INTRODUCTION

In the past three decades, oxytocin (OT) has been raised to the status of a neurotransmitter (Buijs and Van Heerikhuize, 1982; Sofroniew et al, 1981; Vandesande and Dierickx, 1975; see review by Muhlethaler et al (1985)). Beyond participating in the milk ejection reflex and parturition, this neuropeptide is now recognized for its influence on affiliative behaviors and the formation of social bonds (Insel, 1992), memory processing (see review by De Wied (1984)), modulation of stress (Ebner et al, 2005; Windle et al, 2004; Windle et al, 1997), and responses to fear stimuli (Kirsch et al, 2005). OT is a nonapeptide, belonging to the vasopressin-OT family, and is synthesized by neurons in the paraventricular nucleus (PVN) and supraoptic nucleus (SON) of the hypothalamus. These neurosecretory cells send projections to the posterior pituitary, where OT is released into the systemic circulation (Scharrer and Scharrer, 1940; Swanson and Kuypers, 1980). OT-labeled fibers have been identified in various brain regions, including the limbic system and brain stem, indicating direct central effects on specific regions (Caffe et al, 1989; Fliers et al, 1986; Fodor et al, 1992; Sofroniew, 1980).

OT's role in affiliative behaviors suggests it is important in social adaptation. Affiliation encompasses behaviors associated with parental care and reproduction, as well as behaviors that ultimately lead to close proximity of conspecifics, such as grooming. The demonstration of a differential distribution of OT receptors in the monogamous and highly parental prairie vole in comparison to the polygamous and nonsocial montane vole, brought to light the influence of OT on the expression of social behavior (Shapiro and Insel, 1992). In the prairie vole, ligand-binding studies show OT receptors in high concentrations in the ‘reward circuit’ sites, which are absent in the less affiliative montane vole. During the post-partum, when the montane vole displays maternal behaviors towards her offspring, there is a shift in the distribution of OT receptor binding towards the ‘reward circuit’ pattern seen in the prairie vole (Insel, 1992). This suggests that OT shapes rewarding aspects of affiliative behaviors necessary for survival in some species.

In addition to influencing reward circuits, OT also modulates the stress response. OT deficient animals display increased anxiety behaviors (Amico et al, 2004; Mantella et al, 2003), and release more corticosterone after a stressor (Amico et al, 2004). At the same time, OT administration reduces anxiety behaviors in typical stress paradigms, decreasing corticosterone release (Bale et al, 2001; Neumann et al, 2000; Windle et al, 2004, 1997).

A decline in social and ‘affiliative’ behaviors, along with altered stress responses, characterizes major depression (Holsboer et al, 1995; Nelson and Charney, 1981). Although OT is associated with enhanced social behavior and adaptation to stress, the serotonin and other monoamine systems are typically associated with amelioration of social isolation and anxiety found in mood disorders (see review by Castren (2005)). For example, these symptoms resolve after chronic treatment with serotonin reuptake inhibitors (SSRIs), which enhance serotonin transmission by blocking the serotonin transporter (5-HTT). 5-HTT is the molecule responsible for clearing serotonin released in the synapse (Blakely et al, 1994) and alterations in its expression have been linked to mood disorders (Arango et al, 2001; Kugaya et al, 2004). The fact that both OT and serotonin can modulate stress and affiliative responses naturally raises the question of how serotonin might influence the OT system, and the specific brain regions in which this occurs.

As a first step in seeing whether serotonin might exert its therapeutic effects via the OT system, we examined the anatomy of these systems in nonhuman primates. In the present study, we investigate the extent to which there is overlap of 5-HTT-immunoreactive (IR) fibers and OT-labeled cells in the primate hypothalamus utilizing immunocytochemical and double immunofluorescent techniques.

METHODS

One female (MFJ7) Macaque fascicularis and five male Macaque nemestrina (MNJ1, MNJ2, MNJ3, MNJ5, MNJ6) were used in these studies. All experiments were carried out in accordance with the National Institutes of Health guidelines. Experimental design and techniques were aimed at minimizing animal use and suffering, and were reviewed and approved by the University of Rochester Committee on Animal Research. Initial anesthesia was administered by an intramuscular injection of ketamine (10 mg/kg), and a deep anesthesia was induced by intravenous injection of pentobarbital (initial dose 20 mg/kg i.v. and maintained as needed). Animals were deeply anesthetized, and killed by perfusion through the heart with saline containing 0.5 ml of heparin sulfate, followed by 4% paraformaldehyde solution in 0.1 M phosphate buffer (pH 7.4). The brains were then removed, and cryoprotected in increasing gradients of sucrose (10, 20, and 30%). Serial sections of 50 μm were cut on a freezing microtome and transferred into 0.1 M phosphate buffer or cryoprotectant solution.

Immunocytochemistry

Sections from adjacent compartments were selected for immunostaining with OT (Immunostar, Hudson, WI, 1:10 000, rabbit polyclonal) and 5-HTT (MAB Technologies Inc., Stone Mountain, GA, mouse monoclonal 1:150 000) antibodies, and for cresyl violet staining. One compartment was also selected for 5-HT immunostaining for comparison with 5-HTT IR (Immunostar, Hudson, WI, 1:100,000, rabbit polyclonal). Dilution curves were performed before the actual experiments to determine the optimal concentration of each primary antibody. Control sections in which the primary antibodies were omitted resulted in no IR staining of cells or fibers. Sections were first thoroughly rinsed in 0.1 M phosphate buffer (pH 7.2) with 0.3% Triton X-100 (PB-T). Following treatment with endogenous peroxidase inhibitor (EPI), and more PB-T rinses, the tissue was preincubated in 10% normal goat serum (NGS) diluted with PB-T (NGS-PB-T) for 30 min. In the 5-HTT and 5-HT experiments, 1% bovine serum was added to the NGS to reduce background staining. Tissue was then placed in primary anti-sera to OT, 5-HTT, and 5-HT respectively, and incubated for four nights at 4°C. Following rinses with PB-T, the sections were incubated with the appropriate biotinylated secondary antibodies and visualized using the avidin-biotin complex reaction (Vectastatin ABC kit, Vector Labs, Burlingame, CA). Staining was enhanced by incubating sections for 1–3 min in 3, 3′diaminobenzidine tetrahydrochloride (DAB) and 0.03% hydrogen peroxide. Sections were mounted on subbed slides, dehydrated in increasing gradients of alcohol, cleared in xylene and cover slipped. Several immunostained compartments were counterstained with cresyl violet. Following analysis of single-labeled adjacent sections, additional compartments from one female (MFJ7) and two male (MFJ2 and JFJ5) monkeys were chosen for double labeling of OT and 5-HTT. For double-labeling using DAB, sections were first processed for OT and nickel-intensified to yield a blue–black reaction product in OT-labeled cells. Sections were secondarily processed for 5-HTT without nickel intensification to give a light brown labeling of 5-HTT-positive fibers.

Double-Immunofluorescence-Labeling Protocol

Dilution curves were first performed in single label immunofluorescence experiments to optimize fluorescent signal and minimize background staining. After treatment with EPI, and incubation in 10% NGS, selected sections were incubated for four nights in pooled OT (rabbit, 1:10 000) and 5-HTT (mouse, 1:75 000) antibodies at 4°C. Then, tissue was thoroughly rinsed in PB-TX, incubated in 10% NGS-PB-TX for 30 min, and incubated in the dark for four hours in filtered Alexafluor 546 (anti-rabbit IgG antibody) and Alexafluor 488 (anti-mouse IgG antibody) (Molecular Probes Inc., Eugene, OR), to react with the OT and 5-HTT primary antibodies, respectively. Control experiments in which the fluorescent secondary antibodies were ‘reversed’ (eg Alexafluor 546 anti-mouse IgG antibody to 5-HTT primary antibody; Alexafluor 488 anti-rabbit IgG antibody to the OT primary antibody) were also performed to ensure similar patterns of labeling. Tissue was then rinsed for 1 h, mounted on slides, and coverslipped in aqueous medium (Vectashield, Vector Labs).

Analysis

For DAB-immunoreacted compartments, the distribution of OT-labeled cells in the hypothalamus was hand-charted under bright-field illumination using a drawing tube attached to the microscope (× 10 objective). Hand-drawn charts were then entered into the computer using the application Canvas 5.0 (ACD Systems, Victoria, British Columbia, Canada) in conjunction with a drawing tablet (Wacom Technology Corp., Vancouver, WA). 5-HTT-labeled fibers were hand-charted under bright and dark field illumination, also using camera lucida techniques (× 10 objective). Paper charts of labeled fibers and surrounding landmarks, including fiber tracts and blood vessels, were then scanned into the computer using the program Adobe Photoshop 6.0 (300 dpi) (Adobe Inc., San Jose, CA) and imported into the drawing program Canvas 5.0. Finally, the boundaries of the PVN and SON in adjacent Nissl-stained sections were also charted, entered into the computer with the aid of a drawing tablet, and superimposed in transparent layers over charts of OT and 5-HTT-labeled sections. Landmarks such as blood vessels and fiber tracts were used for alignment. 5-HTT and OT-labeled sections counterstained with cresyl violet were examined to confirm these relationships. Sections immunoreacted with 5-HTT and counterstained with cresyl violet were used to assess the relative density of 5-HTT staining across subnuclei. Additional sections double-labeled for OT and serotonin transporter immunoreactivity (5-HTT-IR) using DAB techniques were examined under high power using oil immersion. After analysis of DAB-labeled sections, additional sections double-labeled for 5-HTT/OT with immunofluorescence antibodies were evaluated using fluorescent microscopy to further confirm the relationships among these markers. Images were captured with an Optronic Microfire color CCD and formatted in Adobe Photoshop. In addition, some sections were examined with confocal laser microscopy using Z-stack analysis (2 μm distance between slices) (Leica Microsystems Inc., Exton, PA).

RESULTS

Cytoarchitecture

Since OT-positive cells were mainly confined to the paraventricular and supraoptic nuclei, as previously described (see below), we examined the architecture of these nuclei in cresyl violet stained sections. Admixtures of small and large cells are found throughout the PVN, with relatively subtle differences across subdivisions as has been previously noted (Nauta and Haymaker, 1969; Saper, 1990). The five subnuclei recently described in the human PVN, based on cyto- and chemoarchitectural studies were also appreciated in the monkey (Figure 1). They include the anterior parvicellular subnucleus of the paraventricular nucleus (PaAP), the parvicellular subnucleus of the paraventricular nucleus (PaP), the magnocellular subnucleus of the paraventricular nucleus (PaM), dorsal subnucleus of the paraventricular nucleus (PaD), and posterior subnucleus of the paraventricular nucleus (PaPo) (Koutcherov et al, 2000). Owing to the similarities with the human PVN, we adopted this nomenclature. At the rostral-most part of the PVN, the small PaAP can be seen in the ventral region of the anterior hypothalamus (Figure 1a). It contains small and medium-sized neurons. Dorsal to it is the rostral tip of the PaD, which occupies a relatively large rostrocaudal expanse of the PVN and is seen in subsequent sections. The PaP, which lies medial to the PaAP and PaD, is also comprised of small cells and forms a thin band along the wall of the third ventricle (III) (Figure 1b). Caudal to the level of the PaAP, the PaM is seen (Figure 1b). It is characterized by large densely packed dark-staining cells in Nissl stained sections, but also contains small and medium-sized neurons. The PaD is continuous with the PaM dorsolaterally but is less prominent due to the lower packing density of its large dark-staining cells. Posterior to the descending column of the fornix (f), the wing-like PaPo progressively replaces the PaD, being the only part of the PVN left at the level of the mammillary nuclei (Figure 1c). It contains magnocellular neurons, along with large spindle-shaped cells and small cells.

Figure 1
figure 1

(a–c) Macroscopic images of coronal sections through the monkey hypothalamus, stained with cresyl violet and arranged in rostral-caudal order. Bar=50 μm.

Full size image

For the SON, we use nomenclature previously described by Nauta and Haymaker (1969) and Christ (1969). The SON follows the extent of the optic tract and is subdivided into two main regions, the dorsolateral SON (dlSON) and ventromedial SON (vmSON) subdivision (Christ, 1969; LuQui and Fox, 1976) (Figure 2). A third dorsomedial subdivision is seen in humans, however, we did not appreciate this in the monkey, consistent with previous studies (Christ, 1969). The dlSON appears at the level of the optic chiasm and is the only subdivision seen rostrally (Figure 2a, a′). It contains densely packed magnocellular neurons. At this level, the PVN and SON are bridged by clusters of large multipolar cells, consistent with descriptions of the ‘supraoptic accessory nucleus’ in humans (Saper, 1990) (Figure 2a′, arrow). Further caudal, in the ‘retrochiasmatic’ area, the dlSON moves laterally, and a smaller more diffuse ventromedial subdivision (vmSON) appears (Figure 2b, b′). The vmSON sends streams of cells dorsolaterally, which join the dlSON.

Figure 2
figure 2

The SON subdivisions at rostrocentral (a, a′) and caudal (b, b′) levels of the hypothalamus in coronal view. The dlSON emerges at the optic chiasm, and is composed of densely packed magnocellular neurons (a′ is high power image of boxed area in (a); bar=50 μm). The supraoptic accessory nucleus is seen dorsomedially, in the region between the SON and PVN (arrow). Posterior to the descussation of the optic chiasm, the SON is composed of the vmSON and dlSON. The vmSON is a diffuse collection of neurons embedded in fibers above the optic tract (b′, high power image of boxed area in (b); Bar=50 μm).

Full size image

Distribution of OT-IR Cells in the PVN and SON

OT-positive neurons are found in the PVN, SON and accessory SON (Figure 3). We classified their relative densities according to the relative number of cells per high-powered field (hpf, × 40 objective) across cases: + sparse (1–9 cells/hpf); ++ light (10–20 cells/hpf), +++ moderate (21–80 cells/hpf); ++++ high (>80 cells/hpf) (Table 1). There were no appreciable differences in the distributions of labeled cells across animals, nor was there a notable difference in the female animal compared to the males. Oxytocin immunoreactive (OT-IR) cells are mostly large, with darkly stained cytoplasm and prominent proximal dendritic processes (Figure 3e, large arrows), but we also found a number of medium-sized round cells in both the PVN and SON (small arrows).

Figure 3
figure 3

Schematic of the relative distribution of OT-labeled cells in hemi sections through the primate PVN/SON, arranged rostrocaudally. Arrow in (b) depicts the supraoptic accessory nucleus. Arrow in (c) indicates cascades of labeled fibers descending from the PaM and PaD to join OT-labeled fibers from the SON, which then enter the median eminence. 1 dot=1 cell. (e) Photomicrograph of OT-labeled cells in the PVN showing the typical spindle shaped cells (large arrow) and medium-sized, round cells (small arrow). Bar=50 μm.

Full size image
Table 1 Relative Densities of OT-Labeled Cells and 5-HTT-Labeled Fibers in the PVN and SON Subdivisions
Full size table

In general, OT-IR cells form a rostrocaudal continuum through the PVN (Figure 3 and Table 1). There are only occasional OT-positive cells in the PaAP, the most rostral PVN subdivision (Figures 3a and 4a), a light distribution of OT IR cells in the PaP (Figure 3a and b, 4b), and a moderate density of labeled cells in the rostral PaM (Figures 3b and c, 4c). The density of OT-labeled cells increases in the caudal PaM where it merges with the PaD. The highest density of labeled cells is found in the PaD (Figure 3c, 4c). Posterior to the descending column of the fornix, moderate concentrations of OT-labeled cells are identified in the PaPo (Figures 3d and 4e).

Figure 4
figure 4

Photomicrographs of OT-labeled cells in the PVN subdivisions (a–e) and the SON (f). (a) Sparse concentrations OT-IR cells in the PaAP. (b) OT-positive cells in the PaP along the ventricle. (c) The PaM contains moderate densities of OT-labeled cells. (d) Dense concentrations of OT-IR cells in the PaD. (e) OT-labeled cells in the wing-like PaPo. Bar in (a–e)=50 μm. (f) OT-labeled cells form a ‘cap’ over putative vasopressinergic cells of the dlSON core (area without label, asterisk). Bar=100 μm.

Full size image

The SON contains many OT-labeled cells, which tend to cluster medially in the dlSON (Figure 3a and b). In the retrochiasmatic area, where the SON is subdivided, OT-labeled cells are seen in the vmSON, and flow dorsolaterally in a thin stream to encircle the dlSON, forming a ‘cap’ around the nucleus, avoiding its center (Figures 3d, 4f). Most OT-positive cells in the dlSON are bipolar, whereas relatively smaller, multipolar cells are found in the vmSON.

Distribution of 5-HTT-IR Fibers in the PVN and SON and Overlap with OT+ Cells

5-HTT-containing fibers are widely distributed throughout the primate hypothalamus (Table 1). Overall patterns of 5-HTT-IR were similar to those achieved with 5-HT labeling. However, the fine details of fiber labeling were much harder to appreciate with the latter, since 5-HT is also found in extracellular spaces which can obscure fiber labeling. The relative densities of 5-HTT-labeled fibers are classified based on their distribution throughout the entire hypothalamus (Emiliano et al, in preparation). We classified the densities of fine, varicose 5-HTT-positive fibers based on the degree to which they obscured underlying cellular structures in counterstained sections. Scattered labeled fibers were classified as ‘sparse’ (+); a fine meshwork of labeled fibers with most underlying cells clearly visible was classified as ‘light’ (++); a moderate, relatively homogeneous meshwork of labeled fibers, partially obscuring underlying cells was classified as ‘moderate’ +++; and a high, homogeneous density of labeled fibers that almost completely obscured underlying Nissl stained structures was classified as ‘heavy’ (++++). Based on this classification, concentrations of labeled fibers in the PVN and SON range from sparse to moderate (Table 1 and Figure 5a–d). There was no obvious difference in 5-HTT distribution between the male animals and the one female. The majority of 5-HTT-containing fibers are thin and finely beaded. Thick, non-beaded fibers coursing through the median eminence and in the median forebrain bundle were interpreted as fibers of passage and were not charted.

Figure 5
figure 5

Schematic of the relative distribution of 5-HTT-labeled fibers in hemi-sections through the primate PVN/SON, arranged rostrocaudally.

Full size image

Sparse densities of 5-HTT-labeled fibers are found in the PaAP, and there is a light density of labeled fibers in the PaP (Figure 5a and b). Slightly caudal to this level, moderate densities of labeled fibers occupy the PaP and the rostral PaM. Moderate densities of labeled fibers occupy the PaM and PaD (Figure 5b and c). In caudal-most sections, the PaPo contains a moderate concentration of labeled fibers (Figure 5d). The dlSON displays moderate density of labeled fibers in the dorsal capsular area, with sparse concentrations of 5-HTT-IR fibers in the central core regions (Figure 5a and b). Moderate concentrations of 5-HTT-IR fibers are found caudally in the vmSON and ‘capsular’ dlSON (Figure 5c).

Overall, 5-HTT-labeled fibers and OT-IR cells overlap in parallel degrees across PVN and SON subdivisions (Table 1). In general, all PVN subdivisions containing at least moderate concentrations of OT-positive cells—PaM, PaD, and PaPo—also have modest concentrations of 5-HTT-labeled fibers (Figure 6c and d and Figure 7a–c). These ‘magnocellular’ subdivisions of the primate PVN thus share similar densities of innervation by 5-HTT-labeled axons. Confocal analysis of these regions revealed 5-HTT-labeled varicose fibers closely apposed OT-IR soma and proximal dendrites (Figure 7g, see also Figure 6e and f). In contrast, the PaAP and PaP, which have relatively fewer concentrations of OT-labeled cells, have only a sparse and light distribution, respectively, of 5-HTT-positive fibers (Figure 6a and b). Similarly, in the SON, subdivisions containing the highest densities of OT-labeled cells (vmSON and capsular dlSON) also receive a moderate concentration of 5-HTT-labeled fibers. Subregions relatively devoid of OT-labeled cells (core of the dlSON) receive only a sparse innervation of 5-HTT-labeled fibers. Under high power confocal microscopy, the close relationship of 5-HTT-labeled boutons and OT-labeled cells in the ‘capsular’ dlSON is seen, suggesting synaptic contacts (Figure 7).

Figure 6
figure 6

Immunofluorescent labeling in adjacent sections. (a, b) 5-HTT-IR (a) and OT-IR (b) in the PaAP. The PaAP contains minimal concentrations of 5-HTT-IR fibers and scant OT-positive cells. Bar=50 μm. (c, d) 5-HTT-IR (c) and OT-IR (d) in the PaM and PaD. OT-labeled cells in these subdivisions are moderate to dense, while 5-HTT-IR fibers are of a moderate concentration (asterisks mark blood vessels). Bar=100 μm. (e, f) Double-labeled sections in high power view showing fine varicose 5-HTT-labeled fibers in the PaD in close proximity to OT-positive cells. Black arrows indicate thin 5-HTT-labeled fibers, which can be distinguished from thick beaded OT-containing processes (open arrow). Bar=50 μm.

Full size image
Figure 7
figure 7

Double immunofluorescent labeling. (a–c) Photomicrograph of region in the PaD, labeled for OT-IR (a, red) and 5-HTT-IR (b, green). Merged image (c) shows OT-labeled cell bodies (red, arrows) embedded in fine meshwork of 5-HTT-labeled fibers (green). Bar=50 μm. (d–f) Macroscopic image of the capsular dlSON as it abuts the optic tract. It contains OT-IR cell clusters (a, red), which are overlapped by 5-HTT-labeled fibers (b, green; c, merged image). Bar=100 μm. (g, h) High power confocal images in the PaD (g), and capsular dlSON (h) illustrating the close apposition of 5-HTT-labeled fibers to OT-IR cells (arrows). Bar=50 μm.

Full size image

DISCUSSION

Summary of Findings and Comparison with Previous Primate Studies

Cytoarchitecture of the PVN/SON

The organization of the monkey PVN and SON has several similarities to its human counterpart. Consistent with previous studies we found that there are no clear cellular boundaries among PVN subdivisions, with similar cell types crossing several subcompartments (Christ, 1969; Koutcherov et al, 2000; Nauta and Haymaker, 1969; Saper, 1990). Thus, for example, the PaM, PaD, and PaPo, all contain magnocellular neurons, indicating that the PaM is not the only ‘magnocellular’ subdivision. The SON is composed of a large, relatively compact dorsolateral subdivision (dlSON), and a diffuse, slightly caudal ventromedial subregion (vmSON). We found that the monkey lacks the dorsomedial subdivision (‘dmSON’), as found in the human (Christ, 1969).

Distribution of OT-containing cells

While the range of nomenclatures for the primate PVN subregions makes comparison of studies difficult, the distribution of OT-labeled cells is consistent across studies (Caffe et al, 1989; Dierickx and Vandesande, 1977; Kawata and Sano, 1982; Koutcherov et al, 2000; Saper, 1990) (present results). The OT-positive cells form a ventral-to-dorsal trajectory along the rostrocaudal extent of the PVN, traversing several subdivisions. The rostral PaAP contains only scattered OT-labeled neurons, while the PaP contains a light distribution of OT-labeled cells. OT-IR cells are most highly concentrated in caudal ‘magnocelluar’ subdivisions (the PaM, PaD, and PaPo) with the highest concentration in the PaD. In the SON, the distribution of OT-labeled cells is highest in the vmSON and dorsal ‘capsular’ subregions of the dlSON. These distributions are consistent with results in previous human and nonhuman primate studies (Caffe et al, 1989; Dierickx and Vandesande, 1977; Fliers et al, 1985; Kawata and Sano, 1982).

Distribution of 5-HTT-labeled fibers and overlap with OT-positive cells

The relative densities of 5-HTT-labeled fibers follow the distribution of OT-labeled cells in both the PVN and SON. The PaAP has relatively sparse concentrations of 5-HTT-labeled fibers/OT-labeled cells, the PaP has light concentrations of 5-HTT-labeled fibers/OT-labeled cells, and moderate concentrations of labeled fibers overlap the PaM, PaD, and PaPo, which contain moderate to heavy concentrations of labeled cells. In the SON, the dorsal ‘cap’ of the dlSON and vmSON, which have heavy and moderate concentrations of OT-labeled cells respectively, are innervated by a moderate concentration of 5-HTT-labeled fibers. In contrast, the ventral dlSON (OT-negative) is relatively devoid of labeled fibers.

A single previous study shows a moderate concentration of 5-HT-labeled fibers in the Macaque PVN, with a relatively light concentration of labeled fibers in the dlSON, and a higher concentration of labeled fibers in the vmSON (‘nucleus supraopticus difusus’) (Kawata and Sano, 1982). While the PVN and SON subdivisions were not defined in that study, the overall pattern of 5-HT-IR is consistent with the present results. Our results are also in agreement with earlier studies using immunofluorescent and biochemical techniques, which report moderate levels of 5-HT labeling the region of the primate PVN (Azmitia and Gannon, 1986; Nobin and Bjorklund, 1973; Schofield and Everitt, 1981).

In rats, immunofluorescence studies indicate a moderately high number of serotonin-containing fibers in the rat PVN (Kent and Sladek, 1978). However, more recent studies utilizing 5-HT immunocytochemistry show relatively low concentrations of 5-HT-labeled fibers in the PVN and SON of that species (Sawchenko et al, 1983; Steinbusch and Nieuwenhuys, 1981) (Emiliano, unpublished observations). However, a preferential input of serotonergic fibers to oxytocinergic regions has been noted, similar to the current findings (Sawchenko et al, 1983).

5-HTT/OT overlap as an anatomic substrate for stimulation of OT: Despite an apparent paucity of 5-HT-IR fibers in the rat PVN and SON, numerous pharmacologic studies support this connection. Moreover, the raphe projects to both nuclei (Sawchenko et al, 1983; Tribollet and Dreifuss, 1981). Serotonin receptor stimulation in the PVN causes the release of OT in male and female animals (Bagdy, 1996; Jorgensen et al, 2003b; Moos and Richard, 1983; Saydoff et al, 1991; Van de Kar et al, 1995). Administration of the serotonin-releasing agent D-fenfluramine, results in c-fos expression in OT-containing neurons of both the PVN and SON, indicating that serotonergic transmission activates OT neurons in both those nuclei (Javed et al, 1999). Furthermore, serotonin agonists increase OT mRNA expression in the PVN and SON, suggesting a long lasting effect on OT levels (Jorgensen et al, 2003a). In the present study, the anatomic overlap of 5-HTT-labeled fibers and OT-containing cells in the PVN and SON suggests that serotonin can directly influence the OT system in the primate. While there are few pharmacologic or behavioral studies in monkeys or humans, a recent study shows that OT release is stimulated by fenfluramine administration in healthy men (Lee et al, 2003). An indirect piece of evidence for 5-HT-OT interactions in humans is the well-known prolactin surge associated with fenfluramine challenge (Coccaro et al, 1997; Siever et al, 1984). OT-containing neurons are among the possible mediators of this phenomenon, since OT is a potent prolactin releasing factor (see review by Emiliano and Fudge (2004)).

Mother–infant behaviors as a model of 5-HT/OT interaction: Maternal nursing behavior is one of the few behavioral models of 5-HTT-OT interactions and provides clues on how these connections may promote affiliation and decrease stress. Stimulation of sensory afferents in the nipple are transmitted through the dorsal root ganglia and dorsal horns to the lateral funiculus and the brain stem, activating serotonin cells in the raphe nuclei (Bodnar et al, 2002; Dubois-Dauphin et al, 1985a, 1985b; Fukuoka et al, 1984; Tasker et al, 1986; Tribollet and Dreifuss, 1981). Serotonergic fibers stimulate OT cells in the PVN, which in turn promote prolactin release (Moos and Richard, 1983), based on lesioning and pharmacologic studies (Bodnar et al, 2002; Saydoff et al, 1991). While the release of OT in the systemic circulation causes the ‘let-down’ of milk, stimulation of central OT may reduce anxiety and stimulate reward circuits. Decreased HPA-axis responsiveness to stress has been demonstrated in both lactating animals and humans. For example, cortisol levels are reduced in breast-feeding women in paradigms of social and running stress (Altemus et al, 1995; Heinrichs et al, 2001). This decreased responsiveness to stress has been attributed to OT, as well as to prolactin and other hormones (Carter and Altemus, 1997; Lightman et al, 1993) and may represent an adaptive mechanism to ensure that this important affiliative behavior will remain protected from disruptions. Exogenous OT administration to non-nursing human volunteers also decreases basal and stimulated ACTH and cortisol plasma levels (Chiodera and Coiro, 1987; Coiro et al, 1988; Legros et al, 1982).

While attenuating the stress response, OT also activates reward circuits. Nursing rats activate ‘reward’ regions of the brain in response to pup contact, in contrast to virgin rats. Moreover, nursing dams prefer pup contact to classic drug reward (Ferris et al, 2005). In normal non-nursing animals, exogenous OT administration activates similar reward sites, suggesting that OT is an important chemical mediator of the rewarding effects of nursing (Febo and Ferris, 2005). Taken together, these studies show that OT release through the serotonin-mediated suckling mechanism is a potent activator of reward circuits as well as an inhibitor of HPA-axis-mediated stress responses. To the extent that serotonin activates central OT release, it indirectly influences the reward properties of this affiliative behavior and decreases the stress response.

Functional implications

A core symptom of depression is diminished pleasure and interest (Naranjo et al, 2001; Nelson and Charney, 1981), which is effectively treated by serotonin enhancing drugs. It is possible that the efficacy of SSRIs in restoring interest in social interactions is due, in part, to their actions on the reward circuit via the OT system (Insel, 2003; Swaab et al, 2000; Uvnas-Moberg et al, 1999). SSRIs also attenuate anxiety and depressive behavior, an effect that may also be partly mediated by downstream OT release. While serotonin administration leads to acute changes in OT release and cell activation (Vacher et al, 2002; Fraga et al, 2005; Jensen et al, 1999; Van de Kar et al, 2001), it also results in upregulation of OT mRNA suggesting a delayed but perhaps prolonged effect on basal OT levels (Jorgensen et al, 2003a).

Besides a possible involvement in the amelioration of mood symptoms seen in patients treated with SSRIs, available evidence suggests that serotonin enhancement of OT would play a role in the improvement of sexual functioning seen in recovering depression. In rodents, OT is a potent inducer of penile erection when injected into the lateral ventricle or PVN, a phenomenon that is blocked by the administration of OT antagonists (Arletti et al, 1985). OT facilitates sexual behavior in male and female rats, enhancing receptivity and copulatory performance (Arletti and Bertolini, 1985; Caldwell et al, 1986). Projections from OT cells in the PVN to sacral parasympathetic neurons involved in penile erection (Andersson and Wagner, 1995; Tang et al, 1998; Veronneau-Longueville et al, 1999) provide anatomic substantiation for OT's role in sexual behavior. The sexual dysfunction associated with SSRIs is significant, however (Fava and Rankin, 2002; Rosen et al, 1999), and multiple mechanisms for this have been proposed (see review by Rosen et al, (1999)). It is possible that SSRI-stimulated OT expression may also drive prolactin release in some individuals, a phenomenon that is associated with sexual dysfunction (see review by Emiliano and Fudge (2004)).

The present results provide anatomic support for the idea that serotonin mediates OT responses in higher primates, which in turn may play a role in symptom relief. Future studies aimed at detailing the input/output pathways of this system, serotonin receptor profiles associated with the OT-containing cells, and the ability of SSRIs to increase OT mRNA and circulating/central OT levels will help test these ideas in primate models.