Coordination of social behaviors by the bed nucleus of the stria terminalis

Abstract

The bed nucleus of the stria terminalis (BNST) is a sexually dimorphic, neuropeptide-rich node of the extended amygdala that has been implicated in responses to stress, drugs of abuse, and natural rewards. Its function is dysregulated in neuropsychiatric disorders that are characterized by stress- or drug-induced alterations in mood, arousal, motivation, and social behavior. However, compared to the BNST’s role in mood, arousal, and motivation, its role in social behavior has remained relatively understudied. Moreover, the precise cell types and circuits underlying the BNST’s role in social behavior have only recently begun to be explored using modern neuroscience techniques. Here, we systematically review the existing literature investigating the neurobiological substrates within the BNST that contribute to the coordination of various sex-dependent and sex-independent social behavioral repertoires, focusing largely on pharmacological and circuit-based behavioral studies in rodents. We suggest that the BNST coordinates social behavior by promoting appropriate assessment of social contexts to select relevant behavioral outputs and that disruption of socially-relevant BNST systems by stress and drugs of abuse may be an important factor in the development of social dysfunction in neuropsychiatric disorders.

Introduction

The bed nucleus of the stria terminalis (BNST) is a neurochemically diverse hub of the extended amygdala that integrates sensory, stress, and reward-related information to mediate responses to threats, natural reinforcers, and drugs of abuse. It has been strongly implicated in human psychopathologies such as generalized anxiety disorder (GAD), post-traumatic stress disorder (PTSD), and substance use disorder (SUD), all of which can be characterized by alterations in mood, arousal, sleep, appetite, and social behavior (Avery et al., 2016). Animal studies modeling these disorders have subsequently identified important roles for specific BNST cell types and circuits in mediating a number of the associated symptoms, particularly disordered arousal and mood. Though the BNST largely promotes negative affect and aversion-related behaviors, distinct BNST cell types and circuits can in fact drive reward-related behaviors (Jennings et al., 2013). In an attempt to unify these seemingly dichotomous functions, it has been suggested that the BNST performs what has been termed “valence surveillance” (Lebow & Chen, 2016). This proposes that the BNST’s role in behavior is to assess the positive or negative valence of a given context or stimulus and act appropriately by integrating incoming information about mood, arousal, relevant sensory stimuli, and motivational state.

Anatomically, the BNST acts as a relay station between cortical limbic regions, arousal centers, and reward centers, and displays dense reciprocal connectivity with the amygdala, extended amygdala, and hypothalamus. Though often overlooked as such, the BNST is a critical node in the social behavioral network and interfaces with regions that are integral for social decision making and the processing of multimodal social stimuli (O’Connell & Hofmann, 2011; Lebow & Chen, 2016). However, despite this pattern of anatomical connectivity, the role of the BNST in many social behaviors remains relatively understudied and thus poorly understood at the circuit level. Following a brief overview of BNST anatomy and neurochemical architecture, this review will synthesize what is known about the specific BNST cells and circuits implicated in social behavior in healthy states, stressed states, and states of drug withdrawal and intoxication. In addition, it will cover emerging data demonstrating BNST dysfunction in humans with social anxiety. We note that unless otherwise stated, all rodent studies discussed in this review were performed in commonly used inbred laboratory mouse strains, primarily the C57BL6/J strain. While decades of research suggest that the BNST plays an important role in in a multitude of different social behaviors, including but not limited to maternal care and aggression, intermale aggression, prosocial behaviors, sexual behaviors, and social dysfunction associated with stress and drug exposure, few articles have systematically reviewed these studies. With the idea of “valence surveillance” in mind, we hypothesize that a central function of the BNST in social behavior is to integrate both internal and external evaluations of the conditions of a given social interaction to inform behavior. More specifically, we suggest that the BNST assesses whether a given social target should be ignored, avoided, or engaged with by assessing the identity of the social target and the context of the interaction, the internal motivational state and stress history of the animals, the physical or emotional threat posed by the social interaction, and the reward value of the social interaction.

Anatomy of the BNST

Connectivity

The BNST is a sexually dimorphic nucleus comprised of up to 18 different sub-nuclei, each expressing a distinct combination of receptors, neurotransmitters, and transcription factors as well as sending and receiving projections to and from distinct regions. While extensive efforts have been made to characterize the makeup, connectivity, and function of these sub-regions (reviewed thoroughly in Lebow and Chen 2016), it is not known how they individually integrate information from multiple intra- and extra-BNST sources to drive behavior. This may be due, at least in part, to the technical difficulties associated with targeting and manipulating anatomically small sites with available stereotaxic, viral, and pharmacological methods. Due to these limitations, studies investigating the functional role of BNST sub-regions in behavior or sub-region-specific connectivity patterns often distinguish only between dorsal and ventral or anterior and posterior BNST, although the oval nucleus may also be targeted. While all of these BNST sub-regions interface with regions that are important for social behavior, they do so in distinct ways. The ventral portion of the BNST, for example, displays extensive connectivity with the medial amygdala (MeA) and ventromedial hypothalamus (VMH) and expresses a high density of steroid hormone receptors (Laflamme et al., 1998; Dong & Swanson, 2004a; 2006a; Ni et al., 2016). It also communicates heavily with regions of the mesocorticolimbic reward pathway like the ventral tegmental area (VTA), nucleus accumbens (NAc), and medial prefrontal cortex (mPFC), as well as the dorsal raphe nucleus (DRN) (Dong & Swanson, 2004a; 2006a; Ni et al., 2016). In addition, the ventral BNST receives the densest brainstem-derived noradrenergic inputs of any site in the brain (Forray & Gysling, 2004). While the dorsal BNST also expresses steroid hormone receptors and communicates with the DRN, VTA, NAc, and mPFC, it is also strongly connected with neuroendocrine centers like the paraventricular hypothalamus (PVN) and nodes of the extended amygdala like the central amygdala (CeA) (Dong & Swanson, 2006b; Ni et al., 2016). It is indeed possible that the ventral and dorsal BNST modulate unique aspects of social behavior based on their relatively distinct patterns of connectivity. For example, communication between the ventral BNST and the VMH may specifically modulate aggressive and sexual behavior, while communication between the dorsal BNST and oxytocin (OXT) neurons in the PVN may specifically modulate prosocial approach and avoidance. The anterior BNST is reciprocally connected with the amygdala and hypothalamus and receives notable inputs from the periaqueductal gray (PAG), DRN, VTA, and mPFC (Peyron et al., 1998; Forray & Gysling, 2004; Vertes, 2004; Shin et al., 2008; Herr et al., 2012; Ni et al., 2016). The posterior BNST is the most sexually dimorphic sub-region (Laflamme et al., 1998; Campi et al., 2013) and receives dense inputs from the olfactory bulb (OB), accessory olfactory bulb (AOB), and MeA while sending outputs to the VTA, hypothalamus, and lateral septum (LS) (Dong & Swanson, 2004b; Shin et al., 2008; Ni et al., 2016). Based on these patterns of connectivity, it is likely that the posterior BNST is particularly important for modulating innate sex-specific social behaviors like aggression, mating, and parental care, whereas the anterior BNST plays a role in mediating responses to social threats and stress-induced impairments in social behavior. While this idea has been investigated to some extent, future research is required to elucidate the precise inputs, outputs, sub-regions, and cell populations of the BNST that modulate particular aspects of social behavior.

Neurochemical architecture

The vast majority of BNST neurons release the inhibitory neurotransmitter GABA, form symmetric synapses, and express the vesicular GABA transporter (vGAT) (Kudo et al., 2012; Mazzone et al., 2018; Moffitt et al., 2018; Welch et al., 2019; Rodriguez-Romaguera et al., 2020). Thus, BNST outputs generally inhibit their target cells. However, roughly one-tenth of BNST neurons release glutamate, form asymmetric synapses, and express the vesicular glutamate transporter 2 (vGlut2) (Kudo et al., 2012; Moffitt et al., 2018; Welch et al., 2019; Rodriguez-Romaguera et al., 2020). Although there is also a small population of vesicular glutamate transporter 3 (vGlut3)-expressing neurons in the BNST, these neurons form symmetric synapses, express enzymes necessary for GABA production, and release GABA (Kudo et al., 2012; Welch et al., 2019). BNST neurons are also known to express a large variety of neuropeptides, which include but are not limited to dynorphin (Dyn), enkephalin (Enk), OXT, arginine vasopressin (AVP), corticotropin-releasing factor (CRF), cholecystokinin (CCK), neuropeptide Y (NPY), somatostatin (SOM), pituitary adenylate cyclase-activating polypeptide (PACAP), neurotensin (NTS), and nociceptin (NC) (Kash et al., 2015). Single BNST neurons can express widely varying combinations of the aforementioned neurotransmitters, neuropeptides, and their receptors; thus the identification of transcriptionally-defined “cell types” in the BNST has proven challenging. For example, one recent study that performed single-cell transcriptional profiling of BNST neurons reported that they can be divided into as many as 41 different transcriptional clusters (Welch et al., 2019). Another study suggested that BNST neurons can be divided into 11 distinct transcriptional clusters, and that expression of pre-pro-nociceptin, the precursor for the NC neuropeptide, is distributed across 4 of these 11 clusters (Rodriguez-Romaguera et al., 2020). These two studies likely identified differing numbers of BNST cell types because of the clustering analyses used. Welch et al. employed integrative non-negative matrix factorization to identify both shared and dataset-specific transcripts across datasets. Rodriguez-Romaguera, on the other hand, used a more traditional principal component analysis to cluster cells and did not include transcripts if they were not present in all datasets. Regardless of how many transcriptionally-defined BNST “cell types” there may be, there does appear to be a spatial and functional relationship between transcriptionally similar neurons in the BNST (Moffitt et al., 2018). The BNST is also a major site of release for monoamines, catecholamines, hypothalamic-derived neuropeptides, and steroid hormones (Kash et al., 2015), all of which may dynamically modulate the output activity of BNST neurons in cell type and circuit-specific manners. Furthermore, male and female rodents display marked differences in BNST steroid hormone responsiveness and neuropeptide synthesis (Laflamme et al., 1998; Toufexis, 2007; Campi et al., 2013; Gungor & Paré, 2016). This complexity of sex-specific neurochemical signaling within the BNST combined with its intricate sub-region-specific connectivity may ultimately underlie its capacity to orchestrate a broad range of social behavioral repertoires.

The BNST and aggressive social behaviors

Aggression is a social behavior that evolved to promote the protection of self, resources, progeny, and territory. Though it may be adaptive or maladaptive in animals depending on its severity, aggression in humans is largely maladaptive and is a hallmark of neuropsychiatric disorders like SUD and PTSD. In rodents, aggression is strongly influenced by the sex, sexual history, and stress history of the experimental animal and the social target (Miczek et al., 2001; Miczek et al., 2013). As female rodents primarily engage in aggression during the postpartum period and direct attacks towards males, interfemale aggression has been historically understudied. However, recent work has found that nulliparous female mice co-housed with castrated males will reliably exhibit aggression towards female social targets, which opens up new avenues for studying interfemale aggression (Newman et al., 2019). On the other hand, nearly all male rodents are capable of exhibiting aggression towards other males, and even toward juveniles, throughout their adult lives and in a variety of social contexts (Miczek et al., 2001; Miczek et al., 2013). Prior winning experience increases subsequent intermale aggression, as does social isolation, copulation, and brief exposure to a physically inaccessible male conspecific prior to aggression testing (Valzelli, 1973; Flannelly et al., 1982; Potegal, 1992; Miczek et al., 2001; Hsu et al., 2006). Specific offensive behaviors exhibited during intermale aggression include biting, chasing, tail rattling, and making lateral threats (Blanchard et al., 1984; Miczek et al., 2001). Most aggressive contests also require both animals to engage in defensive behaviors such as crouching and darting (Blanchard et al., 1984). In addition, because aggression requires that animals properly identify male versus female, novel versus familiar, and adult versus juvenile social targets, pheromone signaling plays a large role. Recent studies also suggest that there is a strong motivational component to aggressive behavior, as aggressive male rodents will operantly self-administer for the opportunity of aggression as reinforcement—e.g. they will lever press to gain access to submissive social targets they can attack (Falkner et al., 2016; Golden et al., 2017; Golden et al., 2019). Similarly, aggressive mice form conditioned-place preferences for contexts associated with winning aggressive contests (Golden et al., 2016; Aleyasin et al., 2018; Flanigan et al., 2020). Furthermore, both stress and drugs of abuse modulate the expression of aggressive behavior. For example, social isolation stress as well as withdrawal from alcohol or morphine increases aggression in male rodents (Miczek et al., 2013; Miczek et al., 2015; Piccin & Contarino, 2020). Therefore, because the BNST is a hub between pheromone signaling centers like the AOB, reward centers like the VTA, and aggression initiation centers like the MeA, LS, and VMH, it is well positioned to influence multiple aspects of aggressive behavior.

BNST neurons are broadly activated by aggression-related stimuli as well as aggressive social interactions themselves. For example, male rats that are presented with both a female cage mate and an unfamiliar male while in the bore of an fMRI magnet, which would generally provoke aggressive behavior towards the unfamiliar male, display increased BNST activation (Ferris et al., 2008). This level of activation is higher than that induced by presentation of the female cage mate alone. Male mice also display increased BNST c-Fos expression (a marker of neural activation) following an aggressive encounter (Haller et al., 2006; Lin et al., 2011; Bayless et al., 2019). Interestingly, female California mice exhibiting aggression towards female intruders also display robust increases in BNST c-Fos following aggression, suggesting that the BNST’s role in female aggression may extend beyond that of the maternal context, depending on the mouse strain investigated (Davis & Marler, 2004). This is an important area of research to pursue, as human females engage in aggression outside of the postpartum period yet we know very little about the underlying mechanisms.

Manipulations of GABA, glutamate, CRF, and AVP signaling in the BNST all have effects on aggressive behavior in male rodents. For example, infusion of a metabotropic glutamate receptor 7 (mGluR7) antagonist into the BNST reduces aggressive behavior, and mGluR7 knockout mice display reductions in aggression and reduced BNST activation in response to sex-specific olfactory cues (Masugi-Tokita et al., 2016). Though a GABA-A receptor agonist infused into the BNST of male Syrian hamsters prior to an aggressive contest does not alter defensive responses, if this treatment is applied following a social defeat defensive behaviors are reduced in subsequent contests (Markham et al., 2009). This implies that the BNST regulates experience-dependent flexibility in the expression of aggressive behaviors. The BNST CRF system is likely involved in this process, as natural increases in submissive behaviors following social defeat in male Syrian hamsters are blocked by CRF receptor 1 (CRFR1) or CRF receptor 2 (CRFR2) antagonist infusion into the BNST (Jasnow et al., 2004; Cooper & Huhman, 2010). The role of BNST AVP signaling in aggressive behavior appears to depend on the species tested and the stress history of the experimental animals. For example, highly aggressive male rats display reduced AVP release during aggression, and AVP infusion into the BNST of these male rats functionally reduces aggression (Veenema et al., 2010). Interestingly, antagonism of AVP receptor 1a (AVPR1a) in the BNST in this study did not affect rat intermale aggression. The results of this study are somewhat surprising, as AVP release in other brain regions has been shown to promote aggression in male rats (Dumais & Veenema, 2016; Terranova et al., 2017), and aggressive behaviors in Syrian hamsters like flank marking are associated with increased activity in AVPR1a-containing regions of the BNST (Bamshad et al., 1996). Furthermore, aggression in unstressed, but not stressed, California mice is positively correlated with the number of BNST AVP cell bodies (Steinman et al., 2015). Thus, further research is required to untangle the precise conditions in which BNST AVP neurons drive versus dampen aggression in various rodent species.

OXT neurons in the PVN (which project to the BNST) can drive prosocial or aggressive social behaviors in a manner that is highly dependent on the context of the social interaction. For example, optogenetic stimulation of PVN OXT cell bodies in the resident intruder (RI) test, whereby an experimental mouse (“resident”) is exposed to an unfamiliar conspecific (“intruder”) in the experimental mouse’s home cage, increases the incidence of aggressive behaviors (Miczek et al., 2013; Anpilov et al., 2020). Interestingly, stimulation of these neurons in a group social setting with familiar cage mates increases the incidence of prosocial interactions on the first day of testing but increases aggressive chases in the following days of testing (Anpilov et al., 2020). While there is some evidence that OXT release in the BNST positively modulates intermale aggression, it is unclear which BNST OXT sources (PVN or BNST) and which targets of BNST OXT- and oxytocin receptor (OXTR)-containing neurons are involved. Male mice who have engaged in pup-directed aggression display increased c-Fos expression in BNST OXTR-expressing neurons (Moffitt et al., 2018), and OXTR binding is increased in the BNST of highly aggressive rats (Calcagnoli et al., 2014). PVN OXT mRNA is simultaneously reduced in these highly aggressive rats, suggesting that aggressive behavior is controlled by brain site-specific alterations in OXT signaling (Calcagnoli et al., 2014). It should be tested whether optogenetic stimulation of PVN OXT axons in the BNST modulates aggressive behavior in a similar manner to that of broad PVN OXT stimulation.

Steroid hormone signaling in the BNST has also been implicated in aggressive behavior. Androgen receptors (AR) are activated by circulating testosterone and are expressed in a sexually dimorphic manner in the posterior BNST (Wu et al., 2009). Mice who have won aggressive contests display increased AR mRNA and protein in the BNST (Fuxjager et al., 2010), and increased dominance behavior is associated with estrogen receptor alpha 1a (ESR1a) mRNA expression in the BNST (Greenberg et al., 2014b). Interestingly, BNST AR neurons are not activated during attack (as measured using the in-vivo calcium imaging technique fiber photometry (Gunaydin et al., 2014)), but their activity is increased upon presentation of an unfamiliar male social target (Bayless et al., 2019). While functional manipulation of BNST AR-expressing neurons does not alter maternal aggression in female mice, chemogenetic inhibition or caspase-mediated ablation of BNST AR neurons reduces attacks between males (Bayless et al., 2019). Remarkably, if these neurons are optogenetically stimulated either persistently during a male-male interaction or only upon the start of the interaction, aggression is reduced and mating behavior is initiated towards the male social targets (Bayless et al., 2019). This indicates that BNST AR neurons play a role in aggressive behavior largely via their influence on processes of sex discrimination in a manner that is dependent on the time course of BNST AR neuron activation.

Recent studies have begun to disentangle the roles of specific BNST inputs and outputs in aggressive behavior, particularly the MeA and the VMH. The VMH was originally discovered to be a crucial locus for the initiation of indiscriminate attack behavior in cats and has since been found to play a similar role in rodents (Hess, 1927; Lin et al., 2011). Its activity is tightly controlled by the LS, which suppresses aggression through GABAergic inhibition of the VMH (Brayley & Albert, 1977). In fact, lesions of the LS produce what is known as “septal rage” as a result of VMH disinhibition (Maeda, 1977). The BNST communicates bidirectionally with both the VMH and LS (Dong & Swanson, 2004b; a; 2006a; Shin et al., 2008). When the VMH is stimulated simultaneously with the BNST, there is an increase in active defense behaviors compared to VMH stimulation alone (Shaikh et al., 1986). Infusion of acetylcholine (ACh) into the VMH increases the stimulation threshold necessary for attack, but this does not occur if the BNST is lesioned prior to stimulation (Kono et al., 1986). This suggests that the BNST may play a role in balancing offensive with defensive aggressive behaviors through its connections with the VMH. The dynamic circuit interactions between the BNST and the VMH have recently been characterized and may partly explain the complexity of the BNST’s role in aggression. The output of VMH projection neurons in the VMH core are inhibited by the VMH shell, which receive inhibitory inputs from BNST neurons (Yamamoto et al., 2018). Though BNST neurons also project to the VMH core, BNST-evoked IPSCs are greater in VMH shell neurons. Depending on the firing rate of shell neurons, inputs from the BNST to the VMH can either result in net inhibition or net excitation of the VMH output neurons in the core. Thus, the BNST may modulate aggression via context-dependent interactions with various VMH neuronal populations. The MeA projects to both the VMH and the BNST, and optogenetic stimulation of multiple genetically-defined MeA cell types can influence aggression in opposing ways (Canteras et al., 1995; Pardo-Bellver et al., 2012; Hong et al., 2014; Unger et al., 2015; Padilla et al., 2016). A very recent study found that both aggressive experience and traumatic stress, which produce experience-dependent increases in aggressive behavior, potentiate connections between the MeA and the BNST in an N-methyl-D-aspartate receptor (NMDAR)-dependent manner to promote the initiation, but not the maintenance, of intermale aggressive behaviors (Nordman et al., 2020). The MeA-VMH projection, on the other hand, undergoes these same experience-dependent adaptations to promote the maintenance, but not initiation, of aggression. Remarkably, functional manipulation of either of these circuits does not alter non-aggressive social behaviors. Thus, the MeA-BNST and MeA-VMH circuits contribute to unique aspects of aggressive social interactions. Future studies should investigate whether these BNST circuits are dysregulated in response to stress or drug exposure to drive changes in aggressive behavior.

The BNST and maternal social behaviors

Maternal care and aggression represent two female-specific social behaviors that are integral for the survival and fitness of offspring. Rodent maternal care behaviors include pup licking and grooming, nest building, nursing, and retrieval of scattered pups (also referred to maternal motivation), which support healthy physical and behavioral development of the offspring (Champagne et al., 2001). Maternal aggression occurs in a short (7-10d) window in the early postpartum period and is necessary for pup defense against hostile adult males (Lonstein & Gammie, 2002; Bosch, 2011; 2013). Exposure to stress or drugs of abuse profoundly disrupts maternal behaviors in both humans and rodents. For example, human mothers using cocaine during pregnancy display reduced sensitivity and responsivity to infant-related cues than drug-free mothers (Gottwald & Thurman, 1994). In rodents, chronic cocaine treatment during the postpartum period reduces nursing behaviors, pup licking, and nest building but increases maternal aggression (Johns et al., 1997; Nelson et al., 1998). Current research suggests that the appropriate expression of maternal behaviors requires reductions in innate anxiety in the lactating mother that are driven by adaptations in BNST neuropeptide systems (Lonstein & Gammie, 2002; Bosch, 2011; 2013). While numerous studies indicate that the dysregulation of these BNST neuropeptide systems by stress has detrimental effects on maternal care, maternal motivation, and maternal aggression, few studies have investigated whether dysregulation of these maternal behaviors by drugs of abuse also involves neural adaptations in the BNST.

The expression of several neuropeptide receptors in the BNST varies throughout pregnancy, parturition, and lactation to regulate BNST activity and the expression of subsequent maternal behaviors. These primarily include receptors for the neuropeptides AVP and OXT, which have been widely implicated in maternal behavior in other brain regions like the mPOA (Bosch, 2013). Sources of BNST AVP and OXT include both the PVN and the BNST itself (Steinman et al., 2016; Janeček & Dabrowska, 2019; Kompier et al., 2019). Dams display increased expression of AVP receptor 1a (AVPR1a) and OXT receptor (OXTR) mRNA, as well as increased AVPR1a and OXT receptor binding, during parturition and lactation (Meddle et al., 2007; Naik & de Jong, 2017). During these periods, AVP peptide production in the BNST is stimulated by estrogen on a massive scale, with target regions like the LS and the lateral habenula (LHb) seeing particularly large increases in BNST-derived AVP (Fink et al., 1996). Parturition and lactation also increase c-Fos in the dorsal BNST (Hasen & Gammie, 2005) as well as in BNST neurons expressing OXTR in the posterior BNST (Meddle et al., 2007). Gestational cocaine treatment, which enhances maternal aggression during lactation, increases OXTR binding in the BNST (Jarrett et al., 2006). While antagonism of AVPR1a in the BNST does not alter measures of maternal care like pup retrieval or nursing, it reduces maternal aggression as well as anxiety-like behavior in lactating females, suggesting that BNST AVPR1a signaling is important for promoting maternal aggression but not maternal motivation (Bosch et al., 2010). However, these results are counter to the idea that maternal aggression requires reductions in anxiety, as anxiety and aggression were reduced by the same mechanism. Consistent with the fact that BNST OXTR expression, binding, and neural activation increase during the postpartum period, antagonism of OXTRs in the posterior BNST reduces pup retrieval behavior, suggesting that OXT in the BNST functionally drives maternal care, particularly maternal motivation (Klampfl & Bosch, 2019). On the other hand, administration of OXT directly into the BNST dose-dependently reduces maternal aggression towards unfamiliar males (Consiglio et al., 2005). This highlights a potentially divergent role for BNST OXT signaling in maternal care versus maternal aggression. Alternatively, the different effects of BNST OXT signaling on maternal care versus maternal aggression may be related to the contexts in which these experiments were performed—while pup retrieval experiments were conducted in a novel cage, maternal aggression experiments were conducted in the home cage. Future studies should investigate the anatomical targets of OXT, OXTR, AVP, and AVP1aR-containing neurons in the BNST and determine how they each contribute to various maternal behaviors in projection-specific manners. Moreover, it will be important to determine how extra-BNST sources of OXT and AVP modulate maternal care and aggression in various contexts.

As mentioned above, exposure to stressors or drugs of abuse reduces maternal care and increases maternal aggression. In turn, neglected offspring display increased mortality rates and greater incidence of maladaptive behaviors associated with human neuropsychiatric diseases like GAD, SUD, and depression (Orso et al., 2019). Existing research suggests that the BNST CRF system plays a central role in mediating the effects of stress on maternal behavior. Like AVP and OXT, sources of BNST CRF include both the PVN and the BNST itself (Klampfl & Bosch, 2019). BNST CRF mRNA is decreased during lactation, though the expression of CRFR1 and CRFR2 does not differ between lactating and virgin females (Klampfl et al., 2014; 2016a; Klampfl et al., 2016b). However, signaling via CRFR1 and CRFR2 in the BNST during the postpartum period affects both maternal care and aggression in a sub-nuclei and receptor-specific manner (Klampfl et al., 2014; 2016a; Klampfl et al., 2016b). Direct administration of CRFR1 or CRFR2 agonists into the posterior BNST reduces nursing behavior, while antagonism of these receptors prevents stress-induced impairment of nursing behavior and reduces anxiety-like behavior (Klampfl et al., 2014). Interestingly, maternal aggression is increased by CRFR2, but not CRFR1, agonist infusion into the posterior BNST (Klampfl et al., 2014). While administration of a CRFR1 agonist into the anterior BNST reduces nursing behavior, CRFR2 agonists cause a brief decrease in nursing followed by a long-term increase in nursing; neither manipulation affects maternal motivation, maternal aggression, or anxiety-like behavior (Klampfl et al., 2016a). CRF-binding protein (CRF-BP) is a critical regulator of the CRF system that primarily sequesters the endogenous ligands of CRFR1, CRF and urocortin-1 (UCN1), to decrease CRFR1’s activation (Haass-Koffler 2018). Pharmacological inhibition of CRF-BP in the posterior BNST has no effect on maternal care under baseline conditions, but impairs the reinstatement of maternal care following stress exposure (Klampfl et al., 2016b). This manipulation also enhances maternal aggression under baseline conditions while concurrently increasing anxiety-like behavior. Interestingly, CRF-BP inhibition in the anterior BNST following stress has no effects on maternal care or aggression (Klampfl et al., 2016b). Though these studies suggest that downregulation of CRF signaling in the BNST, particularly the posterior BNST, is critical for the expression of appropriate maternal social behaviors, the mechanisms driving this downregulation remain unknown. Similarly, it is unknown how communication between BNST CRF neurons and their downstream targets is altered by stressors that impact maternal behaviors. There is evidence that the CRF and OXT systems interact during the postpartum period to influence maternal behaviors, but the mechanisms of these interactions have only begun to be untangled. In males, both central and intra-BNST inhibition of CRFR2 increases local OXT release, but in lactating females this effect is only observed with central inhibition of CRFR2 (Martinon & Dabrowska, 2018; Klampfl & Bosch, 2019). This suggests that increased activation of CRFR2 in lactating dams during stress may mediate impairments in maternal behavior via reduction of PVN-derived OXT release in the BNST. As OXT neurons express CRFR2 and CRF neurons express OXTR (Dabrowska et al., 2011), the interactions between these peptide systems within the BNST are likely complex and bidirectional. Future studies should investigate whether these CRF/OXT interactions occur in the context of other social behaviors such as sexual and prosocial behaviors.

The BNST and sexual social behaviors

Sexually reproducing animals display sex-specific mating behaviors that enhance their fitness and reproductive success. In order for these mating interactions to occur appropriately, animals must correctly differentiate between social targets of different sexes as well as display motivation to engage in the sexual encounter. In both humans and rodents, sexual behaviors are profoundly suppressed by both stress and pain (Monga et al., 1998; Hamilton & Meston, 2013; Manzano Nieves et al., 2019; Pitcher et al., 2019). In addition, sexual dysfunction and lack of sexual desire is prevalent among human patients with stress-related neuropsychiatric disorders like depression and GAD, and the degree of dysfunction is associated with the severity of the disorder (Williams & Reynolds, 2006). Interestingly, drugs of abuse may impact sexual behaviors in distinct ways depending on the subject’s sex, the type of drug, and whether the subject is acutely intoxicated or in withdrawal. For example, male mice acutely intoxicated with cocaine display increased sniffing of female social targets but reduced consummatory sexual behaviors (Kohtz et al., 2019b), whereas female mice acutely intoxicated with cocaine display increased avoidance of male sexual targets and reduced sexual receptivity (Guarraci & Clark, 2003; Kohtz et al., 2019a). Withdrawal from chronic cocaine exposure, on the other hand, vastly reduces the motivation to seek out sexual encounters in male and female mice (Guarraci & Bolton, 2014; Barnea-Ygael et al., 2016). While some studies suggest an important role for the BNST in the recognition of potential sexual partners and motivation for/initiation of consummatory sexual behaviors, whether the effects of stress and drugs of abuse on sexual behavior also involve the BNST remains to be tested.

In rodents, BNST neurons are innervated by AOB neurons, which provide contextual information about a given social interaction such as sex and sexual history via pheromone-related signals that arise in the vomeronasal organ (VNO), which is a direct input to the AOB (Davis et al., 1978; Ni et al., 2016). A short (90s) social interaction between an intact male mouse and a ovariectomized female mouse (in which copulatory behaviors do not occur) induces activation of posterior and ventral BNST as well as inputs to the BNST from the AOB in the male mouse (Kim et al., 2015). Moreover, activation of the posterior BNST in this context is positively correlated with the time the male mouse spent engaged in ano-genital sniffing and close following of the female mouse. The level of posterior BNST activation in males interacting with females was also greater than that of males interacting with males, female scents, and novel objects. Consistent with these findings, lesions of the posterior BNST markedly reduce male ano-genital sniffing of female social targets and induce deficits in sexual behavior (Powers et al., 1987; Claro et al., 1995; Liu et al., 1997). In females, posterior BNST lesions reduce urine marking for male odors but not for female odors and increase ultra-sonic vocalizations in the presence of males, but do not affect lordosis behavior (Kirn & Floody, 1985; Martinez & Petrulis, 2011). Though these studies suggest that the posterior BNST might be important for the discrimination of opposite-sex olfactory cues and engagement in sexual behaviors, they unfortunately do not provide information about the specific BNST cell types and circuits that are important for these processes.

Similar to aggressive behavior, AR-containing BNST neurons appear to exert their effects on sexual behavior largely via their involvement in sex discrimination. In sexually naïve males, BNST AR neurons are activated during multiple phases of a copulatory interaction with a female, as measured using the in-vivo cell-type specific bulk calcium imaging technique fiber photometry (Gunaydin et al., 2014; Bayless et al., 2019). Remarkably, these neurons actually appear to encode the sex of social targets, as they respond more strongly to female social targets than male social targets in a manner that is dependent on upstream pheromone signaling. Female BNST AR neuron activity, however, does not change during copulatory interactions with males. Selective ablation or chemogenetic inhibition of BNST AR neurons in males eliminates preferences for female over male urine and disrupts mating, but these manipulations have no effects in females. Together, these findings indicate that in male mice, AR-expressing BNST neurons are critical for both identifying the sex of social targets and engaging appropriately in sex-specific copulatory behaviors. Future studies should investigate whether BNST AR activation itself (as opposed to activation of AR-containing BNST neurons), is important for these behaviors, or whether it merely serves as a convenient transcriptional marker of the relevant neural population.

Recently, it was discovered that the juvenile mouse pheromone ESP22, which drives sexual suppression in adult female mice, drives robust activation of the entire BNST (Osakada et al., 2018). ESP22 is a protein secreted from lacrimal gland cells and is most highly expressed in male and female juvenile mice from birth until 4 weeks of age, after which ESP22 expression and secretion declines sharply (Ferrero et al. 2013). Moreover, activation of inhibitory BNST outputs to the ventrolateral subdivision of the VMH via ESP22 promotes sexual suppression in female mice. While ESP22 has also been shown to promote sexual suppression in male mice, whether this also requires activation of the BNST-VMH pathway has not been tested (Ferrero et al., 2013). Furthermore, it is unknown what neuropeptides are released from BNST-VMH neurons and whether they influence the expression of sexual rejection via downstream signaling in the VMH. BNST AVP neurons regulate interactions with opposite sex conspecifics in a sex-specific manner, but it is not clear if they project to the VMH. Ablation of BNST AVP neurons in males increases urine marking in the presence of a female, but does not alter ultra-sonic vocalizations, copulatory behavior, or investigation of the females or their odor (Rigney et al., 2019). However, in females, this manipulation reduces female receptivity to mounting but does not alter any other social behaviors. Thus, it is possible that AVP neurons in females represent a parallel population within the BNST separate from neurons activated by ESP22 that communicate with regions like the VMH to promote sexual receptivity rather than sexual rejection. To facilitate these investigations, the specific inputs and projection targets of BNST AVP neurons should be comprehensively mapped and compared between the sexes.

The BNST and prosocial behaviors

Same-sex prosocial interactions promote survival and fitness through the sharing of resources and are crucial for maintaining psychological wellbeing. Prosocial deprivation is a major stressor that induces a vast number of behavioral alterations in species from mice to humans, including anhedonia, social avoidance, aggression, and aberrant responses to non-social stressors (Matthews & Tye, 2019). Prosocial interactions are highly rewarding, as rodents will choose to self-administer access to social targets over drugs of abuse (Venniro et al., 2018; Venniro et al., 2020) and form conditioned place preferences for contexts associated with prosocial encounters (Dolen et al., 2013). Specific behaviors engaged in during prosocial interactions in same-sex individuals include sniffing, grooming, huddling, and approaching (Bartal et al., 2011). Like its role in opposite sex social interactions and same-sex aggressive social interactions, the BNST’s role in same-sex prosocial interactions appears to primarily involve its ability to mediate proper discrimination of individual social targets and engage with them appropriately. Furthermore, stressors and drugs of abuse promote adaptations in BNST circuitry that alter subsequent prosocial behaviors, including social vigilance, social recognition, and social preference.

To date, only a handful studies have investigated the role of the BNST in same-sex prosocial interactions between naïve (non-stressed) rodents. Despite this, there is evidence that a variety of BNST neuropeptide systems are involved in cell-type, circuit-, and sex-specific ways, including OXT, CRF, AVP, and orexin/hypocretin. Prosocial interactions with same-sex conspecifics increase both c-Fos and cocaine and amphetamine related transcript (CART) expression in the BNST (Perkins et al., 2017). Non-conditional chemogenetic activation of the BNST reduces same-sex social preference and social recognition in both males and females (Emmons et al., 2020), and pharmacological inhibition of BNST GABA synthesis reduces free prosocial interaction behaviors while increasing anxiety-like behavior (Sajdyk et al., 2008). Similarly, orexin-A infusion into the BNST reduces prosocial behaviors and increases anxiety-like behavior in male mice, an effect that is dependent on NMDAR-mediated depolarization of BNST target neurons (Lungwitz et al. 2010). Repeated sub-anxiogenic doses of UCN1, a ligand of CRFR1, in the BNST causes a long-term (4 week) suppression of prosocial behaviors, an effect that is blocked by pre-treatment with CRFR1 antagonists (Lee et al., 2008). Thus, even in unstressed animals, the BNST CRF system negatively modulates prosocial behavior. Antagonism of AVP1aR in the BNST reduces prosocial behavior in both male and female California mice but only increases anxiety in males (Duque-Wilckens et al., 2016). Ablation of BNST AVP neurons reduces social recognition of same-sex conspecifics in male C57BL6/J mice, but does not affect this behavior in females (Whylings et al., 2020). Hence, there may be a sex-specific and species-specific role for BNST AVP in recognizing same-sex conspecifics. However, it is possible that the lack of effect of BNST AVP neuron ablation on social recognition in females was related to the estrous cycle phase of the female animals at the time of testing, which was not monitored in the study. OXTR antagonist infusion into the posterior BNST reduces social recognition of same-sex conspecifics in both sexes, but OXT infusion only increases social recognition in males (Dumais et al., 2016). The reasons for this finding remain unclear, as there are no differences in social interaction-induced OT release between the sexes (Dumais et al., 2016). In contrast, a recent study found that both males and females display reductions in social approach behavior following infusion of OXT into the anterior BNST (Duque-Wilckens et al., 2020). The discrepancies between these studies may be due to the fact that experimenters did not specifically manipulate intra- versus extra-BNST OXT sources or that they targeted different sub-regions of the BNST (anterior versus posterior). As mentioned in the above sections on other types of social behavior, future studies investigating the role of BNST OXT signaling in prosocial interactions should utilize cell-type and circuit-specific tools that can perform in-vivo manipulations of BNST OXT sources individually and in concert.

Healthy prosocial behaviors are disrupted by a variety of environmental factors, most prominently stress and drugs of abuse. For example, social defeat stress, maternal separation stress, drug withdrawal, and chronic social isolation stress induce social avoidance of and promote social vigilance toward same-sex conspecifics (Overstreet et al., 2002; Blanco-Gandía et al., 2015; Becker et al., 2017; Duque-Wilckens et al., 2018; Matthews & Tye, 2019). Stress and/or drugs of abuse are also significant factors leading to the development of various psychiatric disorders that are characterized by impairments in prosocial behavior, including social anxiety disorder (SAD), substance use disorder (SUD), anti-social personality disorder, and depression (Jamieson et al., 2013; Cacioppo & Cacioppo, 2014; Blanco-Gandía et al., 2015; Castellano et al., 2015). In order to develop more effective treatments for psychiatric patients displaying social behavioral symptoms, a better understanding of the neurobiological substrates that drive these symptoms is required. Given its role in healthy prosocial behaviors (see above) and its responsivity to stress and drugs of abuse (Kash et al. 2015), the BNST likely underlies multiple aspects of stress and drug-induced impairments in prosocial behavior.

Similar to maternal behavior, impairment of same-sex prosocial behavior by stress appears to be mediated, at least in part, by dysregulation of BNST CRF and OXT systems. Intranasal OXT treatment reduces prosocial approach, increases social vigilance (orienting towards social target but not approaching it), and increases expression of the immediate-early gene early growth response 1 (Egr1) in the anterior BNST of female, but not male, California mice exposed to social defeat stress (Duque-Wilckens et al., 2018). Infusion of an OXTR antagonist into the anterior BNST of stressed female California mice increases social approach and reduces social vigilance (Duque-Wilckens et al., 2018). However, inhibition of OXT production in the BNST using a morpholino targeted to OXT mRNA prevents social defeat stress-induced impairments in social approach and social vigilance in both male and female California mice. Remarkably, vicarious social defeat stress in female C57BL6/J mice, whereby a female mouse observes social defeat of a familiar male by a male aggressor, increases social vigilance and the number of OXT-containing cell bodies in the ventral BNST (Duque-Wilckens et al., 2020). Thus, contrary to PVN OXT neurons projecting to the BNST, OXT-producing cells within the ventral and anterior BNST appear to negatively impact prosocial behaviors in both stressed and unstressed states. While the BNST CRF system also appears to negatively influence prosocial behaviors in the context of stress, this idea is informed by a relatively small number of studies. One found that intra-BNST delivery of a dual CRF receptor antagonist normalizes social avoidance following social defeat stress (Jasnow et al., 2004). Another found, somewhat surprisingly, that antagonism of CRF-BP in the BNST, which is generally thought to increase the availability of ligands for CRFR1, prevents reductions in social approach induced by social defeat stress (Vasconcelos et al., 2019). However, there have been some reports that CRF-BP can act in concert with CRFR2 to promote CRF’s effects (Ungless et al., 2003), which could better explain these results.

Some recent studies suggest that the BNST may differentially modulate social behavior depending on the stress history of the experimental animal. Cellular and synaptic adaptations occurring in BNST neurons as a result of stress exposure likely underlie these effects, but this has not been widely investigated. Social defeat stress, which induces prosocial avoidance, reduces GABA synthesis and increases brain derived neurotrophic factor (BDNF) protein and mRNA in the BNST (Greenberg et al., 2014a; Makinson et al., 2015). While intra-BNST infusion of a tyrosine receptor kinase-b (Trk-B) antagonist, which blocks BDNF’s activity, normalizes social avoidance induced by social defeat stress, it has no effect in unstressed mice (Greenberg et al., 2014a). Similarly, non-conditional inhibition of BNST neurons normalizes maternal separation stress-induced reductions in social preference and recognition, but does not have effects in control mice (Emmons et al., 2020). On the other hand, non-conditional chemogenetic activation of BNST neurons has no effect in mice that underwent juvenile maternal separation stress but is successful in reducing social preference in control mice (Emmons et al., 2020). This indicates that maternal separation stress induces such an intense state of BNST hyperactivity that exogenous activation is incapable of impairing social behavior further (a ceiling effect). Future studies should investigate whether cellular, molecular, and synaptic adaptations in particularly stress-responsive BNST neural populations, such as CRF neurons, underlie these effects of juvenile maternal separation stress on adult prosocial behavior.

Social withdrawal in SUD patients negatively contributes to treatment outcomes and increases incidences of relapse to drug and alcohol taking (Wise & Koob, 2014; Heilig et al., 2016). Moreover, social exclusion is associated with higher rates of SUD (Todd et al., 2004), suggesting that the neurobiological systems mediating responses to drugs and social interaction may dynamically regulate each other. While acute drug and alcohol intoxication tends to increase prosocial behavior in rodents due to reductions in anxiety and altered reward sensitivity, withdrawal from chronic drug and alcohol use is associated with reductions in prosocial behaviors (Varlinskaya & Spear, 2015; Becker et al., 2017). While many BNST systems are affected by drug and alcohol exposure (Kash et al., 2015), few of them have been explored as potential mediators of substance-induced social disruption. Despite this, there is some evidence that alterations in BNST serotonin signaling induced by chronic alcohol exposure contribute to alcohol-induced impairments in prosocial behavior during withdrawal. Marcinkiewcz et al. found that systemic treatment with a serotonin 2c receptor (5HT2cR) antagonist normalizes deficits in prosocial behavior and reduces ventral BNST hyperactivity associated with alcohol withdrawal (Marcinkiewcz et al., 2015). Furthermore, alcohol withdrawal increases neuronal excitability in the ventral BNST in a 5HT2cR-dependent manner, and intra-BNST infusion of a 5HT2cR antagonist is sufficient to normalize alcohol-induced social dysfunction. These results suggest that the BNST serotonin system, and particularly 5HT2cRs, are important for mediating the negative effects of alcohol withdrawal on social behavior. Whether 5HT2cRs play a similar role in social dysfunction associated with other drugs of abuse remains to be tested. The reduction of drug-withdrawal-related stress by social interaction may involve CRF signaling in the BNST, as a very recent study found that prosocial encounters can reduce the expression of BNST CRF mRNA induced by chronic cocaine treatment in the conditioned place preference task (Lemos et al., 2020). This indicates that states of heightened stress induced by drug-related adaptations in the BNST may be buffered by social interactions.

Human studies implicating the BNST in social behavior

Despite the importance of social behavior in daily life and its disruption in human disease, very little is known about the BNST’s role in social behavior in humans. This is largely due to the technical limitations of non-invasive neuroimaging techniques like functional Magnetic Resonance Imaging (fMRI), which have historically not provided sufficient spatial resolution to parse the BNST specifically. However, with the recent development of novel tools and methods that improve this spatial resolution, researchers have begun to investigate BNST function in human patients. Indeed, dysregulation of BNST function is a common feature among patients with generalized anxiety disorder and is thought to contribute to hypervigilance and maladaptive responses to uncertain threats (Buff et al., 2017; Brinkmann et al., 2018). This finding was recently extended to patients displaying social anxiety disorder (SAD). For example, patients with SAD display increased phasic BNST responses to temporally unpredictable aversive non-social stimuli (Figel et al., 2019). In addition, SAD is associated with greater functional connectivity between the PFC and BNST, as well as the anterior cingulate cortex (ACC) and the BNST, during exposure to temporally unpredictable aversive non-social stimuli (Clauss et al., 2019). Unfortunately, whether the BNST is also important for mediating aversive responses to social stimuli in SAD patients has not been tested. Interestingly, a very recent study found an association between BNST activation and daily helping (opening doors for people, giving directions, taking photos, etc.) in both normal adults and adults who had engaged in acts of extraordinary altruism (Vekaria et al., 2020). BNST activation during an empathy task predicted acts of everyday helping over a subsequent 14 day period in both groups and was particularly strong for acts of helping strangers and proactive helping. These results may indicate that in healthy patients with relatively low baseline BNST activation, the BNST enhances vigilance for socio-affective environmental cues to promote engagement in prosocial behaviors.

Conclusions and future directions

While much work remains to be done, there is a clear body of evidence implicating the BNST in a wide variety of rodent social behaviors. So far, this evidence suggests that the BNST guides both sex-dependent and sex-independent social behaviors in relatively similar ways. Across social behaviors, the BNST facilitates accurate social recognition and social context assessment through the integration of relevant sensory information: What is the sex, sexual history, stress history, age, and novelty of a given conspecific? The BNST assesses the valence of a given social interaction: Does the conspecific pose a physical or emotional threat, or will the interaction be rewarding? This depends of course on the identities of the animals engaged in the social interaction, their internal motivational states, and the context of the social interaction. Unfortunately, this role of the BNST has been less widely explored than the role of the BNST in social recognition. To elucidate the precise role of the BNST in social reward and motivation, future studies should utilize a wider range of behavioral models, including social conditioned place preference and self-administration. Finally, once the BNST has processed this incoming information regarding internal and external stimuli, it guides appropriate behavioral responses through outputs to specific downstream regions: Should the conspecific be ignored, avoided, or approached? If the conspecific is approached, how should it be engaged with?

As mentioned above, exposure to stress or drugs of abuse induce myriad alterations in BNST neurons that drive negative behavioral states. As a result, BNST manipulations often have different effects on social behavior in stress/drug exposed animals versus control animals. It is likely that exposure to stress or drugs alters the baseline activity of many BNST cell populations such that further exacerbating these BNST alterations has no further impact on stress- or drug-induced social impairments. Conversely, driving these BNST alterations in control animals may have no effect or an entirely opposing effect, as the homeostatic starting point is much different. The mechanisms by which these divergent effects occur likely involve micro-circuit interactions between BNST systems controlling innate social behaviors and BNST systems mediating responses to stress and drugs. Elucidation of the specific BNST systems that are responsible for driving the negative effects of stress and drugs on social behaviors will help inform future treatment options for many neuropsychiatric patients suffering from social dysfunction.

Figure 1:

Diagram of BNST sub-regions according to the Allen Brain Adult Mouse Reference Atlas and list of social behaviors associated with each sub-region; purple regions with solid outline = anterior/dorsal BNST, green regions with solid outline = anterior/ventral, purple regions with dashed outline = posterior/dorsal, green regions with dashed outline = posterior/ventral. Sub-nuclei abbreviations: am = anteromedial, ov = oval, al = anterolateral, fu = fusiform, ju = juxtacapsular, dm = dorsomedial, mg = magnocellular, pr = principal, v = ventral, rh = rhomboid, se = strial extension, d = dorsal, if = intrafascicular, tr = transverse.

List of abbreviations

5HT2cR

Serotonin 2c receptor

ACC

Anterior cingulate cortex

ACh

Acetylcholine

AOB

Accessory olfactory bulb

AR

Androgen receptor

AVP

Arginine vasopressin

AVPR1a

Arginine vasopressin receptor 1a

BDNF

Brain-derived neurotrophic factor

BNST

Bed nucleus of stria terminalis

CART

Cocaine and amphetamine

CCK

Cholecystokinin

CeA

Central amygdala

CRF

Corticotropin-releasing factor

CRF-BP

Corticotropin-releasing factor binding protein

CRFR1

Corticotropin-releasing factor receptor 1

CRFR2

Corticotropin-releasing factor receptor 2

DRN

Dorsal raphe

Dyn

Dynorphin

Enk

Enkephalin

GAD

General anxiety disorder

LHb

Lateral habenula

LS

Lateral septum

MeA

Medial amygdala

mGluR7

Metabotropic glutamate receptor 7

mPFC

Medial prefrontal cortext

NAc

Nucleus accumbens

NC

Nociceptin

NMDAR

N-methyl-D-aspartate receptor

NPY

Neuropeptide Y

NTS

Neurotensin

OB

Olfactory bulb

OXT

Oxytocin

OXTR

Oxytocin receptor

PACAP

Pituitary-adenylate cyclase activating polypeptide

PAG

Peri-aqueductal gray

PTSD

Post-traumatic stress disorder

PVN

Paraventricular hypothalamus

RI

Resident intruder

SAD

Social anxiety disorder

SOM

Somatostatin

SUD

Substance use disorder

Trk-B

Tyrosine receptor kinase B

UCN1

Urocortin 1

vGAT

Vesicular GABA transporter

vGlut2

Vesicular glutamate transporter 2

vGlut3

Vesicular glutamate transporter 3

VMH

Ventromedial hypothalamus

VTA

Ventral tegmental area

Data Availability

There is no original data associated with this manuscript.