Neurochemical Mediation of Affiliation and Aggression Associated with Pair Bonding

Abstract

Background

The neuropeptides vasopressin (AVP) and corticotrophin-releasing hormone (CRH) facilitate while serotonin (5-HT) inhibits aggression. How the brain is wired to coordinate interactions between these functionally opposed neurotransmitters to control behavioral states is poorly understood.

Methods

Pair-bonded male prairie voles (Microtus ochrogaster) were infused with a retrograde tracer, fluorogold, and tested for affiliation and aggression toward a female partner or novel female. Subsequent immunocytochemical experiments examined neuronal activation using FOS and neuro-chemical/-receptor profiles on brain areas involved in the behaviors. Finally, a series of behavioral pharmacological and real-time in-vivo brain microdialysis experiments were performed on male prairie voles displaying affiliation or aggression.

Results

We localized a subpopulation of excitatory AVP neurons in the anterior hypothalamic nucleus (AH) that may gate CRH output from the amygdala to the AH and then the lateral septum to modulate aggression associated with mate guarding. Conversely, we identified a subset of inhibitory 5-HT-ergic projection neurons in the dorsal raphe nucleus to the AH that mediates the spatio-temporal release of neuropeptides and their interactions in modulating aggression and affiliation.

Conclusions

Together, this study establishes the medial extended amygdala as a major neural substrate regulating the switch between positive and negative affective states wherein several neurochemicals converge and interact to coordinate divergent social behaviors.

INTRODUCTION

A critical challenge in the psychiatry field is to determine the neurochemical circuitry underlying an individual’s propensity to transition between prosocial emotional states to physical violence (1). Although preclinical neuroscience has largely focused on examining the function of individual neurochemicals, brain areas, and neuronal mechanisms therein, we know surprisingly little about the neuromodulatory microcircuits regulating emotion (2).

The posterior dorsal medial amygdala (MeAPD) projects to several subdivisions of the hypothalamus (3–5) to regulate various forms of social behavior (3–10). However, the circuitry remains largely undefined beyond these second-order projections. The integrating “command” centers that process sensory input and control descending motor output to program socio-emotional behavior is unclear. Previous work has relied on using traditional laboratory rodents to dissect the neural circuitry involved. However, these animals do not readily display certain types of behavior and may not be appropriate for some investigations (11). For example, most laboratory animals do not exhibit strong social bonds between mates and males and typically do not display paternal behavior or female-directed aggression (12). Because mating naturally induces these behaviors in the socially monogamous prairie vole (Microtus ochrogaster), this rodent species represents a unique animal model to investigate neural circuitry programming pair-bonds (12, 13).

Lesions of the vomeronasal organ (VNO) (14) or MeAPD (15) impair partner preference formation and affiliation in prairie voles. In males, parvocellular AVP neurons in the nucleus circularis (NC) and medial supraoptic nucleus (mSON) are both recruited during aggression (16) and release their contents in the anterior hypothalamic nucleus (AH) activating vasopressin (AVP) V1a-type receptors (V1aRs) to facilitate aggression selectively toward novel conspecifics but not a partner (17). Two weeks of socio-sexual experience also induce structural plasticity of V1aRs to mediate selective aggression (17). Further, viral-vector-mediated gene transfer of V1aRs into the AH, of sexually naïve males, recapitulates pair-bonding-induced aggression (17). Finally, dopamine (DA) signaling in the rostral nucleus accumbens shell (NAcc) is also involved in selective aggression to maintain monogamous pair-bonds (18). However, despite these studies, we know little about how these brain regions, genes, and neurochemicals integrate into a network to control pair-bonding behavior (19).

Because recent work demonstrates regional overlap of molecularly specified neurons in the ventral medial hypothalamus (VMH) that control properties characteristic of emotion states regulating social (20), sexual (21, 22), and aggressive (21–25) behaviors, we investigated whether individual pair-bonding behaviors are encoded via similar or different neuronal systems. Here, we focused on examining the neurotransmitters AVP, corticotrophin releasing hormone (CRH), and serotonin (5-HT) for their roles in regulating behavioral states. We proposed that AVP/CRH facilitate aggression while 5-HT functions to inhibit the activity of the AVP/CRH systems in the AH to “switch” from aggression to affiliation. Our data provide necessary refinement steps towards understanding how multiple neurotransmitter systems interact within neuronal microcircuits to drive male-female attachment.

METHODS AND MATERIALS

Subjects

Subjects were male prairie voles (90–120 days of age) that were either sexually naive or pair-bonded with a female for two weeks which reliably induces partner preferences and selective aggression toward novel conspecifics (16–18) (see supplementary experimental procedures). All experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) at the Florida State University.

Behavioral Assays

Subject’s aggressive behaviors were examined using the resident-intruder test (RIT), a well characterized and ethologically valid model of offensive aggression (26). Briefly, a conspecific intruder was introduced into the home cage of the subject (resident), and the resident was scored for 10 mins for aggressive responses including the frequency of lunges, bites, and chases as well as the duration of affiliative side-by-side contact and anogenital investigation, as previously described (16, 17) (see supplementary experimental procedures).

Monosynaptic Tracer Injection Parameters

Subjects were stereotaxically injected into the AH (coordinates from bregma: posterior .55 mm, lateral ± 0.75 mm, ventral 6.1 mm), rostral NAcc shell (anterior 1.60 mm, lateral ± 1.0 mm, ventral 4.5 mm), lateral septum (LS) (anterior .80 mm, lateral ± 0.61 mm, ventral 4.1 mm), or MeAPD (posterior 1.30 mm, lateral ± 2.70 mm, ventral 7.0 mm), respectively, with glass capillary micropipettes (A–M Systems, Inc., Carlsborg, WA) filled with 2% fluorogold (FG; Fluorochrome, Englewood, CA) and 0.5% cresyl violet dye in 0.01 M phosphate buffer solution (PBS; pH 7.4) under sodium pentobarbital (0.1mg/10g body weight). Injection placement was evaluated by processing sections spanning the target area for FG immunocytochemical detection and cresyl violet dye spread. Data from the subjects with correct injection placement were included in neuroanatomical mapping (Figure S3) (see supplementary experimental procedures).

Brain Microdissection and High-Performance Liquid Chromatography with Electrochemical Detection (HPLC-ECD) Analysis

Coronal brain sections (300-μm) were cut on a cryostat and frost mounted onto microscope slides. Bilateral tissue punches were taken using a 1-mm diameter scalpel under 20-X magnification on a Leica DMRB dissection microscope. Tissue samples were localized to the AH, medial preoptic area (MPOA), and paraventricular nucleus of the hypothalamus (PVN), and stored at −80°C. Subsequently, 5-HT and its metabolite 5-hydroxyindoleacetic acid (5-HIAA) were measured using HPLC-ECD (see supplementary experimental procedures).

Intra-AH Stereotaxic Cannulation and Drug Microinfusion

Subjects were anesthetized with sodium pentobarbital (0.1mg/10g body weight) and then stereotaxically implanted with guide cannula aimed at the AH, as described previously (17, 18). All injections were made using a Hamilton-syringe connected to an automatic micropump. Immediately after a 10 min RIT test, subjects were overdosed with sodium pentobarbital, rapidly decapitated, and their brains were sectioned for histological verification of cannula placement. Subjects with correct cannula placement were included in data analysis (Figure S2 & supplementary experimental procedures).

Brain Preparation, Immunocytochemistry, and Image Analysis

Subjects were anesthetized with sodium pentobarbital and then perfused through the ascending aorta with 0.9% saline, followed by 4% paraformaldehyde in 0.1M PBS. Brains were dissected, post-fixed for 2 hours in 4% paraformaldehyde and then stored in 30% sucrose in PBS. Brains were cut into 30-μm coronal sections on a freezing microtome and floating sections were stored in 0.1M PBS with 1% sodium azide at 4°C until immunostaining.

Different sets of floating brain sections at 150-μm intervals were processed for single- or double-immunoreactive (ir) labeling of fluorogold (FG), FOS, FG/FOS, FG/TH (tyrosine hydroxylase), FG/AVP, FG/5-HT, or FG/CRH. AH sections were processed for double- or triple-ir labeling for AVP, V1aR, 5-HT, 5-HTr1a, CRH, CRHR2, FG, and FOS.

We quantified the co-localization of 5-HTr1a, V1aR, CRHR2, and 5-HT on AVP-, CRH-, and/or FG-expressing neurons in the AH. Leica imaging software profile methods of cell counting were employed and area measurements (square millimeters) were taken on each section analyzed to determine cell densities. Photomicrographs were captured by using a Zeiss Axioskop 2 (Carl Zeiss) microscope with a SPOT RT Slider (Diagnostic Instruments) camera and SPOTTM (version 3.0.6) software. Image files were then stored and subsequently analyzed (see supplementary experimental procedures).

Real-Time In-Vivo Brain Microdialysis with Neurochemical Analyses

Microdialysis probe construction, cannulation, and dialysate collection were previously described (17, 27, 28) (see supplementary experimental procedures). Immediately after RIT, subjects were overdosed with sodium pentobarbital, rapidly decapitated, and their brains were sectioned for histological verification of probe placement. Subjects with correct probe placement in the AH were included in data analysis. Microdialysis samples were processed for AVP and CRH contents using standard ELISA kits and 5-HT using HPLC-ECD.

RESULTS

Neuronal Activation Associated with Opposing Behavioral States

To establish a neural framework of the circuitry associated with individual pair-bonding behaviors, we performed affiliation and aggression assays in males injected with FG. Males displayed offensive aggression toward novel females and social affiliation with their female partner. These robust patterns of selective aggression and affiliation were observed in each of the four tracing groups (Figure 1A & B). No group differences were found in general locomotor activity, social interest, exploration, defense, or courtship behaviors (Tables S1a–d).

Figure 1. Differential Limbic Circuit Activation Associated with Affiliation and Aggression.

(A) Male prairie voles that pair-bonded with a female for two weeks displayed robust aggression against a novel female but (B) high levels of affiliation toward their familiar female partner. This behavioral pattern was consistent across animals that received FG injections into the lateral septum (LS), nucleus accumbens (NAcc), anterior hypothalamic nucleus (AH) or posterior dorsal medial amygdala (MeAPD) with significant main (F_(2,4,20) = 32.58, p < 0.001 for aggression and F_(2,4,20) = 37.29, p < 0.001 for affiliation), but not interaction (F_(2,4,20) = 1.37, ns, for aggression and F_(2,4,20) = 1.79, ns, for affiliation), effects. In these four FG injection groups, aggression against a novel female induced a significant increase in the density of FOS-ir neurons in the AH (C & D) and MeAPD (C–E) while affiliation induced a significant increase in the density of FOS-ir neurons in the DR compared to controls with significant main (F_(2,4,20) = 6.85, p < 0.05 for aggression and F_(2,4,20) = 7.64, p < 0.05 for affiliation), but not interaction (F_(2,4,20) = 1.13, ns, for aggression and F_(2,4,20) = 1.32, ns, for affiliation), effects. In addition, aggression induced a significant increase in the density of neurons double-labeled for FOS-ir/FG-ir in the AH (F) and MeAPD (H) compared to affiliation or controls with significant main (F_(2,4,20) = 5.93, p < 0.05), but not interaction (F_(2,4,20) = 1.68, ns), effects. Conversely, males displaying affiliation toward their female partner had an increased density of FOS-ir/FG-ir double-labeling in the dorsal raphe nucleus (DR) compared to males displaying aggression against a novel female and controls with significant main (F_(2,4,20) = 5.43, p < 0.05), but not interaction (F_(2,4,20) = 1.09, ns), effects (H). Light-field photomicrographs (30μm stack) of neurons labeled for FG-ir (brown cytoplasmic staining), FOS-ir (black nuclear staining) or both in the AH (FG injected into LS; I), MeAPD (FG injected into AH; J), and DR (FG injected into AH; K) of males exposed to a stranger female. The open black circles shown in panels I, J, and K depicts the area at higher magnification (5μm stack) in the inset. F: fornix; OT: optic tract; CA: cerebral aqueduct. Bars indicate means ± standard error of the mean. Asterisks (*) indicate, **: p < 0.01 (A & B). Bars labeled with different letters (C–H) differ significantly by post hoc Student Newman-Keuls (SNK) tests of significance examining both main effects and interactions with analysis of variance p value set to < .05. Scale bar = 100μm. The insert within each panel shows neurons double-labeled for FOS-ir/FG-ir, while scale bar = 10μm. See also Tables S1.

The stereological parameters, anatomical coordinates, and abbreviations for each brain area quantified are summarized in Table S2. Stereological quantification found no significant differences in the density of FG-ir neurons among tracing groups indicating consistent microinjection volumes (Table S3). We have previously identified subsets of neurons selectively activated by the expression of affiliation and aggression (16). Therefore, we focused on this distinct subpopulation by using FOS, the protein product of an immediate early gene, c-fos, to assess neuronal activation in retrogradely-labeled projection neurons recruited during affiliation or aggression. We added a baseline control group of handled males that were not exposed to social stimuli during RIT. Males displaying aggression toward a novel female showed a significantly higher density of FOS-ir in the AH and MeAPD than males displaying affiliation which, in turn, showed a higher density of FOS-ir than controls, and this pattern of FOS-ir was consistent across all tracing groups (Figure 1C–E & I–K). There was a significantly higher density of FG-ir/FOS-ir double-labeled neurons in the AH projecting to the LS (Figure 1F & I) and in the MeAPD projecting to the AH (Figure 1H & J) in males displaying aggression compared to males displaying affiliation and controls. Conversely, there was a significantly higher density of FG-ir/FOS-ir double-labeled neurons in the dorsal raphe nucleus (DR) projecting to the AH in males displaying affiliation than in males displaying aggression or controls (Figure 1H & K).

Neurochemical Microcircuit Connectivity

Multiple-label immunofluorescence experiments were performed to identify the cytochemical phenotypes of FG-ir projection neurons recruited during affiliation (DR-AH) and aggression (MeAPD-AH-LS). FG-ir neurons in the AH or MeAPD did not co-express AVP, tyrosine hydroxylase (TH), gamma-Aminobutyric acid (GABA), glutamate, or oxytocin, but stained positively for CRH (Table S4). Because the data above revealed activation of a MeAPD-AH-LS circuit during aggression, we focused on stereologically quantifying the percentage of FG-ir/CRH-ir double-labeled neurons in the AH and MeAPD from the LS and AH tracing groups. Thirty-nine percent of the total number of FOS-ir/FG-ir neurons in the AH projecting to the LS and eight percent of the total number of FOS-ir/FG-ir neurons in the MeAPD projecting to the AH, of aggressive males, stained positively for CRH (Table 1). Although we found FG-ir and CRH-ir cells in the paraventricular nucleus of the hypothalamus (PVN) and bed nucleus of the stria terminalis (BNST), across each injection group, we did not see any FG-ir/CRH-ir colocalized cells.

Table 1.

Neurochemical Microcircuit Phenotyping

FG Injection	Area	Markers	# Cells	%
LS	AHN	CRH/FOS/FG	152	39*
AHN	MeAPD	CRH/FOS/FG	66	8*
AHN	DRN	5-HT/FOS/FG	58	16*
NAcc	VTA	TH/FG	774	27**

^*Percent of the total FOS-ir/FG-ir double-labeled cells.

^**Percent of the total FG-ir cells.

Further, previous mapping studies indicated the source of 5-HT innervation to the mSON- and PVN-AVP systems in the hypothalamus (30). Here we found that sixteen percent of the total number of FOS-ir/FG-ir neurons in the DR projected to the AH and co-expressed 5-HT in bonded males displaying affiliation (Table 1). Finally, because the catecholaminergic circuit from the ventral tegmental area (VTA) to NAcc has been well established in a variety of species including voles (31, 32), we processed sections spanning the VTA taken from the NAcc FG injection group for TH and FG double-labeling, as an internal control for validating our retrograde tract-tracing methods. Twenty-seven percent of the FG-ir neurons in the VTA projecting to the NAcc co-expressed TH (Table 1), implicating a population of mesolimbic DA-ergic neurons extending from the VTA to the NAcc (33, 34). Together, our data unraveled a novel circuit associated with affective behavior, outside the HPA axis (35).

5-HT and AVP Modulation of Selective Aggression via 5-HT1a Receptor Activation

AVP in the AH mediates offensive aggression in rodent species (36) including prairie voles (17) and this AVP effect is attenuated by activation of the 5-HT system (37, 38). Results from our tract-tracing data suggested that a DR-AH 5-HT circuit was activated during affiliation but not aggression. Therefore, we tested the hypothesis that 5-HT in the AH mediates AVP- or pair-bonding-induced aggression.

To increase accumulation of extracellular 5-HT we used fluoxetine, a commonly used selective serotonin reuptake inhibitor (SSRI), which blocks mating-induced aggression in male prairie voles (39) and agonistic behavior in other species (40–47). Results from meta-analytic studies demonstrated that increased 5-HT had the strongest inhibitory effect on aggression in rodent species when aggression was offensive, fluoxetine was used, injection was intra-peritoneal (i.p.), and treatment was acute (48). Therefore, we followed a similar SSRI injection and dosing regimen.

To induce aggression in sexually naïve males we used intra-AH administration of AVP (500ng/side) that induces offensive aggression in male prairie voles (17). Sexually naïve males that received bilateral intra-AH injections of AVP (in 200 nl CSF) were divided into three groups that received i.p. injections of saline or saline containing a low (1mg/kg) or high (6mg/kg) dose of fluoxetine followed by a 10-min RIT toward a novel female. Both doses of fluoxetine blocked AH-AVP-induced aggression (Figure S1A) while the high dose of fluoxetine also increased affiliation (Figure S1B) relative to saline controls. Fluoxetine treatment did not affect general locomotor activity, social interest, exploration, defense, or courtship behaviors (Table S5).

To assess the effect of fluoxetine treatment on 5-HT activity in the brain, three groups of sexually naïve male prairie voles were injected (i.p.) with saline, a low (1mg/kg) or high (6mg/kg) dose of fluoxetine, and their brain tissue was micro-dissected (Figure S1F) for 5-HT and its metabolite 5-hydroxyindoleacetic acid (5-HIAA) measurement using HPLC-ECD. Treatment of fluoxetine at the high dose increased levels of 5-HT (Figure S1C) and 5-HIAA (Figure S1D) in the AH. Both fluoxetine doses decreased 5-HT turnover indicated by a low 5-HIAA/5-HT ratio in the AH (Figure S1E). This effect of fluoxetine on 5-HT turnover has been corroborated in previous work (39, 49–51).

We then focused on the AH to determine the receptor-specific role of 5-HT attenuating AH-AVP-induced aggression. We focused on 5-HT1a-type receptors (5-HTr1a) because their anti-aggressive properties (52) and site-specific effects in the AH on modulating offensive aggression (53). 5-HT1a receptors were expressed on AVP-ir neurons (Figure 2A, C, & D) surrounded by a dense network of 5-HT-ir boutons (Figure 2B). Sexually naïve males received bilateral intra-AH injections of AVP (500ng/side) in CSF or CSF containing a low (0.5μg/side) or high (5μg/side) dose of a 5-HTr1a agonist (R(+)-8-OH-DPAT) followed by a 10-min RIT toward a novel female. Doses were determined based on previous studies (52). Intra-AH microinjections of the 5-HTr1a agonist attenuated AVP-induced aggression (Figure 2E) and enhanced social affiliation compared to CSF controls (Figure 2F). Treatment of R (+)-8-OH-DPAT did not affect other behaviors (Table S6). These data indicate that 5-HTr1a activation by R (+)-8-OH-DPAT abolishes selective aggression that was pharmacologically induced by intra-AH AVP administration. We then tested whether manipulation of 5-HTr1a in the AH influenced aggression naturally induced by pair bonding (16–18). Pair-bonded males received intra-AH injections of CSF (control) or CSF containing a 5-HTr1a agonist (R(+)-8-OH-DPAT; 5μg/side) or antagonist (p-MPPI; 5μg/side) followed by a 10-min RIT toward a novel female. Compared to CSF controls, intra-AH infusions of the 5-HTr1a agonist abolished aggression (Figure 2G) and facilitated affiliation (Figure 2H). Conversely, blocking 5-HTr1a in the AH enhanced aggression above the CSF. Other behaviors were not affected (Table S7).

Figure 2. AH-5-HT-1a-type Receptor Activation Attenuates AVP- and Pair-Bonding-Induced Aggression.

(A) AVP, (B) 5-HT, (C) 5-HTr1a, and (D) AVP/5-HT/5-HTr1a labeling in the anterior hypothalamic nucleus. AVP/5-HTr1a double-labeled neurons indicated by white arrowheads. For the behavioral experiments (E–H), sexually naïve male prairie voles received intra-AH injections of AVP (500 ng in 200 nl CSF)/side) or AVP with a low (0.5 μg/side) or high (5 μg/side) dose of a 5-HTr1a agonist (R(+)-8-OH-DPAT) followed by a 10-min resident intruder test (RIT) toward a novel female. Both doses of the 5-HTr1a agonist blocked AH-AVP-induced aggression (F_(2,18) = 9.23, p < 0.01) (E) and enhanced affiliation (F_(2,18) = 4.20, p < 0.05) (F) compared to CSF-injected controls. Male prairie voles that were pair-bonded with a female for two weeks were divided into three groups that received intra-AH infusions of CSF (200 nl/side) or CSF containing a 5-HTr1a agonist (5 μg/side) or antagonist (p-MPPI; 5 μg/side) followed by a 10-min RIT. (G & H) Males treated with the 5-HTr1a agonist displayed a significant decrease in aggression (F _(2,20) = 40.83, p < 0.001) and an increase in affiliation (F_(2,20) = 8.01, p < 0.01) compared to CSF-injected controls. (G) In contrast, males treated with the 5-HTr1a antagonist displayed enhanced aggression toward a novel female than did CSF-injected controls (F_(2,20) = 40.83, p < 0.001). Lastly, we tested the behavioral consequences of AVP administration in the AH of pair-bonded males treated with a combination of 5-HTr1a agonist/antagonist infusions. Overall, males displayed significantly higher levels of offensive aggression and low affiliation toward a sexually-receptive female (I & J). Males treated with the 5-HTr1a 5μg antagonist exhibited significantly higher levels of offensive aggression (F _(2,20) = 5.39, p < 0.05; I). The effect of enhancing aggressive responding by blocking AH-5-HTr1a, in intra-AH-AVP infused sexually naïve males (E–H), is abolished in pair-boned - intra-AH-AVP treated males - while 5-HTr1a antagonist treatment in these males enhances offensive aggression (I). Thus, the effect of 5-HT1aR activation on aggression can be blocked by AVP, whereas the effect of AVP cannot be blocked by 5-HT antagonists. Bars indicate means ± standard error of the mean. Bars with different alphabetical letters differ significantly from each other. Scale bar = 10μm.

Lastly, we tested the behavioral consequences of AVP administration in the AH of males treated with a combination of intra-AH 5-HTr1a agonist/antagonist infusions. Pair-bonded males were divided into one of four groups that received intra-AH infusions of AVP (500 ng in 200 nl. CSF/side) in CSF (control, n=8) or CSF containing a 5-HTr1a agonist (R (+)-8-OH-DPAT; 5μg/side, n=9), antagonist (p-MPPI; 5μg/side, n=7), or both (R (+)-8-OH-DPAT + p-MPPI; 5μg/side, n=8) followed by a 10-min RIT toward a novel female. Overall, males displayed significantly higher levels of offensive aggression and low levels of affiliation toward a novel female (Figure 2I & J). Males treated with the 5-HTr1a 5μg antagonist exhibited significantly higher levels of offensive aggression (Figure 2I). No group differences were found in affiliation (Figure 2J) or other behaviors measure (Table S8), extending previous findings (37–39).

Cytochemical Profiling of AH Neurons

As AVP, 5-HT, and CRH coordinate patterns of affiliation and aggression, we histochemically profiled the AH for these neurochemical markers and their receptors (Table 2). Twenty percent of AVP-ir neurons in the AH co-expressed 5-HTr1a-ir (Figure 3A–C, Table 2), which confirm our triple-labeling experiments (Figure 2A–D). Sixteen percent of CRH-ir neurons in the AH co-expressed 5-HTr1a-ir (Figure 3D–F, Table 2) and forty-five percent of AVP-ir neurons co-expressed CRH-type-2 receptors (CRHR2; Figure 3G–I, Table 2). Finally, fourteen percent of CRH-ir neurons co-expressed V1aR-ir (Figure 3J, K, & M, Table 2). In addition, forty-five percent of neurons double-labeled for CRH-ir/V1aR-ir co-expressed FG-ir (Figure 3J–M) in males that received FG injections into the LS.

Table 2.

AHN double-label immunofluorescence neurochemical profiling (# neurons/mm² brain region volume)

# Animals	# Sections	Neurochemical Markers					Neurochem ical Double Labeling
		AVP	CRH	5- HTr1a	CRHR2	V1aR
8	16	179.8±19 .1* (19.76% )**		104.5±13.8 (33.99 %)			35.5±9.4 (AVP/5- HTr1a)
6	12		439.3±35.2 (16.55 %)	122.9±15.5 (59.13 %)			72.7±13.7 (CRH/5HTr 1a)
7	14	142.6±16 .0 (44.99% )			158.3±29.6 (40.53 %)		64.2±8.4 (AVP/CRH R2)
8	16		411.1±28.9 (14.49 %)			348.6±37.8 (25.10 %)	87.4±14.3 (CRH/V1aR )

^*Data are presented as mean ± standard error.

^**Percent maker co-expressing the additional label on the same row.

Figure 3. 5-HT-, CRH-, and AVP-Expressing Neurons/Receptors Co-localize in the AH.

Photomicrographs displaying cytochemical marker fluorescence histochemistry in the anterior hypothalamic nucleus. (A) serotonin (5-HT) 1a-type receptors (5-HTr1a), (B) vasopressin (AVP), and (C) 5-HTr1a/AVP double-labeled neurons indicated by white arrowheads. (D) 5-HTr1a receptors, (E) corticotrophin-releasing hormone (CRH), and (F) 5-HTr1a/CRH double-labeled neurons indicated by white arrowheads. (G) CRH-type-2 receptors (CRHR2), (H) AVP, and (I) CRHR2/AVP double-labeled neurons indicated by white arrowheads. (J) AVP-V1a-type receptors (V1aR), (K) CRH, (L) Fluorogold (FG - injected into the lateral septum), (M) V1aR/CRH/FG triple-labeled neurons indicated by white arrows. Scale bar = 10μm.

AH 5-HT Mediates Neuropeptide Release to Modulate Behavioral Switches

Pair-bonded males were implanted with a microdialysis probe aimed at the AH. After one week recovery, subjects were randomly divided into four pharmacological treatment groups and received reverse microdialysis infusion of CSF, or CSF containing a receptor agonist or antagonist for V1aR, CRHR2, or 5-HTr1a, while their behavior toward a familiar partner or a novel female was examined using RIT (Figure 4A). Microdialysis samples were collected and subsequently processed for AVP, CRH, and 5-HT contents. In control males (CSF), aggression levels were low when they were reunited with their female partner but high when they were exposed to a novel female (Figure 4B), and a reverse pattern was found in affiliative behavior (Figure 4C). Pharmacological inactivation of V1aR or CRHR2 as well as activation of 5-HTr1a in the AH diminished aggression and facilitated affiliation toward novel females. Conversely, activation of V1aR or CRHR2 induced aggression and decreased affiliation toward their female partner (Figure 4B & C). Furthermore, blockade of 5-HTr1a in the AH impaired affiliation toward female partners (Figure 4C). None of the drug compounds affected other behaviors measured (Tables S9–11).

Figure 4. Behaviorally- and Pharmacologically-Evoked Neurotransmitter Release in the AH Reveals Dynamic Regulation of Behavioral Switches.

(A) Real-time in-vivo brain microdialysis paradigm. Pair-bonded males were stereotaxically implanted with a microdialysis probe aimed at the anterior hypothalamic nucleus (AH) and divided into four pharmacological treatment groups for vehicle (CSF) infusions with manipulations of vasopressin (AVP), corticotrophin-releasing hormone (CRH), and serotonin (5-HT) systems, respectively. Microdialysate samples were collected every 30-minutes over a 5-hour period in which males were reunited with their female partner, introduced to a novel female, and then re-exposed to their female partner again. Reverse dialysis of pharmacological compounds were infused during exposure to novel females and re-exposure to their female partner. (B & C) Blockade of AVP-V1a-type receptors (V1aRs) or CRH type-2 receptors (CRHR2s) or activation of 5-HT-1a-type receptors (5-HTr1as) in the AH abolished aggression (F_(3,28) = 56.25, p < 0.001; B) and facilitated affiliation (F_(3,28) = 41.61, p < 0.001; C) toward novel females. Activation of either V1aR or CRHR2 induced aggression (F_(3,28) = 18.26, p < 0.01; B) and reduced affiliation (F_(3,28) = 15.92, p < 0.01; C) toward their female partner, while blockade of 5-HTr1a decreased affiliation (F_(3,28) = 63.74, p < 0.001; C) but did not induce aggression toward their female partner (F_(3,28) = 2.91, ns; B). (D) In CSF control males, 5-HT release increased while AVP/CRH release decreased when males were either reunited or re-exposed (F_(3,28) = 14.52, p < 0.01) to their female partner. A reverse pattern of neurotransmitter release was found when males were fighting novel females (F_(3,28) = 17.49, p < 0.01). (E) Blockade of V1aR in the AH attenuated AVP/CRH release associated with exposure to novel females (F_(3,28) = 13.46, p < 0.01) while activation of V1aR facilitated AVP/CRH release and decreased 5-HT release in the AH during partner re-exposure (F_(3,28) = 15.82, p < 0.01). (F) Blockade of CRHR2 also diminished AVP/CRH release associated with exposure to novel females (F_(3,28) = 14.88, p < 0.01) while activation of CRHR2 enhanced AVP/CRH release and decreased 5-HT release during partner re-exposure (F_(3,28) = 13.37, p < 0.01). (G) Activation of 5-HTr1a attenuated AVP/CRH release associated with exposure to novel females (F_(3,28) = 18.59, p < 0.01) while blockade of 5-HTr1a diminished increased 5-HT release and decreased AVP/CRH release during partner re-exposure. Bars indicate means ± standard error of the mean. Bars with different alphabetical letters differ significantly from each other at p < 0.01. Line time points indicate percent change from baseline ± standard error of the mean. *: p < 0.01. ANT: antagonist, AGO: agonist, B: baseline (with CSF infusions). See also Figure S1D.

Changes in behavioral responses toward a partner or novel female were associated with dynamic neurochemical release patterns in the AH. Enhanced 5-HT release coupled with decreased AVP and CRH release was associated with low aggression and high affiliation (Figure 4D–G). Conversely, a reverse neurotransmitter release pattern was associated with aggression toward a novel female (Figure 4D). These release findings were also observed by activating V1aR or CRHR2 in the AH (Figure 4E & F), which induced aggression and impaired affiliation (Figure 4B & C). Diminished aggression and enhanced affiliation (Figure 4B & C) toward novel females, by V1aR or CRHR2 blockade or by 5-HTr1a activation, were associated with an inhibition in AVP and CRH release (Figure 4E–G). Finally, impaired partner affiliation, by 5-HTr1a antagonism in the AH, was associated with the disappearance of high 5-HT and low AVP/CRH release (Figure 4G).

DISCUSSION

Healthy social relationships are necessary for maintaining human mental health, yet we know little regarding interconnections of brain regions and neurochemical interactions underlying the formation and maintenance of sociality. Using the socially monogamous prairie vole, we provide data, for the first time, illustrating a novel limbic network wherein several neurochemical systems converge to regulate ethologically important behaviors critical for male-female pair-bonding.

Our data indicate that the display of aggression by male voles was associated with activation of a sub-population of neurons in the MeAPD projecting to the AH and ones in the AH projecting to the LS, and those projection neurons in the MeAPD-AH-LS circuit expressed CRH. Males are physiologically stressed when separated from their partner (54, 55) and then presented with a novel conspecific, this extra-hypothalamic CRH stress circuit is engaged to facilitate aggression (16–18). Released CRH in the AH acts on CRHR2 expressed on AVP neurons. CRHR2s are coupled to a stimulatory G-protein signaling cascade (56), activating adenylate cyclase (AC) and increasing cyclic adenosine monophosphate (cAMP) and intracellular Ca²⁺ (57, 58). CRHR2 activation may lead to membrane depolarization, facilitating AVP and CRH release in the AH to enhance aggression. On the other hand, released AVP can bind to V1aRs expressed on CRH neurons. V1aRs are also coupled to stimulatory G-proteins (59, 60) and their activation enhances AC activity and increases cAMP and intracellular ²⁺ Ca (61, 62) within CRH neurons projecting to the LS where CRH is released and acts on CRHR2s (63, 64) to escalate aggression (65, 66). Thus, AH-AVP microinfusion can increase CRH levels and may reduce 5-HT inputs to the AH by stimulating GABA-ergic projections synapsing onto 5-HT neurons in the DR-AH pathway. Further, released AVP in the AH can act directly on local V1aR-expressing neurons to regulate selective aggression (17).

Conversely, when males reunite with their partner, a subset of neurons in the DR that expresses 5-HT and projects to the AH was activated. Released 5-HT in the AH acts on 5-HT1a-type receptors expressed on both AVP and CRH neurons. 5-HT1a-type receptors are coupled to an inhibitory G-protein signaling cascade (67) and their activation results in decreases in AC activity, intracellular Ca²⁺, and cellular depolarization (68, 69), leading to a decrease in AVP/CRH release in the AH. Involvement of a 5-HT-ergic microcircuit in pair bonding behavior is further supported by our data showing that acute treatment of fluoxetine suppressed AVP-induced aggression, reduced AH 5-HT turnover, and enhanced social affiliation in sexually naïve males. Indeed, fluoxetine has been found to block agonistic behavior in several species including humans (42, 70–72) and to abolish AVP-induced aggression in rodents (44, 47) by decreasing levels of AVP in the AH (71, 73).

5-HT has long been considered as an important neurotransmitter in the regulation of impulsive aggressive behavior. Patients with a history of physical violence exhibit low CSF levels of 5-HIAA (74, 75). Low 5-HIAA typically indicates decreased 5-HT release and correlates with aggressive behavior (76) and alcohol-related forcefulness (77) in adults as well as impulsive violent behavior in children (78). In free-ranging rhesus monkeys, low levels of CSF 5-HIAA correlate with increased aggression and risk taking (79, 80). In talapoin monkeys with an established social hierarchy, high levels of 5-HIAA in subordinates are related to their low social status characterized by high levels of withdrawal and diminished aggression (81). Human imaging work has shown that a DR 5-HT brainstem microcircuit is activated when subjects are presented with images of their long-term partner (82).

The notion that 5-HT may inhibit AVP release in the human brain is supported by clinical findings that patients presenting with a personality-disorder and exhibiting a lifetime history of fighting and assault showed a positive correlation between CSF levels of AVP and aggression (83) with a hyporeactive 5-HT system as assessed by fenfluramine challenge (84). Fenfluramine is a 5-HT-releasing drug that normally stimulates prolactin release as a neuroendocrine measure of central 5-HT activity. Patients with a lifetime history of escalated aggression and violence show blunted prolactin release in response to fenfluramine (85, 86). This is also true in macaque monkeys that showed increased aggressive responding that negatively correlated with diminished prolactin release in response to fenfluramine challenge (87). Interactions between AVP and 5-HT have also been implicated in pathological aggression in patients with borderline personality disorders (83) who exhibit impairment in bonding with mates (88, 89) because of partner-directed violence (90). Interactions between CRH and 5-HT also underlie aggression toward offspring in abusive rhesus macaque mothers (91). Because most neuromodulators released by amine- and peptide-containing neurons show remarkable preservation of structure and function within the animal kingdom (92), this evidence in humans and non-human primates, coupled with our results in prairie voles, suggests that prosocial behavior associated with pair-bonds may be sub-served by evolutionarily conserved neurochemical circuitry facilitating affiliation.

In summary, our data establish the medial extended amygdala as a critical neural node in which neurochemicals interact in programming naturally-occurring social behaviors associated with pair bonding (Figure 5). These data illustrate the spatio-temporal precision of neurotransmitter release required for the expression of partner affiliation, the reversible capacity of these neuromodulatory circuits in response to changes in social stimuli, and the great utility of the prairie vole model to study neurotransmitter micro-circuits in mediating and optimizing decision-making and behavioral switch under specific environmental contexts.

Figure 5. Microcircuit “Switch” Mechanism Programming Behavioral State.

(A) Schematic illustrates the neurocircuitry and neurotransmitter circuit phenotypes summarized from monosynaptic neuronal tract-tracing and histochemical experiments. Black arrows indicate anatomical connections and neurochemical projections are color-coded. The anterior hypothalamic nucleus (AH) projects to forebrain areas: ventral pallidum (VP) and bed nucleus of the stria terminalis (BNST) that are involved in pair-bonding behavior; intersects several dopaminergic regions including the ventral tegmental area (VTA), nucleus accumbens (NAcc), caudate putamen (CP), and prefrontal cortex (PFC); integrates olfactory and pheromonal information from the vomeronasal organ through the accessory olfactory bulb (AOB) via the posterior dorsal medial amygdala (MeAPD); receives corticotrophin-releasing hormone (CRH) projections from the MeAPD and serotonergic (5-HT-ergic) input from the dorsal raphe nucleus (DR); and sends CRH projections to the lateral septum (LS). (B) During affiliation, a subset of 5-HT neurons in the DR project axonal collaterals and release 5-HT in the AH. Released 5-HT acts on post-synaptic 5-HT1a-type receptors (5-HTr1a) co-expressed on AVP- and CRH-containing interneurons, in the medial supra-optic nucleus (mSON) and nucleus circularis (NC), to suppress local AVP/CRH release, enhance affiliation, and inhibit aggression. (C) During aggression, a subpopulation of CRH neurons in the MeAPD project dendritic arbors to and release CRH in the AH. Released CRH binds to post-synaptic CRH-type-2 receptors (CRHR2s) co-expressed on the membrane surface of AVP interneurons, in the AH, to facilitate local AVP release. Released AVP then acts on post-synaptic V1a-type receptors (V1aR) co-expressed on a subset of CRH neurons projecting to the LS, where CRH is released and activates CRHR2-expressing neurons to escalate aggression.