Morphologic and cytochemical criteria for the identification and delineation of individual subnuclei within the lateral habenular complex of the rat^†

Semithin section technology

Coronal blocks were embedded in Araldite (24:30 w:w DDSA:CY212 Araldite, Serva, Heidelberg, Germany) and polymerized at 65°C overnight. Complete series of horizontal semithin sections (0.8 μm) were cut with a Leica Ultracut microtome and subsequently stained with 1% toluidine blue, pH 9.3 at 90°C.

Immunocytochemistry

Free-floating 20- to 25-μm horizontal cryostat sections or 40-μm horizontal Vibratome sections were pretreated with 1% sodium borohydride in phosphate buffered saline (PBS, 150 mM sodium chloride in 10 mM phosphate buffer, pH 7.4) for 15 minutes and subsequently permeabilized with 0.3% Triton X-100 for 30 minutes. Appropriately diluted primary antibody (mouse anti-tyrosine hydroxylase [anti-TH], Boehringer, Mannheim, Germany 1:1,000; mouse anti-neurofilament, Affinity Research Products Limited, Derbyshire, UK, 1:10,000; guinea pig anti–γ-aminobutyric acid-B receptor (anti-GABA-B R, Chemicon, Temecula, CA, 1:2,000; rabbit anti-Kir3.2, developed by our group, 1:1,000 in a solution of 10% normal goat serum, 0.3% Triton X-100, and 0.1% sodium azide in PBS) incubated for 24 hours was followed by 12 hours with biotinylated horse anti-mouse IgG, goat anti-rabbit IgG, or goat anti-guinea pig (Vector/Camon, Wiesbaden; diluted 1:2,000 in 0.2% bovine serum albumin in PBS), all at room temperature. After 6 hours with the Elite-ABC complex (Vector Laboratories, Burlingame, CA; diluted 1:1,000 in 0.2% bovine serum albumin in PBS), peroxidase activity was visualized with 1.4 mM 3,3′-diaminobenzidine in a solution of 10 mM imidazole in 50 mM Tris buffer, pH 7.6, supplemented by 0.3% nickelous ammonium sulfate and 0.015% H₂O₂ for 3 minutes. No staining was observed when the primary antibody was omitted. Sections were mounted onto gelatin-coated slides, dehydrated through a graded series of ethanol, and coverslipped in Entellan (Merck, Darmstadt, Germany).

Acetylcholinesterase cytochemistry

Acteylcholinesterase (AChE) activity was visualized by using a modified “Karnowski-protocol” (Schatz and Veh, 1987; Schatz et al., 1992). Briefly, cryostat or Vibratome sections were rinsed in 0.1 M maleate buffer, pH 6.3, and preincubated for 30 minutes in a solution containing 3.15 mM cobalt chloride and 5 mM sodium citrate in 0.1 M maleate buffer. The reaction was started by addition of 1/100 volume of 200 mM substrate solution containing 20 to 40 mM acetylthiocholine chloride and 180 to 160 mM acetylcholine. After 2 hours at 4°C, the incubation was stopped by rinsing the sections in maleate buffer. Subsequently, for double-staining, the immunocytochemical protocol was added.

Counterstaining with cresyl violet

Slide-mounted sections were left in 70% ethanol overnight, rinsed in bidistilled water, and stained for 30 minutes in a solution of 0.2% cresyl violet (cresyl violet acetate, Sigma, Deisenhofen, Germany) in 20 mM acetate buffer pH 4.0. After a short wash in bidistilled water, sections were dehydrated through a graded series of ethanol and cover-slipped in Entellan.

Photographs were taken with a digital camera (Olympus Camedia 4040) on a Leica DMRB microscope (Leica, Benzheim, Germany). Minor adjustments of contrast and brightness were made by using Adobe Photoshop 6.0.

Subnuclei of the medial division of the lateral habenular complex

As might be expected, shape and size of the subnuclei vary somewhat along the rostrocaudal axis of the habenular complex. For the most obvious demonstration of the LHbM and LHbL subnuclei the level at the rostral beginning medial root of the fasciculus retroflexus was selected. It roughly corresponds to the area between plate 32 and plate 33 of the commonly used rodent brain atlas (Paxinos and Watson, 1998).

Superior subnucleus of the medial division of the lateral habenular complex.

Situated just below the stria medullaris, the superior subnucleus (LHbMS) is a cone-shaped area with a ventrally pointing tip (Fig. 1A,B; Table 2). In semithin sections, its pale, relatively small (diameter approximately 6 to 12 μm) round to oval-shaped neurons are embedded in a neuropil, which is largely occupied by obliquely cut dendrites (Fig. 2B). Laterally, the superior subnucleus can be easily delineated from the magnocellular subnucleus of the LHbL (LHbLMc; see below) due to the large nerve cells and the neuropil consisting of many myelinated fibers and large dendrites of the LHbLMc (Fig. 2A,B).

Some of these morphologic features can also be recognized in cresyl violet-stained cryostat sections (Figs. 1C, 2C). The small, round, and densely packed cells of the LHbMS are easily detected and distinguished from the large and more loosely arranged neurons of the LHbLMc (Fig. 2C). No structural information on the neuropil can be gained from these sections.

Additional information is obtained from chemoarchitectural data. When stained with an antibody against the GABA-B receptor (Figs. 1E, 2D), the typical small neurons of the LHbMS are still recognizable, even within a heavily labeled neuropil. Furthermore, the LHbMS displays AChE activity (Figs. 1H, 2E) but is largely devoid of TH (Fig. 1D), neurofilament (NF; Figs. 1H, 2E), and Kir3.2 immunoreactivities (Figs. 1G, 2F).

Morphologic as well as cytochemical data are necessary for a delineation from the surrounding areas. The medial border to the medial habenular complex (MHb) is most obvious in cresyl violet-, Kir3.2-, or AChE/NF- stained sections (Figs. 1C,G,H, 2F). For the identification of the lateral border, the strong GABA-B receptor immunoreactivity yields the most precise demarcation against the weaker-labeled LHbLMc (Figs. 1E, 2D). In addition, NF- and Kir3.2 are also valuable markers here, due to their abrupt increase in immunoreactivity in the LHbLMc area (Figs. 1G,H).

Subnuclei of the lateral division of the lateral habenular complex

The lateral division of the LHb occupies two thirds of the LHb area. Due to the generally large sizes of its neurons and the usually high density of neurofilament protein, the area is obvious by gross inspection of the corresponding sections (Fig. 1C,F).

Parvocellular subnucleus of the lateral division of the lateral habenular complex

The parvocellular subnucleus (LHbLPc) represents another strip of neuronal tissue (Fig. 1A,B) in a characteristic position of the LHbL (Table 2). Dorsally situated, just below or even within the fiber bundles of the stria medullaris (Figs. 1B, 6A), its semithin characteristics closely resemble those of the LHbMS. Like the latter, the LHbLPc consists of small neurons (approximately 6 to 10 μm) and an extremely fine neuropil, which is dominated by dendrites with mainly an oblique orientation. Despite its close vicinity to the stria medullaris, there are only very few myelinated axons inside this subnucleus, making the demarcation from the surrounding fiber bundles easy.

The neuropil of the LHbLPc displays immunoreactivities for most neuropeptides investigated so far (not shown). Its GABA-B receptor immunoreactivity provides exact information about the precise area occupied by the LHbLPc (Figs. 1E, 6C). Small neurons are visible as well as the fine neuropil. Even the typical invaginations of the parvocellular nucleus into the stria medullaris may be recognized (Fig. 6C, compare with 6A). Again, the a-Kir3.2 immunocytochemistry shows a negative mirror image to the GABA-B receptor distribution (Fig. 6D).

With the exception of its dorsal border, the LHbLPc is surrounded by the magnocellular subnucleus of the LHbL. Its delineation poses no problems. In Nissl-stained sections, the small neurons of the LHbLPc are quite easily distinguished, dorsally from the almost neuron-free area of the stria medullaris as well as ventrally from the large neurons of the LHbLMc (Fig. 6B). Although staining with anti-Kir3.2 antibody is not helpful, because positively labeled dendrites of LHbLMc neurons extend into the area of the parvocellular subnucleus, the GABA-B receptor immunoreactivity confirms the morphologically identified borders (Figs. 1E, 6C).

Identification of habenular subnuclei along the rostrocaudal axis of the lateral habenular complex

According to an atlas of the rat brain (Paxinos and Watson, 1998), the first few cells of the medial habenular complex appear approximately 1.9 mm caudal to bregma. Around 200 μm further caudal, the first neurons of the LHbMA, the most rostral subnucleus of the lateral habenular complex, become visible. Another 600 to 700 μm more caudally (approximately bregma −2.7), most other subnuclei of the LHb begin to appear. For the primary identification of the individual subnuclei of the LHb, therefore, a level at approximately bregma −3.2 was selected.

Subsequent to the elaboration of the cytochemical criteria for the identification of the LHb subnuclei at a single selected rostrocaudal position, we tested their validity at additional levels. For this purpose, sections at about bregma −3.0 (Fig. 8A1–D1), bregma −3.3 (Fig. 8A2–D2), bregma −3.5 (Fig. 8A3–D3), and bregma −3.8 (Fig. 8A4–D4) were selected.

From the analysis of semithin sections, it is known that the individual subnuclei tend to slightly change form and size during their rostral to caudal extension, but keep their relative position. These data are well confirmed by our observations, regarding for example the rostrocaudal extensions of the LHbMPc and the LHbMC (Fig. 8A1–A4). Rostrally, these subnuclei form an oval complex, the largest extent running medial to laterally (Fig. 8A1). Along the rostrocaudal course, this largest diameter of the two subnuclei slowly turns by 90° extending into a dorsoventral direction at the most caudal levels (Fig. 8A4).

This finding is also reflected by the cytochemical markers as shown for the staining with anti-Kir3.2-antibodies (Fig. 8B1–B4), with anti–GABA-B receptor antibodies (Fig. 8C1–C4), and with anti-neurofilament antibodies (Fig. 8D1–D4). With Kir3.2 immunoreactivity, the LHbMS, LHbMMg, LHbLPc, LHbLMg, LHbLMc, and LHbLO can be delineated throughout the rostrocaudal extent of the lateral habenular complex. By using GABA-B receptor immunoreactivity, the same subnuclei can be identified, although the LHbLO is more easily detected in Kir3.2 than in GABA-B receptor staining. At rostral levels, the LHbMPc and a part of the LHbMC are demarcated by Kir3.2 and GABA-B receptor immunoreactivity (Figs. 1D,F, 8A2,A3), whereas further caudally (Fig. 8B2–D2,B3–D3,) it is only the LHbMPc that can be identified. When an antibody against neurofilament is chosen (Fig. 8A4–D4), the LHbMPc/LHbMC complex as well as the LHbMS, LHbMMg, LHbLMc, LHbLMg, and LHbLO are easily identified throughout the rostrocaudal extent. These data confirm that the newly elaborated cytochemical criteria are very well suited for the identification of individual subnuclei, not only at a single level, but also along the rostrocaudal extent of the LHb area in the rat brain.

Only few habenular subnuclei can be identified by staining with cresyl violet

Classic neuroanatomic work is predominantly based on the analysis of Nissl-stained material. When this method is applied to coronal sections of the LHb, differences of cell morphology and cell density within the LHb become evident. The medial part of the LHb is characterized by its small and closely packed cells, whereas the lateral part consists of large, more loosely arranged neurons. When the criteria obtained from the analysis of semithin sections are transferred to cresyl violet–stained sections, individual subnuclei of the LHb can be recognized. The small, densely packed neurons of the LHbMPc can be differentiated from the surrounding, somewhat larger neurons of the LHbMPc with their conspicuous nucleoli. The area of the LHbMPc/LHbMC complex itself is surrounded by the large and loosely arranged neurons of the LHbLMc. Similarly, the LHbLPc can be delineated by its small neurons situated within or ventral to the stria medullaris.

On the other hand, due to the similar morphology of LHbLMc and LHbLO neurons, these subnuclei cannot easily be separated from each other in Nissl-stained material. Similar problems preclude the delineation of the LHbMMg from the LHbMC and the LHbLMc from the LHbLB. Taken together, staining with cresyl violet on its own does not provide sufficient criteria for the unequivocal delineation of all LHb subnuclei from each other.

Neither the distribution of acetylcholinesterase nor that of calcium-binding proteins is restricted to individual subnuclei along the rostrocaudal axis of the lateral habenular complex

Cytochemical markers such as AChE cytochemistry or immunocytochemical localization of calcium-binding proteins are increasingly appreciated in recent times (Gerfen et al, 1985; Kincaid and Wilson, 1996; Paxinos and Watson, 1998). Consequently, these were the first markers that were analyzed for their potential value for a delineation of lateral habenular subnuclei.

As expected from the earlier descriptions of their localization, calcium binding proteins such as parvalbumin, calbindin (Celio, 1990), calretinin (Jacobowitz and Winsky, 1991), neuronal calcium sensor protein-1 (De Raad et al., 1995; Pitt et al., 1998), and neuronal visinin-like protein-3 (Pitt et al., 1998; Bernstein et al., 1999) displayed a heterogeneous distribution throughout the lateral habenular complex. In pilot experiments, especially calbindin and calretinin appeared to selectively stain the LHbMPc and/or LHbMS (unpublished observations). The subnucleus-restricted localization of these calcium-binding proteins, however, was not constant along the rostrocaudal axis of the LHb. Altogether, the immunocytochemical localization of these calcium binding proteins did not support the expectations for their potential value as markers for habenular subnuclei.

The visualization of AChE activity provides the basis for a chemoarchitectonic parcellation of many brain structures. The high value of this cytochemical marker, which allows many distinctions such as the easy delineation of the reticular from thalamic core nuclei or the identification of the patch compartment within the striatal matrix area, is appreciated by its representation in the commonly used atlas of the rodent brain (Paxinos and Watson, 1998). From this source, a heterogeneous distribution of AChE activity within the habenular area, which might correspond to individual subnuclei, is already evident.

At rostral levels, AChE is almost homogeneously expressed within the LHb with a somewhat reduced activity in the LHbMS (see Figs. 1H, 2E). Further caudally, however, the LHbMPc, LHbLPc, and LHbLO are exclusively labeled, whereas in the most caudal fourth, the AChE activity is selectively expressed within the LHbMPc and LHbLPc (not shown). Thus, AChE staining can be used to recognize LHbMPc and LHbLPc from the middle to caudal third of the LHb but is of limited value to delineate lateral habenular subnuclei throughout their rostrocaudal extent.

Differential distributions of GABA-B receptor, Kir3.2 channel subunit, and neurofilament protein are most useful for the immunocytochemical identification of subnuclei within the LHb

In pilot experiments, the distribution within the LHb of more than 30 neuroactive molecules was analyzed. Of these, three turned out to be most useful for the immunocytochemical identification of subnuclei within the lateral habenular complex. These were antibodies against the two subunits of the metabotropic GABA-B receptor (Charara et al., 2000), an antibody against one subunit (Kir3.2) of a G-protein modified inwardly rectifying potassium channel (Kir3 family proteins, GIRKs; Reimann and Ashcroft, 1999) and an antibody against the nonphosphorylated neurofilament-H protein. These three markers stain different subnuclei within one section.

GABA-B receptor immunoreactivity is especially strong in the LHbMS, LHbMPc, and LHbLPc. Common to these three subnuclei are small-sized neurons and a neuropil that is dominated by dendrites. Consequently, the receptor seems to be expressed mostly on dendrites rather than on axons. Such a postsynaptic localization argues against a functional role as autoreceptors on axon terminals, which is one of the important functions of this receptor in the brain (Charara et al., 2000).

Most often, the biological effects of the GABA-B receptor are transduced by a direct interaction of G-protein beta/gamma subunits with G-protein regulated inwardly rectifying potassium channels, requiring the localization of all these proteins within the same compartments of individual cells. Surprisingly enough, in our material, the distribution of Kir3.2 immunoreactivity was not congruent but largely complementary to that of the GABA-B receptor. Furthermore, the distribution of the μ-opiate immunoreactivity, which is also known to exert its effect by means of opening of Kir3 channels, corresponded well to that of the GABA-B receptor in the LHbM, but not at all to that of the Kir3.2 protein (not shown). These data suggest that, in the LHb area, the inhibitory effects of the GABA-B as well as of the μ-opiate receptor are mediated by Kir3 channels with compositions devoid of the Kir3.2 subunit. Such a type of channels (Kir3.1/Kir3.4-channels) are known to be responsible for the muscarinergic effects in the heart (Krapivinsky et al., 1998). Alternately, other mechanisms such as the inhibition of adenylate cyclase activity or, as recently shown in the case of 5-HT2 receptors (Carr et al., 2002), the phospholipase C–mediated inhibition of sodium channels have to be considered.

Independently of their molecular functions, the distribution of the Kir3.2 channel protein in the LHb area, due to its largely complementary distribution with respect to the GABA-B receptor, is helpful for the identification of subnuclei. Thus, GABA-B receptor immunoreactivity is strongly displayed in the LHbMS, LHbMPc, a part of the LHbMC, and LHbLPc subnuclei, where Kir3.2 expression is modest or absent. On the contrary, the LHbMMg, LHbLMc, and especially the LHbLO subnuclei show heavy expression of the Kir3.2 protein and are largely devoid of GABA-B receptor immunoreactivity.

Residual ambiguity in areas where the GABA-B receptor and Kir3.2 do not yield sufficient information is relieved with the aid of TH and neurofilament immunoreactivity (see Table 2). The LHbLMc displays intermediate staining for both, GABA-B receptor and Kir3.2 protein. Its strong NF immunoreactivity, however, facilitates the delineation of its medial border vs. the LHbMPc/LHbMC complex. With TH immunoreactivity, the LHbMPc can be easily and precisely demarcated. The majority of these TH-positive fibers is dopaminergic in nature, as they are present after depletion and specific uptake of dopamine and resistant to neurotoxic lesions of the locus coeruleus (Skagerberg et al., 1984). This projection arises primarily from cell bodies in the interfascicular nucleus of the ventral tegmental area. However, only a minority of these VTA fibers, which are directed to the lateral habenular complex, contains dopamine (Skagerberg et al., 1984).

Taken together, staining for GABA-B receptor and Kir3.2 protein combined with the visualization of neurofilament- and tyrosin hydroxylase-immunoreactivities allows the unequivocal identification of most subnuclei within the lateral habenular area. Additional Nissl and AchE staining facilitates the reference to the corresponding tables of the commonly used rodent brain atlas (Paxinos and Watson, 1998).

Why no earlier attempts to cytochemically delineate habenular subnuclei?

The phylogenetically old age of the habenular complex is in line with a considerable number of different neurotransmitters and neuropeptides localized within this nuclear group. Unfortunately, in most of the available literature, only a few sections of the habenular complex are depicted and, often, the precise rostrocaudal position is not indicated exactly. Furthermore, due to the few (sometimes only one) sections shown, the putative rostrocaudal changes of the immunoreactivity cannot be recognized. Available data, therefore, need very cautious interpretation. Examples may be understood as supporting several organizational forms of the lateral habenular area, such as a mediolateral gradient, a division into a medial and a lateral subfield, or subdivision into many subnuclei. Thus somatostatin receptor 1 (Hervieu and Emson, 1998) and substance P (Shinoda et al., 1984) seem to be expressed within the entire LHb, whereas neurotensin (Jennes et al., 1982; Görtzen et al., 1995), serotonin (Geisler et al., 1999), and agmatine (Otake et al., 1998) immunoreactivity is preferentially located in the LHbM. In contrast, parathyroid hormone-2 receptor (Wang et al., 2000) and neuromedin (Honzawa et al., 1987) are primarily detected in the LHbL. Although the localization of immunoreactive fibers and neurons with respect to individual subnuclei of the LHb is usually not very clear, there are examples, in which the distribution of label is very suggestive of a localization within distinct subnuclei. The area of dense Leu-enkephalin–immunoreactive fibers (Shinoda and Tohyama, 1987) apparently corresponds to the LHbMS, whereas the dopamine transporter (Freed et al., 1995) and vesicular acetylcholine transporter (Scháfer et al., 1998) seem to be selectively expressed in the LHbMPc/LHbMC. Furthermore, urocortin III (Li et al., 2002) as well as NCS1 immunoreactivity (Pitt et al., 1998) could be expressed within the LHbMMg.

Thus, there are substances, which are expressed either selective or preferential within the LHbM or the LHbL. On the other hand, there are also neuroactive substances, which reveal a more circumscribed localization within the LHbM or LHbL, and this localization could be restricted to individual subnuclei of the LHb. As mentioned above, however, when analyzed in single sections only, no substantial conclusion can be drawn on the localization of a selected neuroactive substance with respect to a given subnucleus. In contrast, in our investigations, more than 15 complete cryostat or Vibratome series in addition to several semithin series (together approximately 15,000 individual sections) throughout the longitudinal axis of the habenular complex were analyzed. Only the comparison of immunostained material with adjacent Nissl sections and corresponding semithin sections allowed the correlative identification of individual subnuclei within the lateral habenular complex.

Putative functional roles of individual subnuclei

Neurons in a round, circumscribed area in the LHbM are activated by different forms of stress, like immobilization, placement in a novel environment, food deprivation, chronic intermittent hypoxia, and lithium injection (Chastrette et al., 1991; Wirtshafter et al., 1994; Sica et al., 2000; Timonefa and Richard, 2001). This round, circumscribed area closely resembles the appearance of the LHbMPc/LHbMC in shape, size, and position, suggesting a role of this subnucleus in mediating a stress response. This concept is supported by hodologic studies, indicating the area of the LHbMPc/LHbMC as the source of afferents to the dorsal raphe nucleus (Kalén et al., 1985; Gervasconi et al., 2000), a brain structure commonly associated with behavioral stress responses (Amat et al., 2001). On the other hand, it seems to be the same round circumscribed area, which also expresses an increased neuronal activity due to the onset of active maternal behavior (Lonstein et al., 1998; Kalinichev et al., 2000), to stimulating food intake (Olszewski et al., 2000) and to self-stimulation (Hunt and McGregor, 1998). At a first glance, these biological functions appear to be unrelated to each other and the question arises, why and whether one subnucleus is involved in such different functions. A common denominator, however, may be that all of these functions—stress, onset of maternal behavior, stimulated food intake and self-stimulation—are associated with an increased state of wakefulness. An involvement of the LHbMPc/LHbMC in arousal-related behavior is further substantiated by its strong reciprocal connections with the laterodorsal tegmental nucleus (Cornwall et al., 1990), a prominent part of the cholinergic ascending reticular activating system.

Neurons within the area of the LHbL seem to be involved in basal ganglia circuitry. After administration of different dopamine agonists, like apomorphine and amphetamine, neurons are activated that seem to be localized within the LHbLMc (see Fig. 1 in Wirtshafter et al., 1994). Interestingly, it is this subnucleus that receives GABAergic afferents from the rostral pole of the entopeduncular nucleus (Parent et al., 1980; Rajakumar et al., 1993, 1994). It is likely, therefore, that the effects of dopamine agonists on the LHbLMc are mediated by alterations in striatal activity, which are relayed to the LHbLMc through the entopeduncular nucleus.

These data indicate that neurons localized in different circumscribed areas within the LHb are involved in distinct biological functions. These circumscribed areas closely resemble in shape, size, and position some of the newly described habenular subnuclei, suggesting a functional relevance of some of these entities.

Whether these areas really correspond to the described subnuclei and whether the other subnuclei are also involved in distinct biological functions needs further investigation. The present report will advance further functional studies of habenular subnuclei by facilitating the identification on conventional cryostat and Vibratome sections.