The animals were killed 13 days after surgical interventions. After anesthesia with a mixture of 1 ml/kg body weight ketamine (100 mg/ml; CP-Pharma) and 0.6 ml/kg body weight xylazin (20 mg/ml; CP-Pharma) delivered intramuscularly, rats were perfused transcardially with 0.9% saline followed by 300 ml of 4% paraformaldehyde, pH 7.4. The brains were removed, post-fixed overnight in a fresh solution, then cryoprotected with 20% sucrose for 24 h, at 4°C. Fifty μ m thick serial coronal sections were cut from the hypothalamus to the spinal cord on a frigomobile (Frigomobil, Reichert-Jung, Vienna, Austria), and the sections were used for immunohistochemistry and immunofluorescence visualizations of BDA-labeling.

Every fourth section was stained for BDA using either immunoperoxidase or immunofluorescence procedures with or without tyramide-mediated amplification (Adams, 1992; Reiner et al., 2000). Free-floating sections were pretreated in 0.5% Triton X-100 for 1 h at room temperature. Sections were then incubated in avidin-biotin-horseradish peroxidase complex (ABC) at 1:500 dilution (Vectastain ABC Elite kit, Vector Laboratories, Burlingame, CA, USA) for 1 h. For biotin-tyramide amplification, sections were rinsed extensively first in PB and then in 0.05 M Tris buffer at pH 8.2 (TB), then sections were placed in biotin-tyramide solution (1:20,000) prepared in TB containing 0.0015% hydrogen peroxide for 15 min. Then a second ABC incubation was conducted at 1:500 for 30 min. Finally, the BDA-fibers were visualized using nickel enhanced 3,3-diaminobenzidine (Ni-DAB Sigma, St. Louis, MO, USA), a chromogen reaction resulting in a dark blue reaction product. Alternatively, after the ABC reaction FITC-tyramide amplification was performed (FITC-tyramide 1:10,000; 0.0015% H₂O₂ in TB for 10 min). Ni-DAB sections were mounted on gelatin coated slides (Intersan Kft., Budapest, Hungary) and coverslipped with DPX Mountant (Fluka, Switzerland). Fluorescent sections were mounted onto non-coated preclean slides (Intersan Kft.), and coverslipped with Aqua-Poly/Mount (Polysciences, Inc., Warrington, USA).

Sites of tracer deposition within the DLH were depicted on a series of computerized standard drawings of the rat brain from the atlas of Paxinos and Watson (2007) with the help of Adobe Photoshop CS 8.0.

Stained sections were examined with an Olympus BX60 microscope equipped with fluorescent epi-illumination. In a complete set of serial sections with 200 μm interval between the sections (every fourth 50 μm thick section), the density of the BDA-labeled fibers in the lower brainstem was assessed in each section. The relative abundance of BDA-labeled fibers was evaluated by the following grading: absence of labeled fibers (−), very low (±), low (+), moderate (++), and high (+++), see Table 1. The data from individually evaluated brains were grouped based on the given subdivision. The data were estimated in the following aspects: (1) which and how many of the brainstem areas/nuclei/subdivisions showed labeled fibers after injections into the 8 investigated DLH subdivisions; (2) the participation of the 8 DLH subdivisions in the innervations of the 116 brainstem areas/nuclei/subdivisions (how many subdivisions project to a particular brainstem subdivisions); (3) which brainstem areas/nuclei/subdivisions showed up high densities of labeled fibers; (4) evaluation of BDA fiber density in the special cell groups of the brainstem (innervations of biogenic amines, reticular formation, etc., see grouping in Table 1).

Totally, six rats were found with topographically correct injection sites: two in each of the 3 subdivisions of the dorsomedial hypothalamic area.

In the case of periventricular area, the injected areas were relatively small, their average mediolateral diameters did not exceed 250 μm in the coronal plane, but they extended more in rostrocaudal and posteroventral directions. The injected area was located immediately adjacent to the third ventricle. They extended over the ventricle, dorsally. This division showed a small overlap with the dorsal part of the dorsomedial nucleus (Figures 2A,B, 3A).

Regarding to the dorsal part of the dorsomedial nucleus, the sizes of injections varied between 400 and 500 μm (Figures 2B, 3B). In one case, the injection spread in a relatively large area, covering almost the entire dorsomedial part of the dorsomedial nucleus. In the second rat, the injection's site was smaller. In both cases, the tracer spread slightly into the posterior periventricular area.

The sizes of the injections in dorsomedial hypothalamic subdivision were small, between 350 and 400 μm average diameters (Figures 2A,B, 3C). The injections centered in the dorsomedial hypothalamic area between the dorsomedial nucleus and the medial border of the perifornical area. BDA spread slightly into these two neighboring areas.

Eleven injections were classified into the perifornical area (Figures 2A–C, 3D–F): two injections in the rostral (PeFr), six in the caudomedial (PeFcm), and three in the caudolateral (PeFcl) perifornical areas.

The sizes of injections in the rostral perifornical area were 350 and 450 μm. In adult rats, this area occupies a territory between the fornix and the most medial part of the zona incerta from the dorsomedial nucleus until the medial border of the lateral hypothalamic area (Figure 1G). Rostro-caudally, it extends between 2.4 and 2.8 mm caudal to the level of the bregma. In both cases, the injected BDA spread entirely within the boundaries of the area (Figures 2A, 3D).

A large area over the fornix with a rostro-caudal extension between 2.8 and 3.4 mm caudal to the bregma represents the caudal portion of the perifornical area (Figures 1H,I). An imaginary vertical line through the fornix divides this field into medial and lateral areas (Figures 1H,I, 2B,C). Regarding to the caudomedial part of the perifornical area, the average diameter of the injections varied between 250 and 700 μm. In three cases, BDA-positive neurons were located mainly within the boundaries of the area, whereas in the other 3 rats BDA spread also to neighboring portion of the lateral area (Figures 2B,C, 3E).

In the case of the caudolateral part of the perifornical area, the sizes of injections varied between 200 and 450 μm (Figures 2B,C, 3F). The 2 smaller injections centered within the area (Figure 2C), whereas the third, a large one, spread partly into the neighboring LHd (Figure 2B).

Five injections were centered in the lateral hypothalamic area: two injections in the ventral and three in the dorsal parts. (In this study, the term of “lateral hypothalamic area” is restricted to the posterior portion of the “hypothalamus-long” lateral hypothalamic area that extended between 2.4 and 3.4 mm caudal to the bregma level.) In the present study this area was arbitrary divided into dorsal and ventral areas by an imaginary line between the lower edge of the fornix and the centromedial point of the cross-sectional profile of the cerebral peduncle (Figures 1J–L, 2A–C).

Both injections into the ventral part of the lateral hypothalamic area were located entirely inside the boundaries of the area. Their sizes were moderate with average diameters around 400 μm (Figures 2C, 3H).

The sizes of injections into the dorsal part of the lateral hypothalamic area varied between 250 and 350 μm. In two rats, all of the BDA-labeled cells were within the boundaries of the area, whereas, in the other case, a small spread of label was seen in the LHv (Figures 2A,C, 3G).

In general, the locations of BDA-positive fibers in the midbrain (Figure 4A) showed ipsilateral dominance to the unilateral application of the tracer in the DLH (Figures 4C–F).

The most densely network of labeled fibers in the midbrain appeared in the periaqueductal central gray (PAG). They arise mainly in the three subdivisions of the perifornical area and the periventricular subdivision of the dorsomedial hypothalamic area (Table 1). Dense network of BDA-positive fibers was seen in the ventrolateral and the ventral parts (Figures 4B,D), moderate to weak in the other parts of the PAG. Although, a substantial portion of the fibers appeared to be “axons of passage,” there were also interspersed numerous axons with varicosities, indicating possible synaptic contacts.

Numerous axons fanned out in the midbrain lateral and ventral to the PAG, predominantly in the deep mesencephalic (also called cuneiform) nucleus, the precuneiform area, and in the isthmic reticular formation (Figures 4C–E).

The dorsal raphe nucleus established moderate to heavy distribution patterns of BDA-positive fibers originated from the PeF and LHv subdivisions of the DLH (Figures 4C–E). In the median raphe nucleus (also known as midbrain raphe or superior central raphe nucleus) only light or very low projections were observed (Figure 4F). The rhabdoid and paramedian raphe nuclei showed moderate density of BDA fibers (Figures 4E,F, respectively).

The other investigated midbrain nuclei and areas, including the oculomotor and trochlear nuclei showed low fiber densities or were devoid of BDA-positive fibers (Table 1).

Like in the midbrain, the distribution of the BDA fibers was bilateral but with strong ipsilateral dominance.

A moderate density of partly en passant BDA-positive fibers was apparent within the oral part of the pontine reticular formation after injections into the PeFcm and PeFcl, and lighter densities after the other injections. Light to moderate density of labeling was observed in both the oral and caudal parts of the pontine reticular formation especially after injections into PeV, PeFr, and LHv (Table 1).

Labeled fibers were present in the pedunculopontine and the laterodorsal tegmental nuclei after all injections into the 8 subdivisions. Their density was very high after LHv and moderate after PeFcl injections (Table 1).

Dense network of BDA-positive fibers was seen in the Barrington's nucleus but only in PeV injected rats (Figures 5A,C,E). The density of the BDA-labeled fibers in the locus coeruleus was almost equal to those in the Barrington's nucleus. The PeV and the subdivisions of the PeF generated moderate, while the other subdivisions of the DLH only light (or sparse in the case of DMH) projections to the locus coeruleus (Figures 5D,E). In the caudal portion of the locus coeruleus, from where neurons projects to the spinal cord, the labeling of fibers were rare (Figures 5B,F). Ventral to the locus coeruleus, in the subcoeruleus area, especially in its dorsal part, moderately dense fiber network was present. The fibers were seen mainly in rats with injections in the PeFr, the PeFcm, and the LHv (Figures 5C,D). BDA-positive fibers were present in very light to moderate densities in the lateral and medial parabrachial nuclei. The neurons of PeFr and PeFcl innervated moderately both parabrachial nuclei.

In the ventrolateral area of the pons, few fibers were spread throughout the area of the A5 noradrenaline cell group. Few labeled fibers were seen in the pontine raphe and pontine nuclei, some in the sensory and motor nuclei of the pons (Table 1), while in many others, like sensory trigeminal, cochlear, and lateral vestibular nuclei were devoid of BDA-labeled fibers or terminals (not shown).

Although generally, sparse labeled fibers could be traced all along the medulla, their topographical distributions and density were heterogeneous. They showed generally bilateral appearance with ipsilateral dominance.

The subdivisions of the DLH generated light labeling in the gigantocellular reticular nucleus (Figures 6A,C,E). In the vicinity, somewhat lesser density of fibers was observed in the lateral paragigantocellular and the intermediate reticular nucleus. Except neurons in the DMN, labeled fibers arise in the other seven subdivisions of the DLH, but in different densities (Table 1).

From all of the DLH divisions (with exception of DMN) projections were generated to the nucleus of the solitary tract: light ones from neurons of the PeV, PeFr, and LHv, and sparse projections from the other areas (Figures 6B,D). It was not the case for the dorsal motor nucleus of the vagus, where the BDA-labeling was low or sparse.

In the territory of the medullary catecholamine cell groups, despite of the gap of heavy innervations, the BDA fibers originated from more individual subdivisions of the DLH (Table 1). The source of the fibers was especially wide in the case of A2 noradrenaline cell group, which is innervated by all of the investigated DLH subdivisions. BDA-labeled fibers in the caudal part of the A1 noradrenaline (Figures 6B,F) and C1 adrenaline cell groups arised in 7 subdivisions, whereas the rostral part of the A1 noradrenaline and the C2 adrenaline cell groups received fibers from 5 different subregions of the DLH.

Light to moderate density of varicose fibers was present in the raphe magnus nucleus originating from all the subdivisions of the DLH (Figures 6C,E, Table 1). In the raphe pallidus and obscurus nuclei delicate varicose fibers were apparent after the tracer injections within 7 different subregions of the DLH (Figures 6C,E, Table 1). Some of the BDA-positive fibers traveled in the raphe obscurus and the paramedian reticular nucleus nearly perpendicular to the plane of the sections (Figure 6C).

In most cases, after perifornical or lateral hypothalamic injections, BDA-positive fibers were identified among fibers of large neuronal tracts or fasciculi. Their number was low, except in the medial lemniscus, the medial longitudinal fascicle, and the superior cerebellar peduncle, where the density of BDA fibers could be rated as moderate (Table 1).

Several, more than one third of recognized lower brainstem areas/nuclei/subdivisions (Paxinos and Watson, 2007) were devoid of BDA-labeled fibers or terminals. They are not listed on Table 1.

This area consisted of 3 subdivisions. Each of them have been delineated and showed in one of the most popular topographical maps in the rat brain (Paxinos and Watson, 2007).

The posterior periventricular area is labeled as “Pe” (periventricular nucleus), expands from 2.24 to 3.48 mm caudal to the level of the bregma. This area is about 200–300 μm wide along the posterior portion of the third ventricle and arching over it (Figures 1D–F). This portion is called as “PVi” (intermediate hypothalamic periventricular nucleus) by Swanson (1987, 2004) and Hahn and Swanson (2010). The caudal part of the posterior periventricular area (between 2.92 and 3.48 mm) is also called as dorsal part of the posterior hypothalamic area (PHD) by Paxinos and Watson (2007). This is an area with relatively low neural cell density (Palkovits, 1975), cells are comprised of layers between the third ventricle and the dorsomedial nucleus.

The hypothalamic dorsomedial nucleus (Figures 1D–F) is made up of three subdivisions (dorsal, compact, and ventral). The medial and the dorsolateral portions of the dorsal division can be separated by their cellular density (Palkovits, 1975). The medial one is densely packed with neuronal cells, in contrast to the sparsely populated dorsolateral territory. This subdivision (designated as DMD in the atlas of Paxinos and Watson, 2007) extends rostro-caudally between 2.28 and 3.48 mm from the bregma level. Dorsolaterally, it is bordered by the dorsal hypothalamic area (see next).

The dorsomedial hypothalamic subdivision (signed also as dorsal hypothalamic area and labeled DMA by Bernardis and Bellinger, 1993 and DA by Paxinos and Watson, 2007) is comprised by various cell types with low cellular density. It can be separated from the zona incerta (dorsally), dorsomedial nucleus (ventromedially), the perifornical nucleus (ventrolaterally), and partly from the lateral hypothalamic area, laterally (Bernardis and Bellinger, 1993). It extends between 2.48 and 3.24 mm, rostrocaudally, from it is replaced by the most cellular dense posterior hypothalamic nucleus (Paxinos and Watson, 2007).

Originally, this area has been described as perifornical nucleus (Gurdjian, 1927; Christ, 1969; Bleier et al., 1979). Various cell types with high density are located over the fornix and in small areas around the fornix 2.64–3.60 mm caudal to the bregma (Paxinos and Watson, 2007). Actually, the perifornical area is larger than the perifornical nucleus, and extends all directions, occupying the less cell dense territories dorsomedially and dorsolaterally to the fornix. Correctly, the perifornical nucleus (signed PFn Palkovits, 1975, PFX by Geeraedts et al., 1990, PeF by Paxinos and Watson, 2007) consists only a part of the perifornical area (that is labeled PeFLH by Paxinos and Watson, 2007). The “nucleus” and the “area” can be clearly differentiated by their cellular density and somewhat by their cell types (Palkovits, 1975). The perifornical area extends 2.16–3.48 mm caudally to the bregma level (Paxinos and Watson, 2007). Dorsally, the perifornical area fuses with the medial part of the zona incerta. Medially, a relatively loose area separate the perifornical area from the dorsomedial nucleus and dorsomedial hypothalamic subdivision, like laterally from the lateral hypothalamic area (Palkovits, 1975). The perifornical area was divided rostral and caudal portions by Geeraedts et al. (1990).

The rostral perifornical subdivision (2.4 and 2.8 mm caudal to the level of the bregma; Figure 1G) contains relatively large cells (Palkovits, 1975) and higher cellular density. This subdivisions extends not only dorsally, but somewhat medial to the fornix (Paxinos and Watson, 2007).

The caudal part of the perifornical nucleus and area is larger and longer (2.8 and 3.4 mm caudal to the bregma) than the rostral one. It extends more dorsally, up to nearly the mamillothalamic tract. We divided the caudal part into medial and lateral parts by an imaginary line through the fornix (Figures 1H,I). The cellular density in the caudomedial perifornical subdivision is higher than in the caudolateral subdivision. Besides this, the caudomedial subdivision establishes more heterogeneous cell population (Palkovits, 1975).

This area constitutes a segment of the lateral hypothalamic area between 2.4 and 3.6 mm caudal to the bregma level. It is also called “peduncular part” of the lateral hypothalamus (PLH) by Paxinos and Watson (2007). Laterally, it extends to the subthalamic area, where it is bordered by the cerebral peduncle/internal capsule and the optic tract. It is bordered by the perifornical area, dorsomedially, and by the ventromedial nucleus, ventromedially. Dorsally, it is covered by the zona incerta and the Forel field (Paxinos and Watson, 2007). The area is used to be divided into a higher cell-dense ventral and a lesser cell-dense dorsal subdivisions of the lateral hypothalamic area (Geeraedts et al., 1990; Bernardis and Bellinger, 1993; Swanson, 2004; Paxinos and Watson, 2007). However, the actual borderline between these two areas is poorly defined. Two versions have been reported, or indicated in maps: (1) divided the ventral and dorsal parts by an imaginary line between the fornix and the midpoint of the medial borderline of the cerebral peduncle (Bernardis and Bellinger, 1993; Swanson, 2004). This is an oblique line moving upward medial to lateral direction; (2) imaginary horizontal line from the fornix, lateralwards (Geeraedts et al., 1990). We applied the first version in our study (Figures 1J–L).