, ’81; Porrino et al., ’81; Aggletori and Mishkin, ’84; Russchen et al., ’87).The present evidence of a fairly discrete origin in the rat of midline efferents to amygdaloid complex, hippocampal formation, and nucleus accumbens, characterized by a low degree of divergent axonal branching, also indicates a segregation in the thalamic return loops upon limbic targets, as well as upon the “limbic” portion of the striatum. A discrete organization also has been identified in the relationships of amygdala and hippocampus with the basal forebrain. In the monkey, amygdaloid and hippocampal projections terminate in different areas of the basal forebrain (Aggleton et al., ’87). A low degree of multiareal collateralization has been reported in the widespread cortical projections of the basal forebrain in several mammalian species (Price and Stern, ’83; Walker et al., ’85), so that amygdaloid and hippocampal projections are likely to be conveyed upon different cortical targets of the basal forebrain. Further, the subcortical return loop to amygdala and hippocampus takes its origin from different neurons of the basal forebrain (Koliatsos et al., ’88). Together with the present findings on midline efferents, these data strongly indicate that the exchange of information with amygdala and hippocampus at a subcortical level is subserved by parallel channels, whose components seem to “require” a segregation, although they meet a t common crossroads.

Midline relationships with the ependyma The present study reveals a close topographical relationship of thalamo-amygdaloid cells with the ependymal-CSF boundary, and some of these neurons were seen to extend their dendritic arborizations toward the ependymal lining. This interesting feature, currently being investigated at the ultrastructural level, suggests that thalamo-amygdaloid new Tons may be a site of exchange with the CSF. It is noteworthy to recall that the contact between the thalamic midline and the ventricular cavities expands greatly in primates, so that in the human thalamus large portions ‘of the midline nuclei lie at the border with the ventricle, which separates the two hemithalami (Dewulf, ’71). This phylogenetic trend suggests that the structural contiguity with the CSF could represent an important feature of the role played by the thalamic midline. At present there is no clear indication of the substances involved in an exchange between CSF and midline neurons. In general, very little is known concerning the molecules that characterize neurotransmission and neuroregulation of these cells. In the rat, glutamate and/or aspartate have been indicated as putative neurotransmitters of thalamic midline neurons projecting to the ventral striatum (Christie et al., ’87; Fuller et al., ’87; Robinson and Beart, ’88). Re cells projecting to the hippocampal formation and ventral retrosplenial cortex contain butyrylcholinesterase (Robertson and Gorenstein, ’87). The chemical features of thalamoamygdaloid cells remain to be studied. Numerous data indicate that extrinsic peptidergic afferents are involved in the modulation of thalamic midline activity. In several mammalian species, thalamic midline regions are recipient of a dense innervation of substance 1’-, cholecystokinin-, neuropeptide Y-immunoreactive fibers, as well as of a rich enkephalinergic and dynorphinergic innervation (Hunt et al., ’87; Molinari et al., ’87; Battaglia et al., ’88; Herkenham et al., ’88; Hirai and Jones, ’89). Neurotensincontaining cell bodies have been described in the thalamic midline of the rat (Jennes et al., ’82) and the monkey

MIDLINE THALAMOLIMBIC PROJECTIONS

591

oh 0 ~,O O .,.

Ai

00

n. accumbens

AA

arnygdaloid complex

00

hippocampal formation

Fig. 5. The diagram summarizes the distribution of the neuronal populations observed in the present study, which project from the thalamic midline to limbic targets and the limbic-related portion of the striatum.

H.-S. SU AND M. BENTIVOGLIO

592

(Makino et al., ’87), but their efferent projections have not been identified simultaneously.

Functional considerations Midline nuclei have been originally associated with the “nonspecific” thalamic system regulating the electrical activity of the cerebral cortex involved in arousal and alertness (Jasper, ’49).With the advancement in the knowledge of the organization of the thalamocortical diffuse system (see Herkenham, ’86),the various “nonspecific” thalamic sectors have been related also to more “specific” functions. A role of the medial thalamus in memory formation has been repeatedly suggested, but the definition of diencephalic structures playing a critical role in memory processes are still under discussion (see Squire, ’87; Markowitsch, ’88). Experimental lesions of the monkey medial thalamus produced impairment of visual recognition memory (Aggleton and Mishkin, ’83), but these lesions involved the midline nuclei, as well as MD and the anterior thalamic structures. Lesions of MD were considered critical for memory impairment in clinical cases of Korsakoffs syndrome (Victor et al., ’71). However, Mair et al. (’79) ascribed the memory deficit in two amnesic patients to damage to the mammillary nuclei or midline thalamus rather than to MD. Experimental models of amnesia were obtained in primates with combined removal of amygdala and hippocampus (Mishkin, ’78). The present evidence that the thalamic midline represents a locus of prominent output to both amygdala and hippocampus favours its involvement in the circuits subserving the memory system. Through its mosaic of interspersed neuronal subpopulations, which include neurons possibly modulated by substances circulating in the CSF, the thalamic midline could be a critical locus of influence on the activity of basal ganglia, amygdala, hippocampus, and wide areas of the cortical mantle. Thus the thalamic midline could be the effector of a distributed system subserving the higher order processing of thalamic information. Inserted in such a broad distributed network, thalamolimbic midline cell populations could operate on some aspect of memory processing and provide a direct subcortical access to the extrapyramidal system for the message encoded by the “limbic” thalamus.

ACKNOWLEDGMENTS We dedicate this work to Dr. G. Macchi on the occasion of his seventieth birthday. This study was supported by grants from the Italian Ministry of Public Education. Dr. H.S. Su’s leave was supported by fellowships of the Paolo Zorzi Association for Neuroscience and of the Fidia Research Laboratories. The authors are grateful to Dr. A. Rosina (LFCN, CNR, Milan) for having kindly provided facilities for plotting, to Dr. L. Kruger for his helpful suggestions in the preparation of this manuscript, and to Dr. D. Schiff for reading it.

LITERATURE CITED Aggleton, J.P., and M. Mishkin (1983) Memory impairments following restricted thalamic lesions in monkey. Exp. Brain Res. 52r189-197. Aggleton, J.P., and M. Mishkin (1984) Projections of the amygdala to the thalamus in the cynomolgousmonkey. J. Comp. Neurol. 2225648. Aggleton, J.P., M.J. Burton, and R.E. Passingham (1980) Cortical and subcortical afferents to the amygdala of the rhesus monkey (Maeaca mulatta).Brain Res. 190:347-368. Aggleton, J.P., R. Desimone, and M. Mishkin (1986) The origin, course and

termination of the hippocampothalamic projections in the macaque. .I. Comp. Neurol. 243:409-421. Aggleton, J.P., D.P. Friedman, and M. Mishkin (1987) A comparison between the connections of the amygdala and hippocampus with the basal forebrain in the macaque. Exp. Brain Res. 67:556-568. Amaral, D.G., and W.M. Cowan (1980) Subcortical afferents to the hippocampal formation in the monkey. J. Comp. Neurol. 189:573-591. Baisden, R.H., and D.B. Hoover (1979) Cells of origin of the hippocampal afferent projection from the nucleus reuniens thalami. A combined Golgi-HRP study in the rat. Cell Tiss. Res. 203:387-392. Battaglia, G., R. Spreafico, and A. Rustioni (1988) Substance P-immunoreactive fibers in the thalamus from ascending somatosensory pathways. In M. Bentivoglio and R. Spreafico (eds): Cellular Thalamic Mechani5m.s. Amsterdam: Elsevier, pp. 365-374. Beckstead, R.M. (1978) Afferent connections to the entorhinal area in the rat as demonstrated by retrograde cell-labeling with horseradish peroxidase. Brain Res. 252249-264. Beckstead, R.M. (1984) The thalamostriatal projection in the cat. J. Comp. Neurol. 223:313-346. Bentivoglio, M., G. Macchi, and A. Albanese (1981) The cortical projections of the thalamic intralaminar nuclei, ae atudied in cat and rat with the multiple fluorescent retrograde tracing technique. Neurosci. Lett. 26r510. Bentivoglio, M., D. Minciacchi, M. Molinari, A. Granato, R. Spreafico, and G. Macchi (1988) The intrinsic and extrinsic organization of the thalaniic intralaminar nuclei. In M. Bentivoglio and R. Spreafico (eds): Cellular Thalamic Mechanisms. Amsterdam: Elsevier, pp. 221-237. Berendse, H.W., P. Voorn, A. te Kortschot, and H.J. Groenewegen (1988) Nuclear origin of thalamic afferents of the ventral striatum determines their relation to patch/matrix configurations in enkephalin-immunoreiictivity in the rat. J. Chem. Neuroanat. 1:3-10. Carman, J.B., W.M. Cowan, and T.P.S. Powell (1963) The organisation of cortico-striate connexions in the rabbit. Brain 86:525560. Christie, M.J., R.J. Summers, J.A. Stephenson, C.J. Cook, and P.M. Beart (1987) Excitatory amino acid projections to the nucleus accumbens septi in the rat: A retrograde transport study utilizing D [3H]aspartate and [SHIGABA. Neuroscience 22:425439. Cornwall, J., and O.T. Phillipson (1988) Afferent projections to the dorsal thalamus of the rat as shown by retrograde lectin transport. 11. The midline nuclei. Brain Res. Bull. 21:147-161. De Olmos, J.S., and W.R. Ingram (1972) The projection field of the stria terminalis in the rat brain. An experimental study. J. Comp. Neurol. 146303-334. Dewulf, A. (1971) Anatomyof the Normal Human Thalamus, Topometrymd Standardized Nomenclature. Amsterdam: Elsevier. Faull, R.L.M., and W.R. Mehler (1985) Thalamus. In G. Paxinos (ed): The Rat Nervous System. London: Academic Press, pp. 129-168. Friedman, D.P., L. Li, and L.G. Ungerleider (1989) Origins of thalamic projections to layer I of the cerebral cortex of the monkey. Sac. Neurosci. Abstr. 15:311 (Abstract). Friedman, D.P., J. Bachevalier, L.G. Ungerleider, and M. Mishkin (1987) Widespread thalamic projections to layer I of primate cortex. SOC. Neurosci. Ahstr. 13.951 (Abstract). Fuller, T.A., F.T. Russchen, J.L. Price (1987) Sources of presumptive glutamergic/aspartergic afferentsto the rat ventral striatopallidal region. J. Comp. Neurol. 258:317-338. Groenewegen, H.J., N.E.H.M. Becker, and A.H.M. Lohman (1980) Subcortical afferents of the nucleus accumbens septi in the cat, studied with retrograde axonal transport of horseradish peroxidase and bisbenzimid. Neuroscience 5:1903-1916. Groenewegen, H.J., E. Vermeulen-Van der Zee, A. te Kortschot, and M.P. Witter (1987) Organization of the projections from the subiculum to the ventral striatum in the rat. A study using anterograde transport of Phaseolus vulgaris leucoagglutinin. Neuroscience 23:103-120. Herkenham, M. (1978) The connections of the nucleus reuniens thalami: Evidence for a direct thalamo-hippocampal pathway in the rat. J. Comp. Neurol. 177:589-610. Herkenham, M. (1986) New perspectives on the organization and evolution of nonspecific thalamocortical projections. In E.G. Jones and A. Peters (eds): Cerebral Cortex, Vol. 5. New York: Plenum Press, pp. 403445. Herkenham, M., S. McLean, S.L. Moon., and C.B. Pert (1988) Autoradiography of receptor distributions suggests an endocrine function for neurotransmitters: Evidence from comparative anatomy of the opiate system. In M. Bentivoglio and R. Spreafico (eds): Cellular Thalamic Mechanisms. Amsterdam: Elsevier, pp. 417432. Hirai, T., and E.G. Jones (1989) A new parcellation ofthe human thalamus on the basis of histochemical staining. Brain Res. Reviews 14:l-34.

MIDLINE THALAMOLIMBIC PROJECTIONS Hunt, C.A., K.R. Seraogy, C.M. Gall, and E.G. Jones (1987) Cholecystokinin innervation of rat thalamus, including fibers to ventroposterolateral nucleus from dorsal column nuclei. Brain Res. 426257-269. Insausti, R., D.G. Amaral, and W.M. Cowan (1987) The entorhinal cortex of the monkey. 111. Subcortical afferents. J. Comp. Neurol. 264;39&408. Irle, E., and H.J. Markowitsch (1982) Connections of the hippocampal formation, mammillary bodies, anterior thalamus and cingulate cortex. Exp. Brain Res. 47~79-94. Jasper, H.H. (1949) Diffuse projection systems: The integrative action of the thalamic reticular system. Electroencephalogr. Clin. Neurophysiol. 1:405420. Jayaraman, A. (1985) Organization of thalamic projections in the nucleus accumbens and the caudate nucleus in cats and its relation with hippocampal and other subcortical afferents. J. Comp. Neurol. 231:396420. Jennes, L., W.E. Stumpf, and P.W. Kalivas (1982) Neurotensin: topographical distribution in rat brain by immunohistochemistry. J. Comp. Neurol. 210211-224. Jones, E.G. (1985) The Thalamus. New York: Plenum Press. Jones, E.G., and H. Burton (1976) A projection from the medial pulvinar to the amygdala in primates. Brain Res. 104,142-147. Kelley, A.E., and V.B. Domesick (1982) The distribution of the projection from the hippocampal formation to the nucleus accumbens in the rat: An anterograde and retrograde-horseradish peroxidase study. Neuroscience 7:2321-2335. Kelley, A.E., and Stinus, L. (1984) The distribution of the projection from the parataenial nucleus of the thalamus to the nucleus accumbens in the rat: An autoradiographic study. Exp. Brain Res. 54t499-512. Koliatsos, V.E., L.J. Martin, L.C. Walker, R.T. Richardson, M.R. DeLong, and D.L. Price (1988) Topographic, noncollateralized basal forebrain projections to amygdala, hippocampus, and anterior cingulate cortex in the rhesus monkey. Brain Res. 463:133-139. Krettek, J.E., and J.L. Price (1977) Projections from the amygdaloid complex to the cerebral cortex and thalamus in the rat and cat. J. Comp. Neurol. 172:687-722. Macchi, G., and M. Bentivoglio (1986) The thalamic intralaminar nuclei and the cerebral cortex. In A. Peters and E.G. Jones (eds): Cerebral Cortex, Vol. 5. New York: Plenum Press, pp. 355-401. Macchi, G., M. Bentivoglio, M. Molinari. and D. Minciacchi (1984) The thalamo-caudate versus thalamo-cortical projections as studied in the cat with fluorescent retrograde double labeling. Exp. Brain Res. 54~225-239. Mair, W.G.P., E.K. Warrington, and L. Weiskrantz (1979) Memory disorder in Korsakoff s psychosis. Neuropathological and neuropsychologicalinvestigation of two cases. Brain 102349-783. Makino, S., H. Okamura, N. Morimoto, J. Abe, N. Yanaihara, and Y. Ibata (1987) Distribution of neurotensin-like immunoreactivity in the diencephalon of the Japanese monkey (Macaca fuscata). J. Comp. Neurol. 26Ot552-563. Markowitsch, H.J. (1988) Diencephalic amnesia: A reorientation towards tracts? Brain Res. Reviews 13:351-370. Martinez-Moreno, E., A. Llamas, C. Avendano, E. Renes, and F. ReinosoSuarez (1987) General plan of the thalamic projections to the prefrontal cortex in the cat. Brain Res. 407:17-26. Mehler, W.R. (1980) Subcortical afferent connections of the amygdala in the monkey. J. Comp. Neurol. 190:733-762. Meibach, R.C., and A. Siege1 (1977) Thalamic projections of the hippocampal formation: Evidence for an alternate pathway involving the internal capsule. Brain Res. 134:l-12. Mesulam, M.M. (1978) Tetramethylbenzidine for horseradish peroxidase neurohistochemistry: A non carcinogenic blue reaction product with superior sensitivity for visualizing neural d e r e n t s and efferents. J. Histochem. Cytochem. 26:106-117. Mishkin, M. (1978) Memory in monkeys severely impaired by combined but not separate removal of amygdala and hippocampus. Nature 273.297-298. Molinari, M., S.H.C. Hendry, and E.G. Jones (1987) Distribution of certain neuropeptides in the primate thalamus. Brain Res. 426.270-289. Musil, S.Y., and C.R. Olson (1988a) Organization of cortical and subcortical projections to anterior cingulate cortex in the cat. J. Comp. Neurol. 272.203-218. M u d , S.Y., and C.R. Olson (1988b) Organization of cortical and subcortical projections to medial prefrontal cortex in the cat. J. Comp. Neural. 272219-241. Nauta, W.J.H., and D.G. Whitlock (1954) An anatomical analysis of the non-specific thalamic projection system. In J.F. Delafresnaye (ed): Brain Mechanisms and Consciousness. Springfield, I L Blackwell,pp. 81-103.

593 Newman, R., and S.S. Winans (1980) An experimental study of the ventral striatum of the golden hamster. I. Neuronal connections of the nucleus accumbens. J. Comp. Neurol. 191:167-192. Niimi, K., H. Matsuoka, T. Aisaka, and Y. Okada (1981) Thalamic afferents to the prefrontal cortex in the cat traced with horseradish peroxidase. J. Hirnforsch. 22221-242. Ottersen, O.P., and Y. Ben-Ari (1979)Afferent connections to the amygdaloid complex of the rat and cat. I. Projections from the thalamus. J. Camp. Neurol. 187t401-424. Phillipson, O.T., and A.C. Griffitha (1985) The topographic order of inputs to nucleus accumbens in the rat. Neuroscier.ce 16275-296. Porrino, L.J., A.M. Crane, and P.S. Goldman-Rakic (1981) Direct and indirect pathways from the amygdala to the frontal lobe in rhesus monkeys. J. Comp. Neurol. 198t121-136. Price, J.L., and D.G. Amaral (1981) An autoradiographic study of the projections of the central nucleus of the monkey amygdala. J. Neurosci. 1t1242-1259. Price, J.L., and R. Stern (1983) Individual cells in the nucleus basalisdiagonal band complex have restricted axonal projections to the cerebral cortex in the rat. Brain Res. 269:352-356. Riley, J.N., and R.Y. Moore (1981) Diencephalic and brainstem afferents to the hippocampal formation of the rat. Brain Res. Bull. (7437-444. Robertson, R.T., and C. Gorenstein (1987) “Non-specific” cholinesterasecontaining neurons of the dorsal thalamus project to medial limbic cortex. Brain Res. 404282-292. Robertson, R.T., and S.S. Kaitz (1981) Thalamic connections with limbic cortex. I. Thalamocortical projections. J. Comp. Neurol. 195t501-525. Robinson, T.G., and P.M. Beart (1988) Excitant amino acid projections from rat amygdala and thalamus to nucleus accumbens. Brain Res. Bull. 20:467-471. Room, P., and H.J. Groenewegen (1986) Connections of the parahippocampal cortex in the cat. 11. Subcortical afferents. J. Comp. Neurol. 251:451473. Russchen, F.T. (1982) Amygdalopetal projections in the cat. 11. Subcortical afferent connections. A study with retrograde tracing techniques. J. Comp. Neurol. 207:157-176. Russchen, F.T., D.G. Amaral, and J.L. Price (1987) The afferent input to the magnocellular division of the mediodorsal thalamic nucleus in the monkey, Macaca fascicularis. J. Comp. Neurol. 256t175-210. Scheibel, M.E., and A.B. Scheibel (1967) Structural organization of nonspecific thalamic nuclei and their projection toward cortex. Brain Res. 1:43-62. Segal, M. (1977) Afferent8 to the entorhinal cortex of the rat studied by the method of retrograde transport of horseradish peroxidase. Exp. Neurol. 57:75&765. Squire, L.R. (1987) Memory and Brain. New York Oxford University Press. Swanson, L.W., and W.M. Cowan (1977) An autoradiographic study of the organization of the efferent connections of the hippocampal formation in the rat. J. Comp. Neurol. 172t49-84. Thompson, S.M., and R.T. Robertson (1987) Organization of the subcortical pathways for sensory projections to the limbic cortex. I. Subcortical projections to the medial limbic cortex in the rat. J. Comp. Neurol. 265r175-188. Veening, J.G. (1978) Sucortical afferents of the amygdaloid complex in the rat: An HRP study. Neurosci. Lett. 8t197-202. Victor, M., R.D. Adams, and G.H. Collins (1971) The Wernicke-Korsakoff Syndrome. Philadelphia: Davis. Vogt, B.A., D.N. Pandya, and D.L. Rosene (1987) Cingulate cortex of the rhesus monkey. I. Cytoarchitecture and thalamic afferents. J. Comp. Neurol. 262256-270. Vogt, B.A., D.L. Rosene, and D.N. Pandya (1979) Thalamic and cortical d e r e n t s differentiate anterior from posterior cingulate cortex in the monkey. Science 204r205-207. Walker, L.C., C.A. Kitt, M.R. DeLong, and D.L. Price (1985) Noncollateral projections of basal forebrain neurons to frontal and parietal neocortex in primates. Brain Res. Bull. 15:307-314. Wouterlood, F.G., M.P. Witter, and E. Saldana (1989) PHA-L-tracing of thalamic afferents to the hippocampus and parahippocampal region. Light and electron microscopy. SOC.Neurosci. Abstr. 15t1245 (Abstract). Wyss, J.M., L.W. Swanson, and W.M. Cowan (1979) A study of subcortical afferents to the hippocampal formation in the rat. Neuroscience 4t463476. Yanagihara, M., K. Niimi, and K. Ono (1987) Thalamic projections to the hippocampal and entorhinal areas in the cat. J. Comp. Neurol. 266t122141.