Hypothalamic control of innate social behaviors

Abstract

Sexual, parental, and aggressive behaviors are central to individual’s reproductive success and species survival. Thus, hardwired neural circuits have evolved to support the inborn expression of these behaviors. Three interconnected medial hypothalamic nuclei, including the medial preoptic nucleus (MPN), ventrolateral part of the ventromedial hypothalamus (VMHvl), and ventral premammillary nucleus (PMv), are key driver of social behaviors and collectively referred to as the medial hypothalamus reproductive behavior control column (RBCC). RBCC cells integrate diverse internal cues indicative of animals’ physical fitness and reproduction readiness and adjust animals’ physical activity, hence the chance of social encounters. RBCC further engages the mesolimbic dopamine system to maintain social interest and reinforce cues and actions time-locked with social behaviors. Additionally, we propose a RBCC and brainstem dual-controlling system for generating moment-to-moment social actions. This review summarizes the recent progress regarding the identities of RBCC cells and their pathways driving different aspects of social behaviors.

Introduction

The medial hypothalamus houses three heavily interconnected nuclei, MPN, VMHvl, and PMv that constitute the reproductive behavior control column (RBCC) (1). The RBCC is indispensable for all social behaviors sub-serving reproduction, including mating, parenting, and fighting. It is a part of the larger social behavior network (SBN), proposed initially by S. Newman in 1999 based on rodent studies (2) and later extended by J.L. Goodson to other vertebrate species (3). Anatomically, RBCC is highly evolutionarily conserved, recognizable in mammals, birds, reptiles, amphibians and fish (3, 4). Immediate early gene mapping suggests similar response patterns during social behaviors across species (3, 5). Here, we will summarize recent studies, mainly in mice, regarding the molecular identity, functions and responses of RBCC cells. Interested readers could refer to these reviews (4, 6) for RBCC function in non-rodent species.

The review will be separated into two parts. The first part summarizes recent studies regarding the functions and responses of molecularly defined RBCC cells during each social behavior. We will also offer our views regarding the general principles of RBCC organization and the relationship among different social circuits. The second part of the review discusses how RBCC controls each stage of social behaviors, from increasing social exploration, maintaining social interest, to promoting consummatory social behaviors. In particular, we propose that RBCC acts as a “permissive gate” to allow specific brainstem/spinal cord motor circuits to respond to immediate and simple sensory inputs to drive moment-to-moment social actions.

The role of RBCC in sexual, parental, and aggressive behaviors

Male sexual behaviors:

Sexual behaviors differ qualitatively between sexes. Thus, the male and female sexual behavior circuits have little in common. For males, the medial preoptic area (MPOA), which includes MPN and its surrounding region, has been recognized as a site of paramount importance since early 1960 (7, 8). Recently, single cell/nucleus RNA sequencing revealed high molecular heterogeneity of MPOA cells, leading to a series of studies to refine the molecular identities of mating-related MPOA subpopulation (9, 10). In 2018, Wei et al. showed that optogenetic activation of MPOA estrogen receptor alpha expressing cells (MPOA^Esr1) could elicit male mounting albeit with relatively low efficiency (50% animals)(11). In 2021, Kairgo et al. reported that activating GABAergic MPOA^Esr1 (MPOA^Esr1∩Vgat) cells can induce male mounting with 100% efficiency (12). Most recently, Bayless et al. pinpointed tachykinin receptor 1 (Tacr1) expressing cells (MOPA^Tacr1), likely a subset of MPOA^Esr1∩Vgat cells (10), as the key population for male sexual behaviors (13). Strikingly, optogenetic activation of MOPA^Tacr1 cells induces intromission-like behaviors towards a toy mouse (13). Importantly, Tacr1 is itself functionally critical for the behavior as antagonizing MPA Tacr1 suppresses male mounting (13).

The role of VMHvl in male sexual behaviors is minor. Optogenetic activation of VMHvl^Esr1 cells at a low intensity induced mounting in male mice, but it was not coupled with USV production and hence being interpreted as a dominance instead of sexual behavior (12, 14). Although ablating VMHvl progesterone receptor-expressing cells (VMHvl^PR) or chemogenetic inhibition of VMHvl^Esr1 cells (Esr1 and PR overlap nearly 100% at VMHvl) suppressed male mounting (12, 15), optogenetic inhibition of VMHvl^Esr1 cells did not disrupt ongoing copulation, suggesting that VMHvl’s role in male sexual behavior is likely limited to the initiation phase (14). This functional result is supported by the response pattern of VMHvl cells: while the cells increase activity during mounting, they decrease activity during intromission (16, 17).

PMv receives dense inputs from the medial amygdala – a social odor and pheromone processing region, and thus is considered the sensory relay of RBCC (18). Indeed, PMv expresses abundant c-Fos after conspecific odor presentation and shows moderate activity increase during the male-female investigation (17, 19). However, PMv cells do not respond during any phases of male copulation, and PMv deficits fail to impair male reproduction (17, 20, 21).

Thus, MPOA is the primary site mediating male mating while VMHvl and PMv likely play minor roles in early phase of male-female interaction (Fig. 1). MPOA^Tacr1 cells represent the most refined population for male sexual behaviors (Fig. 2).

Fig. 1. The role of RBCC regions in various social behaviors.

White indicates that the function of the region is to be determined. Gray indicates the region has no function. Purple indicates that the region suppresses the behavior. Yellow indicates that the region promotes the behavior. Note that the MPOA and the VMHvl and PMv generally play opposite roles in each social behavior.

Fig. 2. The known molecular markers for social behavior–relevant cells in the RBCC.

Female sexual behaviors:

Female sexual behavior is remarkably simple in its motor output. As males mount, females stay stationary with back arching downward to facilitate penile insertion, a posture known as lordosis (22). Since 1970, many lesion and stimulation experiments demonstrated a critical role of the VMHvl in female sexual behaviors (22). Similar to studies in males, recent efforts focused on refining the molecular identity of VMHvl cells relevant to female sexual behaviors. Towards this goal, VMHvl^PR cells were identified as necessary for female sexual receptivity in mice (15, 23). However, activating VMHvl^PR cells failed to facilitate lordosis (23). This surprising negative result is likely because female VMHvl contains two subdivisions, both expressing abundant PR and Esr1, but only the lateral subdivision (VMHvll) is strongly activated during female sexual behaviors (24). When the VMHvl cells expressing Cckar (VMHvll^Cckar), a lateral subdivision-specific gene, are chemogenetically or optogenetically activated, female sexual receptivity rapidly increases (25, 26). In mice, female sexual receptivity is tightly coupled with ovulation, i.e., the estrus cycle. The in vivo responses and physiological properties of VMHvll^Cckar cells, including intrinsic excitability and synaptic inputs, vary over the estrus cycle, suggesting that these cells likely support the cyclic changes in female sexual behaviors (26).

PMv is also critical for female reproduction, but its role differs from VMHvl. Specifically, PMv enables metabolic signals to modulate the reproductive neuroendocrine axis. Leptin, an adipocyte-derived hormone, signals energy reserve levels and triggers sexual maturation by acting on PMv leptin receptors (Lepr)(27). Upon sensing leptin, PMv^Lepr cells activate kisspeptin-expressing cells to release gonadotropin release hormone (GnRH) and trigger the onset of puberty (20, 27). In adults, PMv impairment decreases ovulation frequency, an abnormality that could occur naturally during leptin deficiency caused by food deprivation (20, 28). Thus, PMv modulates sexual readiness based on the energy reserve level. During food shortage, PMv signals the hypothalamic–pituitary–ovarian (HPO) axis to slow down female reproduction.

MPOA is consistently found to suppress female receptivity, possibly through its strong inhibitory inputs to VMHvl (12, 29). MPOA lesion or site-specific Esr1 knockdown increases lordosis, while electric stimulation of MPOA has the opposite effect (30–34).

Thus, contrary to male mating circuit, female sexual behaviors are mainly mediated by VMHvl and PMv, while MPOA likely plays a negative role (Fig. 1). VMHvll^Cckar and PMv^Lepr represents the most refined populations for female sexual behavior so far (Fig. 2).

Parental behaviors:

In rodents, all parental behaviors, except nursing, could be exhibited by both males and females with some quantitative differences (35). Thus, perhaps unsurprisingly, the hypothalamic regions controlling parental behaviors are largely similar between sexes. Since 1974, the MPOA has been recognized as a crucial region for controlling maternal behaviors (36, 37). In 2014, Wu et al. identified MPOA cells expressing galanin (MPOA^Gal) as a critical population for regulating parental behaviors, especially grooming, in male and female mice (38). In 2018, Wei et al. and our study found that both male and female MPOA^Esr1 cells in mice are necessary, sufficient, and naturally active during pup approach and retrieval (11, 39). Later, multiplexed error robust fluorescence in situ hybridization (MERFISH) revealed that calcitonin receptor (Calcr), which is expressed in a subset of Esr1 cells, preferentially marks MPOA cells activated by parental behaviors (9). Following this finding, Yoshihara et al. reported in 2021 that silencing MPOA^Calcr neurons or knocking down Calcr in the MPOA suppresses maternal behaviors (40). Gal, Esr1 and Calcr are expressed in both GABAergic and glutamatergic MPOA cells. However, the glutamatergic MPOA cells likely do not promote parental behaviors as optogenetic silencing of MPOA glutamatergic cells increase, instead of decreasing, pup retrieval in female mice (41). Lastly, MPOA cells not only drive parental behaviors, but also suppress hostile behaviors towards the pups. During lactation, MPOA^Esr1 cell excitability increases, causing stronger inhibition of infanticide-driving BNSTpr^Esr1 cells and enabling the drastic switch from hostile to caring behaviors towards the young (42).

In contrast to the pivotal role of MPOA in parental care, VMHvl and PMv are dispensable. VMHvl inactivation or PMv lesion did not cause any deficit in maternal care (24, 43). In vivo recording found no activity change of VMHvl^Esr1 cells during maternal behaviors (24). Recent studies further suggested a negative role of VMHvl in parental behaviors. Activating MPOA connecting VMHvl cells or perifonical urocortin-3 (Ucn3)-expressing cell (PeFA^Ucn3) projection to the VMH strongly suppressed pup investigation in virgin female mice (42, 44).

Altogether, MPOA, but not VMHvl and PMv, is the key RBCC region to drive parental behaviors in both sexes (Fig. 2a). MPOA^Calcr cells represent the most refined population for parental behaviors (Fig. 2b).

Aggressive behaviors:

While the tendency to attack differs between sexes, its motor pattern does not, at least in rodents (45). Thus, the neural substrates of aggression are qualitatively similar in males and females. Studies in the last decade firmly established the central role of VMHvl in male and female aggression (46). Inactivating or ablating VMHvl cells abolishes natural inter-male aggression and maternal aggression in mice (14, 15, 24, 47). Conversely, activating VMHvl cells, especially those expressing Esr1 or PR, promotes attack towards both natural and suboptimal targets (14, 24, 47) regardless of the subject’s social status, housing condition, or testing context (48). Simultaneous recording of 13 limbic regions, including 5 in the hypothalamus, revealed the VMHvl as the region with the largest and fastest activity increase during attack onset, highlighting its crucial role in the behavior (17). Furthermore, recent works highlighted the flexibility of VMHvl cell responses. For example, with sexual experience, the male- and female-induced activation patterns in the VMHvl become more distinct (49). With winning experience, the VMHvl cells show long-term potentiation of excitatory synaptic inputs (50). When an arbitrary motor action, e.g., nose poke, is associated with future opportunities to attack, VMHvl cells increase activity prior to poking (51). Lastly, Yang et al. recently reported increased VMHvl cell activity when animals witness fights between others (52). Thus, the VMHvl cells carry diverse aggression-related information, including aggressive motivation, aggression-provoking sensory cues, the motor execution of attacks, and fighting experiences of own and others. Notably, only the posterior VMHvl is related to aggression; the anterior VMHvl mediates conspecific self-defense and social fear (53, 54).

In female mice, only medial VMHvl (VMHvlm) is relevant for aggression (24). Using activity-dependent single-cell RNA sequencing (Act-seq), Liu et al. identified Npy2r, a VMHvlm-biased gene, as a genetic marker for the female aggression population (55). When Npy2r+, but not Npy2r-Esr1+ VMHVl cells, are optogenetically activated, virgin female mice attack various social targets, including adult males (55). VMHvl^Npy2r cells show reproductive state-dependent activity changes, with the highest response during lactation when the female aggression level peaks (55).

PMv is also a critical site for both male and female aggression. Several recent studies targeted dopamine transporter (DAT) expressing cells in the PMv (PMv^DAT) and showed aggression can be bi-directionally modulated in male and female mice (21, 56, 57). In vivo recording shows that PMv cells, like VMHvl cells, respond preferentially to aggression-provoking social cues (17, 21, 58). However, PMv differs from the VMHvl in that it responds more during social investigation than attack (17, 21). Lesioning the PMv drastically reduces aggression-induced c-Fos in the VMHvl, suggesting that PMv is upstream of VMHvl, likely relaying conspecific cues (43). Notably, regardless of sex, PMv^DAT cells are quiescent and hyperpolarized in non-aggressive mice and become spontaneously active and depolarized in aggressive mice (56, 57), suggesting the aggressiveness of an animal could be encoded by the biophysical properties of PMv cells. It is worth mentioning that although DAT is a good marker for PMv as it uniquely expresses in the PMv among neighboring regions, PMv^DAT cells do not appear to synthesize dopamine and it remains unclear whether DAT+ cells are more involved in aggression than DAT- cells (58).

Recent studies support an aggression-suppressing role of MPOA^Esr1 cells through their inhibition onto VMHvl glutamatergic cells (12, 29). Rostral MPOA^Esr1 cells are activated during male sexual behaviors while caudal MPOA^Esr1 cells preferentially respond to superior male opponents (29). Thus, MPOA could suppress aggression towards inappropriate target in various contexts. Indeed, when MPOA^Esr1 to VMHvl terminals are optogenetically inactivated, male mice attack social targets indiscriminately, including superior males, and are ended up inflicting more defeat (29).

Altogether, VMHvl and PMv are central for male and female aggression while MPOA suppresses the behavior (Fig. 1). VMHvl^Esr1/PR in males and VMHvl^NPY2r in female the best-defined populations for aggression (Fig. 2).

General organization of RBCC cells.

When considering the circuits for all social behaviors together, several general conclusions emerge. First, VMHvl and PMv have similar social functions while opposite to MPOA (Fig. 1). For example, VMHvl and PMv promote aggression in both sexes and sexual behaviors in females, while MPOA activation suppresses these behaviors (12, 14–16, 24, 26, 29, 55). MPOA promotes parental behaviors while VMHvl reduces them (11, 38–40, 42, 59). The only exception is male sexual behaviors, of which the initiation is promoted by both MPOA and VMHvl, although MPOA increases while VMHvl decreases activity during intromission (12, 14, 17). The functional relationship of MPOA, PMv, and VMHvl probably root in the neurotransmitter types of the cells. While PMv and VMHvl are overwhelmingly glutamatergic, socially relevant cells in the MPOA are likely GABAergic (12, 41). As these three regions are heavily reciprocally connected, activating MPOA social cells should suppress PMv and VMHvl cells, while PMv and VMHvl cells should facilitate activation of each other (29, 58).

Second, cells supporting distinct social behaviors often occupy different subregions within the VMHvl or MPOA (Fig. 2). For example, lateral and medial VMHvl mediate female sexual and aggressive behaviors, respectively (24, 26, 55). Rostral and caudal MPOA cells in males are activated during female and dominant male interactions, respectively (29). Fos Catfish mapping suggests that mating and parental care-induced c-Fos are largely distinct in the MPOA and the former could be more medially located (see Figure 2 in (38)). At the single cell level, majority of cells show biased, though not necessarily exclusive, response during one social behavior (39, 47, 49, 60). Thus, individual MPOA and VMHvl cells are likely preferentially involved in one social behaviors and cells with similar response patterns tend to cluster. Whether such topographic organization of social behaviors also exists in the PMv remains to be elucidated.

Third, fighting cells and mating cells in the RBCC antagonize each other in both sexes. In males, activating mating-related MPOA^Esr1 cells suppresses inter-male aggression whereas activating fighting-related VMHvl^Esr1 cells suppresses male-female mounting and female urine-elicited USV (12, 29). In females, activating fighting-related VMHvl^Npy2r cells suppresses sexual receptivity (as the female attacks the male) while activating mating-related VMHvl^Cckar cells suppresses female aggression (26, 55). This agonistic relationship is not simply due to motor incompatibility. In lactating females, VMHvll^Cckar cell activation could not increase female sexual receptivity possibly due to changes in the downstream circuit but remains effective in suppressing maternal aggression (26). The relationship between parenting cells and mating/fighting cells are less understood as these three behaviors rarely examined together. Limited evidence suggests that these cells could operate independently. For example, after VMHvll^Cckar are inactivated, there is no change in maternal behaviors (26). Rather than counteracting with mating/fighting cells, parenting cells form strong mutual inhibition with infanticide cells: activating infanticide cells suppresses parenting cells and vice versa (42). These circuit relationships make sense when considering the social target of each behavior. As mating and fighting are both directed to adult conspecific with similar physical features, the mutual inhibition between these circuits ensure one behavior output dominates. In contrast, parenting and mating/fighting are activated by highly distinct social targets, making the cross-activation unlikely and hence mutual inhibition unnecessary.

Lastly, as the number of studies increase, the “molecularly defined populations” for each behavior will keep growing. Gal, Esr1 and Calcr expressing cells in the MPOA have all been found important for driving parental behaviors based on recording and functional studies (38–40) (Fig. 2). These cells are neither distinct nor identical. MPOA^Gal cells partially overlap with MPOA^Esr1 cells but largely different from MPOA^Calcr cells while Calcr is expressed in a subset of MPOA^Esr1 cells (9). How to determine whether a gene truly marks the social behavior population or a random population containing some social behavior relevant cells by chance? Indeed, if a behavior is the dominant output of a brain region, activating a random subset of cells in the region, marked by any one of thousands of expressing genes, may drive the behavior. For a gene to be considered a relevant maker, we think two criteria should be met. First, the gene-positive cells should show higher activity change during the behavior than the gene-negative cells. Second, functional manipulation of the gene-positive cells should influence the behavior more profoundly than the gene-negative cells. For example, Esr1-positive MPOA cells show higher responses than Esr1-negative during parental care (39). Activating Esr1-positive but not negative cells in the VMHvl can drive male aggression (14). Additionally, it is unlikely, although not absolutely impossible, that a gene is uniquely expressed in the social behavior population coincidently. More plausibly, it is expressed for a reason, e.g. to modulate the behavior output. Thus, a good gene marker is probably also functionally important for the behavior. Indeed, Esr1 or Calcar knockout in the MPOA impairs parental behaviors while antagonizing MPOA Tacr1 impairs male mating (13, 40, 61). Based on these rationales, receptors of neuropeptides or hormones that modulate a social behavior uniquely could be the top candidates for marking specific social behavior populations.

RBCC coordinates multiple aspects of social behaviors

We propose that the RBCC achieves social behavior control by (1) promoting exploration to increase the probability of social encounter; (2) maintaining social interest when a social target is encountered; (3) permitting the consummatory social actions at the appropriate times. Here, we will review the neural pathways downstream of the RBCC that mediate each aspect of these behavioral controls.

Social encounter probability.

Animals must physically encounter each other to engage in social behaviors. We consider the first role of RBCC to modulate an animal’s physical activity, hence the likelihood of social encounters, based on an animal’s internal state. In support of this idea, animals increase locomotion when the MPOA^Vgat, VMHvl^Esr1, or VMHvl^Nkx2−1 cells are artificially activated (62–65). Conversely, loss of Esr1 or Nkx2–1 in the VMHvl reduces physical activity and causes obesity in female mice (63, 66). VMHvl^Esr1 projection to dorsal raphe and MPOA^Vgat projection to PAG appears important from RBCC control of physical activity (64, 65) (Fig. 3).

Fig. 3. Pathways extended from the RBCC that mediate various aspects of social behaviors.

RBCC expresses abundant neuromodulator, neuropeptide, and hormone receptors, allowing various small molecules indicative of animals’ reproductive, metabolic, circadian, and other internal states to adjust the ongoing cell activity and, consequently, animals’ physical activity (Fig. 3). For the reproductive state, estrogen is a crucial hormone signaling ovarian activity. Estrogen supplement drastically increases the excitability of MPOA and VMHvl cells (67, 68). This estrogenic modulation likely underlies the increased spontaneous activity of VMHvl cells during estrus (26, 69) and the increased activity of MPOA cells during motherhood (39, 42). In addition to sex hormones, metabolic cues also modulate VMHvl^Esr1 cell activity. For example, 24-hour fasting decreases the excitability of VMHvl^Esr1 cells, which can be reversed with refeeding (65). Glucose may mediate these changes as nearly all VMHvl cells are responsive to glucose in vitro (70). The circadian clock could also modulate the ongoing VMHvl activity. Melanocortin receptor 4 (MCR4), a female VMHvl enriched gene, increases expression when the estrogen level is high, enabling nighttime melatonin to elevate VMHvl cell activity and promote exploration during proestrus when the female is getting ready to mate (71, 72). Beyond the internal cues, the social environment could also influence the ongoing activity of RBCC cells. Recently, Fukumitsu et al. found that prolonged social isolation decreases amylin and its receptor expression in the MPOA and reduces social-seeking behaviors (73). Thus, the ongoing RBCC cell activity is dynamically modulated by the internal state and external environment, which in turn adjusts an animal’s physical activity level and the probability of social encounter.

Social interest level.

Once the animals encounter each other, the next step is to recognize the identity of the social target. Although VMHvl and MPOA lesion or inhibition consistently change social preference, this likely does not reflect a deficit in sex discrimination given that the manipulated animals often show reversed instead of no preference, suggesting they remain capable of discriminating males vs. females (26, 74, 75). Indeed, sex identity information is widely distributed in the limbic system, including regions upstream of the medial hypothalamus (17, 60, 76, 77). Thus, the male and female-induced neural activation patterns should remain different even without RBCC. Therefore, we reason that the RBCC deficit-caused changes in social preference likely reflect decreased interests in certain social targets instead of an inability to discriminate between sexes.

RBCC likely promotes and maintains social interest by engaging the mesolimbic dopaminergic pathway (Fig. 3). Dopamine release in the nucleus accumbens (NAc) during initial social encounters is positively correlated with the total interaction time (78). Inhibiting ventral tegmental area (VTA) dopamine neurons decreases social interaction with an unfamiliar social target and the number of nose-pokes an animal is willing to emit to access a conspecific (79–81). Conversely, optogenetic activation of the VTA dopaminergic projection to the NAc drastically increases social interaction time (78, 80).

RBCC, especially MPOA, likely engages the mesolimbic dopaminergic pathway by direct projection to the VTA. Indeed, optogenetic activation of the MPOA to VTA terminals evokes dopamine release in the NAc and promotes interaction with pups or potential mates (39, 67). Additionally, activating medial amygdala GABAergic projection to the MPOA increases NAc dopamine release and promotes nose poking for social targets (82). ChR2-assisted circuit mapping revealed that MPOA^Esr1 cells preferentially target VTA GABAergic cells, suggesting a disinhibition mechanism underlying the increased dopamine release (39). Similar to MPOA, VMHvl also promotes social seeking and approach. In male mice, optogenetic activation of VMHvl cells shortens the latency to nose poke for a weaker male intruder, whereas inhibiting the cells has the opposite effect (83). Activating female VMHvll^Cckar cells promotes social interest in males (26). VMHvl activation may evoke dopamine release directly through its glutamatergic projection to VTA dopaminergic cells (VTA^DA) or indirectly via its projection to MPOA, although this remains to be investigated experimentally (84).

It is worth noting that MPOA/VMHvl-VTA^DA-NAc circuit is also essential for the reinforcing property of consummatory social actions (Fig. 3). During social behaviors, DA level increases in the NAc, and animals subsequently develop a preference for the contexts, actions, or specific conspecific cues associated with these experiences (78, 85–87). When VTA DA release or NAc dopamine receptor is blocked, the social experience-induced reinforcement diminishes (86–89). Conversely, when the socially relevant cells in the MPOA and VMHvl are optogenetically activated, it quickly enhances the preference of the stimulation-coupled contexts and reinforces the actions preceding the stimulation (26, 41, 67). Taken together, it suggests that the increased RBCC activity during consummatory social actions leads to NAc dopamine increase and reinforces sensory cues and motor actions that are time-locked to the behavior.

Consummatory social actions.

Sexual, aggressive, and parental behaviors are recognized based on their stereotyped and unique motor patterns. RBCC output to the midbrain, especially periaqueductal gray (PAG), has been recognized as a key pathway for the motor execution of social actions (90–93) (Fig. 3). VMHvl mainly targets the dorsomedial and lateral PAG, matching the male aggression-induced c-Fos pattern, whereas MPOA mainly targets ventrolateral PAG, consistent with male mating-induced c-Fos (84, 90, 91, 93, 94). PMv projection to PAG is relatively sparse (92, 93). Interestingly, RBCC largely avoids the dorsolateral PAG, a region important for defense (93). PAG then relays the hypothalamic signal to spinal cord motor neurons directly or indirectly through the pons and medulla (95).

The role of PAG in female sexual behaviors and aggression, two VMHvl-mediated social behaviors, has been well established. PAG lesion in rats reduces both behaviors while electric stimulation of PAG has the opposite effect (22, 96, 97). More recently, we found that activating VMHvl to the PAG pathway induces attack in male mice, while inactivating PAG blocks VMHvl stimulation-induced attack (98). Interestingly, after PAG inactivation, animals remain highly engaged with the intruder and show attack-like behavior, e.g., lunge, but fail to complete the attack sequence (98). Consequently, the opponent walks away unharmed. Thus, PAG lesion impairs the motor execution of attack but not the underlying aggressive motivation.

The role of PAG in male sexual behaviors and parental behaviors, two MPOA-mediated social behaviors, are less clear. For male mating, c-Fos increased in the PAG (94), but large lesions in PAG accelerate instead of suppressing mounting behaviors (99). For parental behaviors, caudal PAG lesion reduces kyphosis, a supine posture during nursing, but does not affect active parental behaviors, such as pup retrieval (100). Rostral PAG (rPAG) lesioned animals still initiate retrieval but have trouble releasing pups held in the mouth (101). Optogenetic manipulation of MPOA^Gal to PAG projection specifically affects pup grooming (59). Thus, different subregions of PAG are likely involved in different aspects of maternal behaviors and midbrain regions outside of PAG may drive active parental actions. Indeed, early lesion studies suggested that MPOA engages ventral pallidum (VP) through its projection to VTA to mediate pup retrieval (102).

We also hope to offer our general view regarding the RBCC control of social actions (Fig. 4). Electrophysiological recording showed that medial hypothalamic cells do not carry information regarding specific movements during social behaviors. For example, individual VMHvl cell increases activity during the entire attack sequence regardless of whether the animal is lunging, biting, or tumbling (103). Thus, RBCC cells could not instruct moment-to-moment movement; instead, the cells likely serve as a “permissive gate” that allows the motor neurons to drive an action based on immediate sensory cues. We speculate the existence of a brainstem-spinal cord reflex circuit for each social action, e.g., bite, retrieve, or mount, which takes in sensory inputs and activates the local motor neurons to drive coordinated muscle movements (Fig. 4). Visual inputs may enter the circuit at the brainstem level, e.g., through superior colliculus (104), while spinal cord sensory neurons presumably receive somatosensory inputs. The brainstem-spinal cord circuits, however, are likely unable to operate on their own either because the motor neurons are under tonic suppression or the inputs from sensory neurons are not sufficiently robust. Only when RBCC is activated to remove the inhibition or boost the excitatory drive, the motor neurons can respond to the acute sensory inputs. During social encounters, the increased activity from specific cells in the RBCC determines which brainstem spinal cord circuits are permitted to operate. However, the final release of the action should be triggered by the acute sensory inputs to the brainstem or spinal cord. Importantly, we speculate that the sensing cells in the brainstem spinal cord circuits have limited ability to integrate information from various sensory modalities and thus are activated promiscuously if ungated. Therefore, when the medial hypothalamic cells are artificially activated, the animals initiate social actions towards improper targets (12, 13, 47). For example, VMHvl activation can induce attacks towards an inflated glove (47). Although the glove has no resemblance to a mouse in its smell, sound, or shape, its soft and bouncy surface provides sufficient somatosensory input to trigger the brainstem spinal cord biting circuit once its gate is opened by artificial VMHvl activation.

Fig. 4. The RBCC and brainstem–spinal cord dual control of social behaviors.

In rodents, the RBCC receives many internal state and conspecific olfactory cues to determine the type of social behaviors to be executed. The RBCC provides permission to specific brainstem–spinal cord cell groups so that those cells can respond to acute sensory visual and somatosensory inputs and generate motor outputs.

A similar conceptual framework for generating lordosis was proposed by Donald Pfaff (22). He suggested that a series of brain modules mediate lordosis. The spinal cord modules, coordinated by the brainstem module, control the muscle movements in response to the somatosensory inputs to the spinal cord during male mounting while the output of the hypothalamus module to the brainstem is essential for ensuring the lordosis only occurs during estrus (22). We now know that female VMHvl activity increases during estrus and with the proximity of a male (23, 24, 26, 69). Thus, the increased VMHvl output provides a window of opportunity for the brainstem spinal cord circuit to drive lordosis upon receiving the somatosensory cues during male mounting. The dual hypothalamic and brainstem control system ensures that the social action is supported by the animal’s physical and reproductive states and directed towards the right social target (based on hypothalamus input), and happens at the right moment (based on the acute sensory inputs to the brainstem and spinal cord). We speculate that this dual controlling system is a common feature in generating innate social actions, and the same principle may apply to learned social actions that vary rapidly with the opponent’s behaviors (Fig. 4). As we gain a better understanding of the connectivity between the RBCC and the brainstem-spinal cord circuit relevant for each social action, this dual-control system shall be further tested and specified in future studies.

Concluding remarks:

RBCC orchestrates all innate social behaviors sub-serving reproduction. MPOA drives male sexual, paternal, and maternal behaviors, while VMHvl and PMv promote female sexual behavior and aggression in both sexes. MPOA and VMHvl/PMv may have an antagonistic relationship, as indicated by their opposing roles in multiple social behaviors. At the baseline, RBCC adjusts an animal’s physical activity, i.e., the chance of social encounter, based on the hormonal and metabolic signals. Upon encountering a social target, RBCC engages the dopamine system to sustain the social interest and reinforce the actions and contexts leading to the successful completion of social behaviors. PAG has emerged as a critical midbrain relay for executing VMHvl-driving social behaviors, but the midbrain region that transforms MPOA signal into motor actions remains elusive. Regardless of the exact circuit, we propose a hypothalamic and brainstem/spinal cord dual-controlling system for the motor execution of each social action. In this model, medial hypothalamus activity determines the broad behavior category, i.e., aggression, based on the animal’s internal state and opponent’s social identity, and opens the gates to allow specific brainstem-spinal cord circuits to respond to the immediate sensory cues and drive moment-to-moment motor output. Though not discussed here, RBCC also drives a suite of autonomic responses to prepare the body for the social actions, and triggers neuropeptide and hormone releases essential for the reproduction following mating. Lastly, it is worth mentioning that although the RBCC circuit is hardwired developmentally, it remains plastic. The input-output relationship of the circuit can be shaped through experience during development and adulthood, enabling widely different tendencies in expressing social behaviors across individuals (105).