Transcriptionally defined amygdala subpopulations play distinct roles in innate social behaviors

Lischinsky, Julieta E.; Yin, Luping; Shi, Chenxi; Prakash, Nandkishore; Burke, Jared; Shekaran, Govind; Grba, Maria; Corbin, Joshua G.; Lin, Dayu

doi:10.1038/s41593-023-01475-5

Download PDF

Article
Open access
Published: 09 November 2023

Transcriptionally defined amygdala subpopulations play distinct roles in innate social behaviors

Julieta E. Lischinsky ORCID: orcid.org/0000-0003-1664-6642¹,
Luping Yin¹,
Chenxi Shi^1,2,
Nandkishore Prakash³,
Jared Burke^1,4,
Govind Shekaran^1,4,
Maria Grba^1,4,
Joshua G. Corbin³ &
…
Dayu Lin ORCID: orcid.org/0000-0003-2006-0791^1,4,5

Nature Neuroscience volume 26, pages 2131–2146 (2023)Cite this article

12k Accesses
6 Citations
54 Altmetric
Metrics details

Subjects

Abstract

Social behaviors are innate and supported by dedicated neural circuits, but the molecular identities of these circuits and how they are established developmentally and shaped by experience remain unclear. Here we show that medial amygdala (MeA) cells originating from two embryonically parcellated developmental lineages have distinct response patterns and functions in social behavior in male mice. MeA cells expressing the transcription factor Foxp2 (MeA^Foxp2) are specialized for processing male conspecific cues and are essential for adult inter-male aggression. By contrast, MeA cells derived from the Dbx1 lineage (MeA^Dbx1) respond broadly to social cues, respond strongly during ejaculation and are not essential for male aggression. Furthermore, MeA^Foxp2 and MeA^Dbx1 cells show differential anatomical and functional connectivity. Altogether, our results suggest a developmentally hardwired aggression circuit at the MeA level and a lineage-based circuit organization by which a cell’s embryonic transcription factor profile determines its social information representation and behavioral relevance during adulthood.

Chromatin accessibility during human first-trimester neurodevelopment

Article Open access 01 May 2024

Formation of memory assemblies through the DNA-sensing TLR9 pathway

Article Open access 27 March 2024

Unsupervised restoration of a complex learned behavior after large-scale neuronal perturbation

Article 29 April 2024

Main

Innate social behaviors, such as mating, fighting and parenting, are indispensable for the survival and propagation of a species and, therefore, are present widely in the animal kingdom. These behaviors are considered innate as they can take place without learning, although the efficiency in performing these behaviors can be improved with repeated execution¹. The developmental mechanisms for the establishment of innate social behaviors and the role of experience in shaping these circuits remain poorly understood.

An array of interconnected brain regions, collectively called the social behavior network (SBN), was proposed to be important for diverse social behaviors^2,3. The medial amygdala (MeA), especially its posterior division (MeAp), is considered a key node of the SBN based on its connectivity, activity, gonadal hormone receptor expression and numerous lesion studies². MeA is the primary recipient of accessory olfactory bulb (AOB) inputs, the exclusive relay of the vomeronasal organ (VNO) specialized in detecting pheromones⁴. Volatile information also reaches MeA cells via the cortical amygdala^5,6. Consistent with the strong olfactory inputs, immediate early gene mapping, in vivo electrophysiological recordings and Ca²⁺ imaging have all revealed increased MeA activity after exposure to conspecific and heterospecific chemosensory cues^6,7,8,9,10. Unsurprisingly, MeA lesion causes deficits in multiple social behaviors, including male sexual behavior, aggression and maternal behaviors^{11,12,13,14,15}. These studies collectively support an important role for the MeA in processing and relaying olfactory information related to conspecifics.

Recent functional experiments suggest a more direct role of the MeAp in driving social behaviors. Hong et al.¹⁶ first showed that optogenetic activation of GABAergic MeAp cells acutely induced mounting or attack in male mice depending on stimulation intensity. Later, Unger et al.¹⁷ reported that silencing or ablating aromatase-expressing MeAp cells decreased aggression in both sexes. Work from three recent studies^18,19,20 found that activation of the projection from MeA^Npy1r, MeA^D1R or MeA^CaMKII cells to the bed nucleus of the stria terminalis (BNST) promoted male aggression. Beyond aggression, MeA GABAergic cells were also found to drive pup grooming, infanticide and allogrooming^21,22.

These results raised several questions regarding the MeA function in social behaviors. First, are there dedicated MeA subpopulations for distinct social behaviors, or can any random subsets of MeA cells generate any social behavior in a context-dependent and intensity-dependent manner? An answer to this question remains unclear as activating multiple subpopulations of MeA cells can all induce aggression^17,18,19,20, whereas activating the same GABAergic MeA population induces diverse social behaviors^16,21,22. Second, how much of the MeA cell response is developmentally hardwired versus determined by adult experience? Choi et al.⁷ found that MeA cells relevant for social behaviors and predator defense are marked by different members of the Lhx family of transcription-factor-encoding genes, suggesting developmental hardwiring of social versus non-social signals. However, recent imaging studies revealed that MeA cell responses to social stimuli could be altered with adult experience, suggesting that the exact social response of MeA cells may not be pre-determined¹⁰.

Taken together, despite being a central node of the SBN, how the MeA mediates social behaviors remains elusive. We previously identified two distinct MeA populations that arise from separate embryonic lineages in the telencephalic preoptic area (POA), marked by the transcription factors Dbx1 and Foxp2 (refs. ^9,23). In adults, although Dbx1 is no longer expressed in the MeA, Dbx1 lineage cells remain distinct from Foxp2-expressing cells despite being spatially intermingled⁹ (Fig. 1). Addtionally, these two subpopulations differ in their gene expression patterns and intrinsic electrophysiological properties⁹. Therefore, we reason that these two developmentally distinct and transcriptionally defined subpopulations could provide a unique opportunity to address whether social cue representation and social function of MeA cells are predetermined by their developmental lineage. In the present study, we compared the neuronal responses, functions and connectivity of MeA^Dbx1 and MeA^Foxp2 cells in male social behaviors and revealed the response pattern of MeA^Foxp2 cells over development. Our findings highlight the interplay between nature (development) versus nurture (experience) in shaping social sensory representations, supporting a framework by which neuronal function and circuitry are developmentally defined.

**Fig. 1: MeA^Foxp2 and MeA^Dbx1 cells are essentially non-overlapping transcriptionally defined subpopulations.**

Results

Distribution of MeA^Dbx1 and MeA^Foxp2 cells in male mice

To visualize the spatial distribution of MeA^Dbx1 and MeA^Foxp2 cells in adults, we crossed Dbx1^cre mice²⁴ with a ZsGreen reporter line (Ai6)²⁵ and immunostained for Foxp2. MeA^Dbx1 cells make up approximately 28% of total posterior MeA cells (MeAp, bregma level −1.4 mm to −2.1 mm) and are found in both dorsal and ventral subdivisions (MeApd and MeApv) (Fig. 1a–c). In comparison, MeA^Foxp2 cells are relatively fewer, constituting only 10% of MeAp cells and largely absent from caudal MeA (Fig. 1a–c). Between MeApd and MeApv, both MeA^Dbx1 and MeA^Foxp2 cells show a dorsal bias, with approximately twice as many cells in MeApd than MeApv (Fig. 1d). Along the medial-lateral axis of the MeApd, the MeA^Foxp2 cells are generally located more laterally than MeA^Dbx1 cells as reflected by their longer distances to the optic tract at multiple bregma levels (Extended Data Fig. 1a–f). Along the dorsal-ventral axis of the MeApd, the MeA^Foxp2 cells are located more dorsally than the MeA^Dbx1 cells (Extended Data Fig. 1g–l). Notably, consistent with our previous study, MeA^Dbx1, and MeA^Foxp2 are largely distinct, even when they occupy the same MeA region (Fig. 1c–e). Of all MeA^Foxp2 and MeA^Dbx1 cells, only 1.8% are double positive.

MeA^Dbx1 and MeA^Foxp2 cells in male mice are predominantly GABAergic

Our previous study showed that both MeAp Foxp2 and Dbx1-derived cells are enriched with markers for inhibitory neurons—for example, calbindin and nNOS^9,23. To determine the neurotransmitter type of MeA^Foxp2 cells more directly, we immunostained Foxp2 in Vgat^cre;Ai6^+/− and Vglut2^cre;Ai6^+/− male mice. We observed that 76% of the total MeA^Foxp2 subpopulation is GABAergic, whereas only 5% is glutamatergic (Fig. 1f,g). Within MeAp subdivisions, there is a significantly higher percentage of Foxp2⁺Vgat⁺ double-labeled cells in the MeApd (84%) than in the MeApv (66%) and a higher percentage of MeA Foxp2⁺Vglut2⁺ cells in the MeApv (12%) than in the MeApd (2%) (Fig. 1f,h,i). We next performed in situ hybridization for Vglut2⁺ and Vgat⁺ mRNA in Dbx1^cre;Ai6^+/− male mice and found that the MeA^Dbx1 subpopulation is also primarily GABAergic. Eighty-two percent of Dbx1-derived cells are Vgat⁺, and only 12% are Vglut2⁺ (Fig. 1j,k). Within MeA subdivisions, 87% and 7% of MeApd Dbx1-derived cells are Vgat⁺ and Vglut2⁺, respectively, whereas 66% and 27% of MeApv Dbx1-derived cells express Vgat and Vglut2, respectively (Fig. 1j,l,m). We note that Vgat⁺ and Vglut2⁺ MeA^Dbx1 (or MeA^Foxp2) cells do not add up to 100%. This could be because (1) some MeA^Dbx1 or MeA^Foxp2 cells are non-neuronal—previous work suggested that approximately 80% of the total MeA^Foxp2 cells express NeuN, a neuronal marker⁹; and (2) some Vgat⁺ or Vglut2⁺ cells might be missed due to their low mRNA or protein levels. Overall, these data suggest that both MeA^Dbx1 and MeA^Foxp2 are primarily GABAergic.

Distinct MeA^Foxp2 and MeA^Dbx1 cell responses to social sensory cues

To address whether MeA^Foxp2 and MeA^Dbx1 are hardwired to respond to different social cues, we recorded the Ca²⁺ activity of each population in naive adult male mice using fiber photometry while presenting various social stimuli in a pseudo-random order (Fig. 2a). Naive mice are animals without any social interaction with other conspecifics, except with their dams and littermates. To ensure that any response difference is not due to behavior differences toward different social stimuli, we head-fixed the recording animals and presented anesthetized social stimuli along a linear track so that the onset, offset and duration of the stimulus presentation were precisely controlled (Fig. 2a). To record MeA^Foxp2 cells, we injected a GCaMP6f virus into the MeA of Foxp2^cre male mice²⁶ (Foxp2^GCaMP). To record MeA^Dbx1 cells, we generated Dbx1^cre;LSL-FlpO mice. In these animals, the transient Cre expression during embryogenesis, when Dbx1 is expressed, drives permanent Flp expression, allowing targeting of Dbx1-derived cells in adult mice²⁷. We then injected a GCaMP6f virus into the MeA of Dbx1^cre;LSL-FlpO male mice (Dbx1^GCaMP) (Fig. 2b). Histological analysis revealed that 88% of GCaMP6f cells express Foxp2 in Foxp2^GCaMP mice, whereas only 5% of GCaMP6f cells were co-labeled with Foxp2 in Dbx1^GCaMP mice, confirming the specificity of the recorded populations (Fig. 2c,d).

**Fig. 2: Distinct responses to social cues of MeA^Foxp2 and MeA^Dbx1 cells in head-fixed naive mice.**

MeA^Foxp2 cells in naive male mice showed robust GCaMP6 increases only during the presentation of adult males but not adult females, pups or objects (Fig. 2e,g,i,k). In contrast, MeA^Dbx1 cells showed the highest activity increase during the adult female presentation (Fig. 2f,h,j,k). MeA^Foxp2 and MeA^Dbx1 cells also differed in their response dynamics. MeA^Foxp2 cells returned to the baseline activity slowly (>10 s) after the removal of the male stimulus, whereas the MeA^Dbx1 cell activity returned to the baseline quickly (<3 s) (Fig. 2g,h). Overall, MeA^Foxp2 cells showed male-specific and slow decaying responses, whereas MeA^Dbx1 cells showed broad responses to social cues (Fig. 2g–k). These results strongly support distinct response patterns of MeA^Foxp2 and MeA^Dbx1 cells to social stimuli independent of fighting or mating experience.

Distinct MeA^Dbx1 and MeA^Foxp2 cell responses during social behaviors

Next, we examined responses of male MeA^Foxp2 and MeA^Dbx1 cells during social behaviors in freely moving male mice to address whether the cells increase activity only to sensory cues—for example, during investigation—or also during the action phase of the behavior—for example, attack and mount (Fig. 3a). Before recording, all test animals went through up to 12 interactions with an adult male and a female (once per day) to ensure behavior stability. During recording, a non-aggressive adult BALB/c male intruder, a sexually receptive female, a pup and a novel object were introduced into the home cage of the recorded mice, one at a time, with 5 min in between (Fig. 3b). MeA^Foxp2 cells significantly increased activity upon introducing a male, more than the responses during the introduction of any other social and non-social stimuli, and the activity remained elevated when the male was present (Fig. 3c–h). During each episode of male investigation and attack, MeA^Foxp2 cells also showed a significant activity increase (Fig. 3h). To address whether the activity increase during attack simply reflects the elevated activity during investigation, we separated investigation trials based on whether they were followed by attack or not. We found that MeA^Fxop2 cell activity at the offset of investigation is higher in investigation-followed-by-attack trials than investigation-not-followed-by-attack trials, suggesting that the elevated activity during attack is not simply due to sensory inputs during the preceding investigation (Extended Data Fig. 2a). In contrast to the strong response during male interaction, MeA^Foxp2 cells showed either no change or slightly suppressed activity during female investigation and all phases of sexual behaviors (Fig. 3d,h). Similarly, no activity change was observed during pup interaction, supporting a highly adult male-specific response of MeA^Foxp2 cells (Fig. 3e,h). Notably, the adult male-specific response of MeA^Foxp2 cells is not limited to one particular strain. We found that BALB/c and C57BL/6 males evoked similarly strong responses, significantly higher than those evoked by C57BL/6 or 129S4/SvJae females (Extended Data Fig. 3a–f).

**Fig. 3: Differential response patterns of MeA^Foxp2 and MeA^Dbx1 cells during fighting and mating in socially experienced male mice.**

In contrast to the response pattern of MeA^Foxp2 cells, MeA^Dbx1 cells in experienced male mice showed activity increases to all social stimuli (Fig. 3i–k). Upon initial intruder introduction, MeA^Dbx1 cells increased activity regardless of the identity of the intruder, and the responses to adult females and pups were significantly higher than those to objects (Fig. 3m). During investigation, MeA^Dbx1 cell activity also increased to a similar extent to all social targets (Fig. 3n). Although MeA^Dbx1 cells showed significant activity increase during inter-male attack, we did not find the responses during investigation-followed-by-attack trials and those during investigation-only trials to differ, suggesting that MeA^Dbx1 cell activity during attack could be largely due to activity increases during investigation preceding attack (Fig. 3i,n and Extended Data Fig. 2b). During copulation, the activity of MeA^Dbx1 cells did not increase during mounting but slightly increased during intromission (Fig. 3j,n). During ejaculation, MeA^Dbx1 cells increased activity robustly, significantly higher than the responses during any other behaviors (Fig. 3j,n). No activity increase of MeA^Dbx1 cells was observed when males attacked pups (Fig. 3k,n). MeA^Dbx1 cells did not respond during object investigation, whereas MeA^Foxp2 cells showed a slight suppression in activity during object investigation (Fig. 3f,h,l,n), supporting the social-specific response patterns of the cells.

The sexual and aggressive behaviors of MeA^Foxp2 and MeA^Dbx1 male mice are largely similar, except that MeA^Dbx1 attacked more frequently (Extended Data Fig. 4). Thus, changes in neuronal activity observed during recordings are unlikely due to behavioral differences. Overall, male MeA^Foxp2 cells show highly specific responses during both the investigatory and action phases of behaviors toward a conspecific adult male, whereas MeA^Dbx1 cells respond to diverse social cues and during ejaculation.

Refinement of MeA^Foxp2 cell responses with adult social experience

In some Foxp2^GCaMP animals, we performed Ca²⁺ recordings during freely moving social interactions before repeated social experiences. Two of 14 naive male Foxp2^GCaMP mice briefly attacked a male intruder in the 10-min testing period, and others only investigated the intruders (Extended Data Fig. 4a,b). Similar to our recordings in head-fixed naive animals, MeA^Foxp2 cells responded specifically during male investigation (Fig. 4a–f). We then directly compared MeA^Foxp2 cell responses between naive and experienced animals. The experienced animals were divided into two groups: mice that attacked a male intruder in the recording session (experienced aggressors) and those that did not (experienced non-aggressors).

**Fig. 4: Comparison of MeA^Foxp2 cell responses in naive versus non-aggressive and aggressive experienced male mice.**

The activity of MeA^Foxp2 cells in experienced aggressors increased faster and with higher consistency during male investigation than in naive animals (Fig. 4g–j). MeA^Foxp2 cells responded (Z_increase > 1 during investigation) in approximately 32% of trials in naive animals. In comparison, this number increased to 55% in experienced aggressors (Fig. 4k). Among the responsive trials, the average latency to respond in experienced aggressors is approximately half of that in naive animals (Fig. 4l). The mean activity increase during male investigation is significantly higher in experienced aggressors than in naive animals (Fig. 4m). The MeA^Foxp2 cell responses in experienced non-aggressors generally fell in between those in naive and experienced aggressors. Nevertheless, the male preference index (PI) did not differ among these three groups (Fig. 4n). Furthermore, the average duration per investigation episode was similar across the groups (Fig. 4o). These results suggest that, although adult aggressive experience is not required for the male-specific responses of MeA^Foxp2 cells, it refines the response by improving its trial-to-trial consistency and temporal precision.

The male-specific response of MeA^Foxp2 cells exists before puberty

To further address whether the male-specific MeA^Foxp2 cell responses are developmentally hardwired or established through adult experience, we recorded the responses of MeA^Foxp2 cells to social stimuli during early life. Puberty (P30–P38) is a critical development period when aggression emerges^28,29,30. Thus, we focused on MeA^Foxp2 cell responses before puberty (P25), at the onset of puberty (P30–P32) and after puberty (P40–P44). To achieve this goal, we injected Cre-dependent GCaMP6f virus into the MeA of P11 Foxp2^cre mice and placed a 400-µm fiber just dorsal to the MeA at P24 and allowed for 24-h recovery before recording at P25 (Fig. 5a–c). Behaviorally, recorded juvenile Foxp2^cre mice (P25) spent a similar amount of time investigating the intruders as age-matched Foxp2^cre mice that had not undergone surgery, suggesting sufficient recovery (Extended Data Fig. 5a). Histological analysis confirmed high levels of GCaMP6 expression at P25 (Extended Data Fig. 5b).

**Fig. 5: MeA^Foxp2 cell responses before, during and after puberty in developing male mice.**

We recorded the Ca²⁺ activity of MeA^Foxp2 cells when the animals were exposed to an anesthetized adult male and adult female mouse or a pup at P25, P30–32 and P40–P44 (Fig. 5d). To minimize the impact of social experience, all animals were singly housed after weaning at P21. We found that MeA^Foxp2 cells in P25 juvenile male mice already showed higher activity during close interaction with an adult male than an adult female or pups (Fig. 5e). However, when we consider the average activity of the entire recording session, the presence of either an adult male or female, but not a pup, caused an elevation in GCaMP6 activity (Fig. 5e5). At P30–P32, a similar male-biased response was observed during close interaction, whereas the overall GCaMP6 activity during the recording session was not significantly elevated regardless of the intruder (Fig. 5f5). At P40–P44, the difference between male and female responses during investigation increased (Fig. 5g), and this trend continued at >P56 (Fig. 5h). At >P56, the presence of a male caused a significant elevation in the GCaMP6 activity, whereas female and pup presence caused either no change or slightly suppressed activity (Fig. 5h,i). As a result of gradually decreased activity to non-male social cues, the MeA^Foxp2 cells become increasingly tuned to adult males over development (Fig. 5j), although male-biased responses are seen at all ages (Fig. 5k). Altogether, these results suggest that MeA^Foxp2 cells are predisposed to preferentially responding to male-related sensory information even before puberty, and the discriminability between adult male and non-male cues is further refined after puberty by reducing responses to non-adult male cues.

Differential inputs to MeA^Foxp2 and MeA^Dbx1 cells

Given the differential responses of MeA^Foxp2 and MeA^Dbx1 cells to social cues, we next asked whether these two populations receive inputs from different brain regions by performing monosynaptic rabies virus tracing. We injected Cre-dependent or Flp-dependent adeno-associated viruses (AAVs) expressing TVA-mCherry and rabies G protein into the MeA of Foxp2^cre or Dbx1^cre;LSL-FlpO male mice and, 4 weeks later, EnvA-ΔG rabies virus expressing GFP (Fig. 6a–d). We found that the major inputs to MeA^Foxp2 are from other amygdala nuclei, including the posterior amygdala (PA), central amygdala (CeA) and BNST (Fig. 6e–g). In contrast, MeA^Dbx1 cells receive inputs mainly from primary olfactory relays, including AOB, cortical amygdala (CoA) and the piriform cortex (Pir) (Fig. 6e,h,i). Hypothalamus, mainly the medial preoptic area (MPOA) and zona incerta (ZI), provided moderate inputs to both MeA^Foxp2 and MeA^Dbx1 cells (Fig. 6e–i). Sparsely retrogradely labeled cells from both MeA^Foxp2 and MeA^Dbx1 cells were also observed in the hippocampus, striatum and pallidum (Fig. 6e–i).

**Fig. 6: Differences in the anatomical and functional inputs of MeA^Foxp2 and MeA^Dbx1 cells for sensory processing.**

The lack of retrogradely labeled cells in the AOB from MeA^Foxp2 starter cells was particularly surprising given that the MeA is the primary target of the AOB (Fig. 6e–g)^4,31,32. To further dissect the inputs from the AOB to MeA^Foxp2 cells, we performed optogenetic-assisted circuit mapping. We expressed ChrimsonR-tdTomato in the olfactory bulb, virally labeled MeA^Foxp2 cells with GFP (Fig. 6j,k) and visualized MeA^Dbx1 cells using Dbx1^cre;Ai6 mice (Fig. 6l,m). Four weeks after injection, we prepared brain slices containing the MeA and recorded the responses of GFP⁺ MeA^Foxp2 and MeA^Dbx1 cells to 605-nm, 1-ms light pulses. Among 23 recorded MeA^Foxp2 cells, we observed light-evoked excitatory postsynaptic currents (oEPSCs) in only two cells, and 18 of 23 recorded cells showed light-evoked inhibitory postsynaptic currents (oIPSCs) (Fig. 6n,o). In contrast, 18 of 33 MeA^Dbx1 cells showed oEPSCs, and 29 of 33 showed oIPSCs (Fig. 6t,u). The oIPSCs of MeA^Dbx1 and MeA^Foxp2 cells were similar in magnitude, and both were of long latencies (>10 ms) (Fig. 6p–s). Bath application of TTX or TTX + 4-AP completely abolished oIPSCS in both populations, suggesting that they are polysynaptic connections (Fig. 6o,u–w). oEPSCs of MeA^Dbx1 cells were of short latency (~4 ms) (Fig. 6s), and their amplitude did not change after TTX + 4-AP application, supporting that AOB cells provide monosynaptic excitatory inputs to MeA^Dbx1 cells (Fig. 6u,x).

These results confirmed that AOB targets MeA^Foxp2 and MeA^Dbx1 cells differently. The fact that MeA^Foxp2 cells receive minimum direct inputs from the AOB and other primary olfactory relays suggests that sensory information reaching MeA^Foxp2 cells is likely more processed, which may explain the higher response selectivity of MeA^Foxp2 cells than MeA^Dbx1 cells.

Activating MeA^Foxp2 cells is sufficient for aggression in naive mice

To understand the functional importance of MeA^Foxp2 and MeA^Dbx1 cells in social behaviors, we bilaterally injected Cre-dependent and Flp-dependent hM3Dq viruses into the MeA of Foxp2^cre and Dbx1^cre;LSL-FlpO naive male mice, respectively (Foxp2^hM3Dq and Dbx1^hM3Dq) (Fig. 7a,b). Control animals were injected with mCherry virus in the MeA (Foxp2^mCherry and Dbx1^mCherry). Three weeks later, we intraperitoneally (i.p.) injected saline and clozapine-N-oxide (CNO) on two consecutive days and, 30 min later, introduced a pup, an adult male and a female intruder into the cage sequentially, each for 5–10 min with 5 min in between (Fig. 7c). To determine whether MeA^Foxp2 activation could result in increases in aggression in mice that are not spontaneously aggressive, we first tested animals’ baseline aggression level after saline injection on day 1, followed by CNO injection on day 2. Although only four of 10 Foxp2^hM3Dq male mice attacked a male intruder after saline injection, all Foxp2^hM3Dq males attacked the intruder after CNO injection (Fig. 7d,e). In comparison, only four of eight control Foxp2^mCherry mice initiated attack after CNO injection (Fig. 7d,e). The total attack time of Foxp2^hM3Dq males significantly increased after CNO injection (Fig. 7f), although the latency to attack did not decrease in animals that attacked on both days (Extended Data Fig. 6a). Possibly due to increased aggression, Foxp2^hM3Dq mice spent less time investigating the male intruder after CNO injection (Fig. 7g). For most Foxp2^hM3Dq animals (9/10), we tested the aggression level again on day 3 after saline injection and observed a significantly lower total attack duration in comparison to that of day 2 after CNO injection, further suggesting that the longer attack duration after Foxp2^hM3Dq activation is not simply due to a natural increase in aggression with repeated resident–intruder (R–I) tests (Extended Data Fig. 6b). Furthermore, CNO-induced attack was not due to an increase in general arousal, as locomotion did not differ between saline-injected and CNO-injected days (Extended Data Fig. 6c). Notably, the increased aggression is adult male specific, as we did not observe an increase in infanticide after activating MEA^Foxp2 cells (Extended Data Fig. 6d). The overall pup interaction was also unchanged (Extended Data Fig. 6e). Similarly, male sexual behaviors, including female investigation, mounting and intromission, were not affected by MEA^Foxp2 activation (Extended Data Fig. 6f–l). Control Foxp2^mCherry animals showed no significant change in any social behavior after CNO injection compared to saline injection (Fig. 7d–g and Extended Data Fig. 6d–l).

**Fig. 7: MeA^Foxp2 cells bi-directionally modulate territorial aggression, whereas MeA^Dbx1 cells do not.**

We noticed that Dbx1^cre;LSL-FlpO male mice tend to be more aggressive than Foxp2^cre male mice, possibly due to their difference in genetic background (Extended Data Fig. 7)^26,27. Twelve of 18 Dbx1^cre;LSL-FlpO animals (combining mCherry and hM3Dq groups) attacked the intruder during the first encounter (after saline injection), whereas only six of 18 Foxp2^cre animals did so (Extended Data Fig. 7b). Notably, there was no difference between Dbx1^hM3Dq and Dbx1^mCherry groups in the percentage of animals that attacked (Fig. 7i). The latency to attack and attack duration also did not differ on CNO-injected and saline-injected days in both Dbx1^hM3Dq and Dbx1^mCherry groups (Fig. 7j and Extended Data Fig. 8a), although Dbx1^hM3Dq male mice investigated the male intruder less after CNO than saline injection (Fig. 7k). Activating MeA^Dbx1 cells did not change the probability of infanticide, male sexual behaviors or locomotion significantly (Extended Data Fig. 8b–k). Thus, MEA^Foxp2 cells can specifically drive inter-male aggression in even non-aggressive naive male mice, whereas activating MEA^Dbx1 cells does not promote any specific social behaviors to a significant level.

Inhibiting MeA^Foxp2 cells reduces aggression in experienced animals

We next asked whether MeA^Foxp2 and MeA^Dbx1 cells are necessary for social behaviors, including inter-male aggression. We injected Cre-dependent and Flp-dependent hM4Di-mCherry into the MeA of Foxp2^cre and Dbx1^cre;LSL-FlpO male mice, respectively (Foxp2^hM4Di and Dbx1^hM4Di). Control animals were injected with mCherry virus (Fig. 7l,m). Three weeks after viral injection, all animals went through repeated R–I tests until they showed stable aggression (Fig. 7n). Then, we i.p. injected saline and CNO on separate days in a randomized order and, 30 min later, tested the behaviors against a male and a receptive female intruder, each for 10 min (Fig. 7n). After CNO injection, Foxp2^hM4Di mice spent more time investigating and less time attacking the male intruder (Fig. 7o–q). The latency to attack increased significantly (Fig. 7r). Foxp2^mCherry mice showed no difference in male investigation or attack duration between CNO-injected and saline-injected days (Fig. 7o–r). In contrast, CNO injection in Dbx1^hM4Di mice did not result in significant changes in male investigation, aggressive behaviors or latency to attack (Fig. 7s–v). CNO injection in Foxp2^hM4Di or Dbx1^hM4Di mice caused no change in female investigation or any aspects of male sexual behaviors except an increase in mount number in both Dbx1^hM4Di and Dbx1^mcherry groups (Extended Data Figs. 7m–s and 8l–r). These results suggest that MeA^Foxp2 cells specifically modulate inter-male aggression, whereas MeA^Dbx1 cells do not.

Differential outputs of MeA^Dbx1 and MeA^Foxp2 cells

As MeA^Dbx1 and MeA^Foxp2 cells play distinct roles in driving social behaviors, presumably through their differential connections on downstream cells, we next asked whether these two MeA subpopulations differ in their projections using anterograde virus tracing (Fig. 8a–d). We observed that both MeA subpopulations project mainly to other extended amygdala areas, such as PA, CoA, posterior BNST (BNSTp) and medial hypothalamus (MH) (Fig. 8e–i and Extended Data Fig. 9). The average density of projections originating from MeA^Dbx1 and MeA^Foxp2 did not differ significantly in any specific amygdala or hypothalamic nucleus (Fig. 8i). Nevertheless, we observed some differences in projection patterns. The MeA^Dbx1 cells generally provide more inputs to AVPV and MPOA subnuclei, structures related to sexual behaviors^33,34, than the VMHvl and PMv, structures related to aggression³. The overall fiber density from MeA^Dbx1 cells in MPOA + AVPV is approximately twice that in VMHvl + PMv (Fig. 8i, j). In contrast, the density of the fibers originating from MeA^Foxp2 cells is similar in the sexual-behavior-related and aggression-related medial hypothalamic regions (Fig. 8i,j). Furthermore, a close analysis of the BNST revealed differential patterns of projections from MeA^Dbx1 and MeA^Foxp2 cells. Whereas MeA^Dbx1 cells target the principal nucleus of the BNST (BNSTpr) primarily, MeA^Foxp2 cells project more densely to the interfascicular part of the BNST (BNSTif) (Fig. 8k).

**Fig. 8: Outputs of MeA^Foxp2 and MeA^Dbx1 cells.**

Discussion

In this study, we showed that two MeA subpopulations, marked by the expression of different transcription factors from early development, play distinct roles in social behaviors, show differential input and output patterns and are responsive to different conspecific sensory cues. The male-specific responses of MeA^Foxp2 cells exist before puberty and aggression onset, suggesting that these responses are largely developmentally hardwired. The reliability and temporal precision of MeA^Foxp2 cell responses improve with adult social experience, demonstrating distinct roles of nature versus nurture in setting up the response patterns of this population.

MeA^Foxp2 and MeA^Dbx1 cell activity and function in social behaviors

Our previous work identified two developmentally distinct GABAergic MeA subpopulations marked by the expression of Dbx1 and Foxp2 (refs. ^9,23). They differ in sex steroid hormone receptor expression, ion channel composition and intrinsic electrophysiological properties^9,35. Our current study further revealed their distinct functions in social behaviors that are well matched with their differential connectivity and in vivo response patterns. These results suggest that social circuits at the MeA could be largely hardwired according to transcription-factor-defined genetic programs.

MeA^Foxp2 cells responded strongly during both male investigation and attack. Functionally, chemogenetic activation of MeA^Foxp2 cells promoted attack even in non-aggressive male mice. Anatomically, MeA^Foxp2 cells receive very little direct input from the AOB. Thus, despite the long-recognized role of the MeA in pheromone processing^6,7,10,36,37, MeA^Foxp2 cells appear to be more involved in facilitating aggressive actions. However, our population fiber photometry recording does not provide single-cell resolution. Therefore, it remains possible that distinct subsets of MeA^Foxp2 cells process olfactory cues and mediate aggressive actions.

In contrast to MeA^Foxp2 cells, MeA^Dbx1 cells are tuned to broad social cues, including those from males, females and pups and respond robustly during ejaculation but minimally during attack or other copulatory behaviors. Consistent with the lack of activity increase during attack, inactivation of MeA^Dbx1 cells does not impair male aggression, and chemogenetic activation of MeA^Dbx1 cells does not promote attack. Given that MeA^Dbx1 cells are three times more abundant than MeA^Foxp2 cells and more excitable⁹, yet fail to promote attack, we conclude that aggression generation requires activation of a specific transcriptionally defined subpopulation instead of any subset of MeA GABAergic cells.

Given the response pattern of MeA^Dbx1 cells, we considered their primary role in processing social cues during the investigatory phase. However, animals with inactivated MeA^Dbx1 cells properly directed their attack toward males and mount toward females, suggesting that MeA^Dbx1 cells are dispensable for sex discrimination. This negative result is possibly due to the existence of other extended amygdala populations that can readily distinguish male and female cues³⁸—for example, MeA^Foxp2 and aromatase cells in BNSTpr³⁹. Interestingly, we found that MeA^Dbx1 cells robustly increase activity during male ejaculation. Recent works found that Esr2⁺ cells in the BNST are also highly activated during ejaculation and functionally important for diminished sexual motivation after ejaculation^38,40. Here, we observed a tendency of decreased mounting and intromission duration after activating MeA^Dbx1 cells (Extended Data Fig. 6p,r), hinting at the possible role of MeA^Dbx1 cells in suppressing sexual motivation as BNST^Esr2 cells. This hypothesis, however, requires further investigation in future studies.

Previous work showed that MeA GABAergic cells are activated during pup-directed attacks and can promote infanticide²². However, neither MeA^Foxp2 nor MeA^Dbx1 cells increased activity during pup-directed aggression or affected infanticide when artificially activated, suggesting that MeA^Foxp2 cells are specialized for adult-directed aggression. Therefore, other GABAergic subclasses likely exist for driving infanticide and remain to be identified.

Developmentally wired versus experientially wired

There is an ongoing debate regarding whether the responses of cells in the SBN are developmentally hardwired or established through adult social experience. In the VMHvl, an essential region for male aggression^41,42,43, individual cell responses to male and female cues overlap extensively in naive adult male mice and diverge only after repeated interaction with females⁴⁴. In contrast, aromatase-expressing cells in male BNSTpr preferentially respond to female cues over male cues, even in naive animals³⁹. Ca²⁺ imaging in the MeA revealed that approximately half of MeA cells are tuned to one stimulus in naive animals, and, after sexual experience, the proportion of cells that are responsive to the opposite sex increases, denoting experience-dependent activity refinement¹⁰. In our study, MeA^Foxp2 cells showed strong male-biased responses in naive animals, suggesting that male olfactory inputs are developmentally wired to target MeA^Foxp2 cells. However, the responses of MeA^Foxp2 cells in naive males are slow and inconsistent and only become fast and consistent after repeated social interactions and aggressive encounters, suggesting that adult social experience plays an important role in refining this hardwired circuit to improve its input (sensory cue)–output (spiking) transformation efficiency.

How is the male-specific response of MeA^Foxp2 cells achieved during development? The classical ‘organization/activation’ model states that gonadal hormones act in two phases to establish sex-specific circuits^45,46,47. First, during the organization stage, gonadal hormones during prenatal development set up the basic structure and connection of the circuit. Then, the circuits are activated by gonadal hormones during puberty to generate appropriate sex-specific social behaviors. In male mice, puberty occurs between P30 and P38 when testosterone spikes and aggression emerges^28,45. The fact that male-biased responses of MeA^Foxp2 cells exist before puberty suggests that male cues have been wired preferentially to MeA^Foxp2 cells during the organization stage. After puberty, MeA^Foxp2 cells show enhanced male-biased responses due to decreased responses to non-male social cues, suggesting further circuit refinement possible through local inhibition. Altogether, we propose that the response specificity of MeA^Foxp2 cells during development is achieved through a multi-stage process, including pre-pubertal hardwiring, pubertal refinement and adult social experience-dependent potentiation. Future microcircuit studies could help further validate this model and its generality in the SBN.

Social behavior circuits beyond MeA

In mice, olfactory inputs are the most essential for determining the identity of a conspecific—for example, its sex, age, social ranking and health state (for example, sickness)⁴⁸. Because MeA^Foxp2 cells receive little direct input from the AOB and other primary olfactory relays, we speculate that MeA^Foxp2 cells obtain highly ‘processed’ olfactory information from the PA. Recent works revealed that PA cells projecting to the VMHvl are crucial for territorial aggression, and these cells are activated during both male investigation and attack^49,50. The PA also projects strongly to MeA; however, whether this projection is essential for aggression remains to be explored. On the contrary, MeA^Dbx1 cells receive abundant inputs from AOB and other primary olfactory processing regions, which could be responsible for the broad and fast responses of MeA^Dbx1 cells to various social cues.

At the output level, MeA^Dbx1 and MeA^Foxp2 cells project to distinct pBNST subnuclei: MeA^Dbx1 primarily to the BNSTpr and MeA^Foxp2 cells mainly to the BNSTif. Miller et al.¹⁹ recently demonstrated that MeA cells that express D1R target the BNSTif primarily, and activating MeA^D1R-BNST projections increased territorial aggression. This supports the relevant role of BNSTif in aggression and a potential downstream mechanism by which MeA^Foxp2 cells mediate aggressive action. Additionally, MeA^Foxp2 cells project relatively more densely to the VMHvl and PMv than MeA^Dbx1 cells. Given that VMHvl and PMv are central for male aggression^42,43,51, the more robust projection of MeA^Foxp2 cells to these regions is consistent with the essential role of MeA^Foxp2 cells in male aggression.

Our study identified a developmentally hardwired circuit at the MeA to process male information essential for initiating aggression. We revealed the distinct contribution of development versus experience in social information processing and highlighted a lineage-based organization strategy that enables the same SBN to drive diverse social behaviors².

Methods

Mice

All animal procedures were approved by the Institutional Animal Care and Use Committee of NYU Langone Health under protocol IA16-01416. Adult and juvenile experimental and stimulus mice were housed under a 12-h light/dark cycle (10:00 to 22:00 dark) with water and food ad libitum. Holding and experimental room temperatures were maintained at 20–22 °C and humidity kept between 30% and 70% (average ~45%). After surgical procedures, all experimental animals were single housed. The Foxp2^cre mice were originally provided by Richard Palmiter (now Jackson Laboratory stock no. 030541)²⁶. The Dbx1^cre mice were originally provided by Alessandra Pierani and crossed to the Flp-excised and Cre-inducible LSL-FlpO mouse line or to the Ai6 mouse line (Jackson Laboratory stock no. 028584 and no. 007906, respectively)^24,25,27. Both Foxp2^cre and Dbx1^cre mice are black, whereas the fur color of LSL-FlpO mice is agouti. Stimulus animals were C57BL/6N and 129S4/SvJae group-housed females, pups (P1–P7) and group-housed BALB/c males purchased from Charles River Laboratories and bred in-house. Females were considered receptive if an experienced male could mount and intromit the female in at least three attempts. No statistical methods were used to pre-determine sample sizes, but our sample sizes are similar to those reported in previous publications^49,52,53.

Viruses and stereotaxic surgery

For fiber photometry experiments, we injected 100 nl of AAV2-CAG-Flex-GCaMP6f (2.21 × 10¹³ vector genomes per milliliter (vg/ml) or 1.82 × 10¹² vg/ml; UPenn Viral Core) unilaterally into the MeA (AP = −1.5 mm, ML = 2.15 mm, DV = −5.1 mm) of Foxp2^cre male mice. For Dbx1^cre;LSL-FlpO mice, we injected either 100 nl of AAV8-Ef1a-fDIO-GCaMP6f (1 × 10¹³ vg/ml; kindly provided by Naoshige Uchida) or 120 nl of mixed AAV9-Ef1a-fDIO-Cre (2.5 × 10¹³ vg/ml; Addgene) and AAV2-CAG-Flex-GCaMP6f (1:2; 2.21 × 10¹³ vg/ml; UPenn Viral Core) or 150 nl of AAV2-Ef1a-fDIO-GCaMP6f (4.1 × 10¹² vg/ml; UNC Vector Core) into the MeA. For fiber photometry recordings in Foxp2^cre juvenile mice, we injected 100 nl of AAV1-CAG-Flex-GCaMP6f (9.4 × 10¹² vg/ml; UPenn Viral Core) unilaterally into the developing MeA (AP = −0.7 mm, ML = 2.03 mm, DV = −5.05 mm). For chemogenetic experiments, we bilaterally injected 400–600 nl of AAV1-Ef1a-DIO-hM4D(Gi)-mCherry, 150 nl of AAV2-hSyn-DIO-hM3D(Gq)-mCherry or 150–600 nl of AAV2-hSyn-DIO-mCherry (3 × 10¹² vg/ml, 5.1 × 10¹² vg/ml and 5.6 × 10¹² vg/ml, respectively; Addgene and UNC Vector Core) into the MeA of Foxp2^cre mice. For chemogenetic experiments in Dbx1^cre;LSL-FlpO mice, we injected 300 nl of AAVDJ-hSyn-fDIO-hM4D(Gi)-mCherry, 50–60 nl of AAV2-Ef1a-fDIO-hM3D(Gq)-mCherry (Vigene) and 60–120 nl of AAV2-Ef1a-fDIO-mCherry (2.65 × 10¹³ vg/ml, 1.84 × 10¹³ vg/ml and 1.1 × 10¹³ vg/ml, respectively; Addgene). For monosynaptic retrograde rabies experiments in Foxp2^cre mice, we injected unilaterally into the MeA 200–250 nl of mixed AAV1-CA-Flex-RG and AAV8-Ef1-Flex-TVA-mCherry (1:1; 3 × 10¹² vg/ml and 5.4 × 10¹² vg/ml; UNC Vector Core) and, 4 weeks later, 800 nl of EnvA G-Deleted Rabies-eGFP (2.26 × 10⁸ and 1.07 × 10⁸ transforming units per milliliter (TU/ml); Salk Viral Vector Core). For monosynaptic retrograde rabies experiments in Dbx1^cre;LSL-FlpO mice, we injected mixed 110–120 nl of AAV8-Flex(FRT)-G and AAV8-Flex(FRT)-TVA-mCherry (1:1; 1.82 × 10¹³ vg/ml and 1.39 × 10¹³ vg/ml; Stanford Gene Vector and Viral Core) and, 4 weeks later, 800 nl of EnvA G-Deleted Rabies-eGFP (Salk Viral Core). We also unilaterally injected 80–100 nl of AAVDJ-Ef1a-fDIO-EYFP (2.1 × 10¹² vg/ml; UNC Vector Core) into the MeA of Dbx1^cre;LSL-FlpO mice for anterograde tracing experiments. For Chr2-assisted circuit mapping, we injected 150 nl of AAV2-Flex-GFP (3.7 × 10¹² vg/ml; UNC Vector Core) unilaterally into the MeA of Foxp2^cre mice and 40–200 nl of AAV9-hSyn-ChrimsonR-tdTomato (5.5 × 10¹² vg/ml; Addgene) unilaterally into the olfactory bulb (AP = 4.45 mm, ML = 0.25 mm, DV = −1.55 mm) of Foxp2^cre and Dbx1^cre;Ai6^+/− mice. EnvA G-deleted Rabies virus titers were 2.26 × 10⁸ and 1.07 × 10⁸ TU/ml.

During surgery, adult male mice were anesthetized with isoflurane (2%) and then placed in a stereotaxic apparatus (Kopf Instruments). For fiber photometry recordings in juvenile mice, P11 pups were anesthetized with isoflurane (2%) and placed in a stereotaxic apparatus modified with a neonatal anesthesia head holder and zygoma ear cups (Kopf Instruments). The virus was then delivered into the target region of interest through a glass capillary using a nanoinjector (World Precision Instruments). For fiber photometry experiments in adults, a 400-μm fiber optic cannulae with ceramic ferrule (Thorlabs, FT400EMT, CF440-10 or RWD, R-FOC-L400C-50NA) was placed 0.3 mm dorsal to the viral injection site and cemented with adhesive dental cement (C&B Metabond, S380). For juvenile experiments, juveniles at P24 were implanted with the optical fiber in the MeA (AP = −0.7 mm, ML = 2.03 mm, DV = −4.75 mm). Histology analysis was performed for all animals, and only animals with correct virus expression and fiber placement were used for the final analysis.

Behavioral assays and analysis

Behavior was recorded by two synchronized top and side cameras (Basler, acA640-100gm) at 25 frames per second and digital video recording software (StreamPix 5, NorPix) in a dark room with infrared lights. Behaviors were manually annotated on a frame-by-frame basis by using a custom MATLAB function called ‘BehaviorAnnotator’ (https://github.com/pdollar/toolbox).

For male–male interactions, we annotated investigation, groom and attack. For male–female interactions, we recorded investigation, mount, intromission and ejaculation. For male–pup interactions, we recorded investigation, groom and infanticide. For fiber photometry analysis, investigation and groom were combined as ‘Investigation’. ‘Investigation’ was considered nose contact with any body part of the target mouse. ‘Groom’ was classified as when a mouse has its front paws holding the back or face of the target mouse or pup and licking either face or back. ‘Attack’ was determined as a series of actions by which the male mouse lunges, bites, chases and pushes the target mouse. ‘Mount’ was defined as a series of fast movements by which the male mouse places its front paws on the target mouse and positions itself on top of the target mouse. ‘Intromission’ was annotated as deep rhythmic thrusts following the mount. ‘Ejaculation’ was considered when the male stops deep thrusting and freezes in place for several seconds while firmly holding the female mouse and then slumping to the side. ‘Infanticidal behavior’ was considered as biting the pup that results in tissue damage. For fiber photometry and chemogenetic analysis, pup investigation and groom were combined as ‘pup investigation’.

In this study, experiments were not performed in a blinded manner as the experimental conditions were clear to the experimenters and the analyses were carried out using a recording system not subject to human bias. During behavioral annotations, the experimenter was blinded to the GCaMP6 signal or to the behavioral response.

Fiber photometry

Foxp2^cre and Dbx1^cre;LSL-FlpO randomly selected male mice aged 2–8 months were used for adult fiber photometry recordings. Foxp2^cre male mice starting at age P25 were used for juvenile fiber photometry experiments. For adult head-fixed experiments, the mice were naive and did not have any interactions with other conspecifics outside of their littermates and dams. The recording mouse was head-fixed using a 3D-printed head ring and placed on a 3D-printed wheel⁵⁴ for a minimum of 3 d of training before testing. Each stimulus was presented five times for 10 s with a 40-s interval in between presentations and a minimum of 5-min break in between different stimuli. Male and receptive female stimulus mice were anesthetized with ketamine (100 mg kg⁻¹) and xylazine (10 mg kg⁻¹).

Fiber photometry was performed as previously described ^49,55,56. Fiber photometry signals were collected by using a custom Tucker-Davis Technologies (TDT) program, OpenEx. To analyze changes in Ca²⁺ activity, the MATLAB function ‘msbackadj’, with a moving window of 25% of the total recording time, was used to obtain the instantaneous baseline signal (F_baseline). The instantaneous ΔF/F was calculated as (F_raw − F_baseline) / F_baseline. The z-score of the ΔF/F (Fz) was obtained using the MATLAB function ‘zscore’ for the whole trace. The peri-event histograms (PETHs) were calculated by aligning the Fz of each trial to either the onset or offset of each behavior. In recordings of head-fixed naive male mice (Fig. 2), the Fz peak was calculated by obtaining the average of the maximum value during stimulus presentation. In recordings of freely moving animals, the introduction Fz peak was calculated by obtaining the maximum value during the first 100 s of stimulus introduction into the resident’s cage. The Fz mean was calculated by obtaining the average of the mean values during specific behaviors during the specified intruder presentation window. The male PI was calculated as (Z_{investigate male} − 0.5 × (Z_{investigate female} + Z_{investigate pup})) / (|Z_{investigate male}| + 0.5 × |Z_{investigate female} + Z_{investigate pup}|); the female PI was calculated as (Z_{investigate female} − 0.5 × (Z_{investigate male} + Z_{investigate pup})) / (|Z_{investigate female}| + 0.5 × |Z_{investigate male} + Z_{investigate pup}|); and the pup PI was calculated as (Z_{investigate pup} − 0.5 × (Z_{investigate male} + Z_{investigate female})) / (|Z_{investigate pup}| + 0.5 × |Z_{investigate female} + Z_{investigate male}|).

When recording from freely moving mice (Figs. 3–5), a receptive female and an adult male mouse were introduced into the cage for 10 min. A pup was placed in the cage for 5 min. The male intruder was placed in the cage for >10 min until the recording mice elicited more than six attacks, without exceeding 1 h in the cage. A receptive female was introduced until 5 min after the recording mouse ejaculated.

When comparing naive freely moving and experienced male mice responses, the ‘latency to respond’ was calculated as the time lapse from behavior onset to when the response reaches Z ≥ 1. The ‘% of trials respond’ was calculated as the percentage of trials that reached Z ≥ 1. ‘Investigation time per trial (s)’ was calculated as the average duration of all male investigation trials. Heat maps were constructed as F_Z − Fz at time 0 for each trial.

Chemogenetic-mediated activation and silencing

For chemogenetic activation experiments, randomly selected naive male mice with no prior social experience except with their dam and littermates were used. On day 1, male mice were i.p. injected with saline. Thirty minutes after injection, video recordings started. After a 5-min baseline period, a pup intruder was placed into the cage for 5 min, followed by a 10-min presentation of an adult male and a receptive female, with 5-min breaks between stimulus. On day 2, male mice were i.p. injected with 1 mg kg⁻¹ of CNO (Sigma-Aldrich, C0832), and stimuli presentation was repeated as on day 1. On day 3, Foxp2^cre male mice were i.p. injected with saline, and stimuli were introduced as on days 1 and 2.

For chemogenetic silencing experiments, experimental male mice were trained to attack by introducing an adult male mouse daily for 10–30 min per day until they could consistently attack within a 10-min period. Mice were then i.p. injected with saline or CNO (1 mg kg⁻¹) on interleaved days for one or two rounds. Thirty minutes after injection, behavioral recordings started, and, after a 5-min baseline period, an adult male or a receptive female was introduced into the cage for 10 min each, with a 5-min break.

Animal body tracking

The velocity (pixels per frame) of each animal after 30 min of saline or CNO i.p. injection was obtained during the first 5 min of the chemogenetic assay before the introduction of any stimulus. The location of each animal was tracked using the top-view camera recordings and analyzed using a custom-written MATLAB GUI and code (https://github.com/pdollar/toolbox)⁴².

Distribution of MeA^Foxp2 and MeA^Dbx1 cells across the medial-lateral and dorsal-ventral axis

MeApd^Foxp2 and MeApd^Dbx1 cells were counted using Adobe Photoshop. The optic tract was used to calculate the medial-lateral cell distribution, whereas the dorsal edge of the MeAp was used to calculate the dorsal-ventral cell distribution. The ‘point to distance’ MATLAB function was used to calculate the distance from a point (that is, MeA^Foxp2 or MeA^Dbx1 cells) to a line (that is, the optic tract or the dorsal edge of the MeApd). Wilcoxon matched-pairs rank tests were performed to compare the distance distributions of MeApd^Foxp2 and MeApd^Dbx1 cells.

Triple in situ hybridization RNAscope

For in situ hybridization, brains were perfused with 1× PBS and fresh frozen in dry ice. Brains were embedded in O.C.T. compound (Sakura, 4583) and cut in 20-µm sections using a cryostat (Leica, CM1950) and placed directly into slides (Fisherbrand Superfrost Plus microscope slides, Thermo Fisher Scientific, 22-037-246). Every sixth section containing the MeA (bregma −1.4 mm to −2.1 mm) was used for staining. Using the manufacturer’s protocol (Advanced Cell Diagnostics)⁵⁷, slides containing the sections were dehydrated with several steps of ethanol and digested with proteinase, followed by hybridization of the mixed target probes for GFP (538851-C2), Slc32a1 (319191) and Slc17a6 (319171-C3). Slides were then stained with DAPI and coverslipped. Sections were imaged using a confocal microscope (Zeiss, LSM 800).

Immunohistochemistry and imaging analysis

Mice were anesthetized and perfused with 1× PBS, followed by 20 ml of 4% paraformaldehyde (PFA). Brains were fixed in 4% PFA for 6–12 h at 4 °C, dehydrated in 15% sucrose overnight, embedded in O.C.T. compound (Sakura, 4583) and cut into 50-µm sections using a cryostat (Leica, CM1950). Every third section was used for immunohistochemistry performed as previously described⁹. In brief, free-floating sections were incubated with primary antibody and blocked in 10% normal donkey serum (Jackson ImmunoResearch, 017-000-121) overnight. Brain sections were washed, placed in secondary antibody and blocked for 4–16 h. Then, sections were washed, mounted (Thermo Fisher Scientific, 12-550-15) and coverslipped using Fluoromount mounting media with DAPI (Thermo Fisher Scientific, 00-4959-52). Primary antibodies used were rabbit anti-Foxp2 (1:500, Abcam, ab16046), rat anti-GFP (1:1,000, Nacalai, 04404-84) and rabbit anti-mCherry (1:1,000, TaKaRa, Living Colors DsRed Polyclonal Ab 632496). Secondary antisera used were donkey anti-rat Alexa Fluor 488 (1:300; Jackson ImmunoResearch, 712-545-150) and donkey anti-rabbit Cy3 (1:1,000, Jackson ImmunoResearch, 711-165-152). Sections were imaged using a slide scanner (Olympus, VS120) or a confocal microscope (Zeiss, LSM 800) and counted manually using Adobe Photoshop. Cells stained with DAPI were automatically counted using the ‘analyze particles’ feature in ImageJ software and manually corrected.

Monosynaptic retrograde rabies input mapping

To determine the inputs to MeA^Foxp2 and MeA^Dbx1 cells, we injected randomly selected adult male mice with Cre-dependent or Flp-dependent AAV-G and AAV-TVA-mCherry viruses and, 4 weeks later, with EnvA G-Deleted Rabies-eGFP. After 7 d, mice were perfused, and every one in three brain sections was collected (50-µm thickness sections). Starter cells were considered TVA-mCherry and Rabies-eGFP double positive. Upstream Rabies-eGFP cells were then counted using ImageJ software. Owing to proximity with the MeA starter cell location, the lateral hypothalamus (LH), anterior MeA and anterior amygdalar area (AAA) were excluded from the analysis. Brains with more than 70% of starter cells in the MeA were considered for further analysis. Regions with more than 2% of total inputs to MeA^Foxp2 and MeA^Dbx1 cells were included in Fig. 6.

Output axonal projection mapping

To determine the projection patterns of MeA^Foxp2 and MeA^Dbx1 cells, every one in three brain sections was collected (50-µm thickness sections) and analyzed in Adobe Photoshop as previously described⁵⁵. In brief, the average pixel intensity of each region of interest (I_raw) was obtained, and the background intensity from the contralarateral side (l_background) was substracted (I_signal) and normalized to the maximum l_signal across all brain regions (I_norm). The average I_norm was then calculated for all animals to obtain the average axonal projection intensity for each terminal field. Animals with more than 65% of starter cells in the MeA were considered for analysis. Regions with more than 0.2 normalized intensity were included in Fig. 8. VMHvl was shown in the analysis given its well-established role in territorial aggression^42,58,59. The LH and anterior MeA were excluded from the analysis owing to their proximity to the starter cells.

Brain slice electrophysiology

For AOB to MeA circuit mapping experiments, we injected AAV2-Flex-eGFP and AAV9-hSyn-ChrimsonR-tdTomato into the MeA and the AOB, respectively, of Foxp2^cre+/− male mice or AAV9-hSyn-ChrimsonR-tdTomato into the AOB of Dbx1^cre+/−Ai6^+/− male mice. Whole-cell patch-clamp recordings were performed on MeA slices from all mice.

Mice were anesthetized with isoflurane, and brains were removed and submerged in ice-cold cutting solution containing (in mM): 110 choline chloride, 25 NaHCO₃, 2.5 KCl, 7 MgCl₂, 0.5 CaCl₂, 1.25 NaH₂PO₄, 25 glucose, 11.6 ascorbic acid and 3.1 pyruvic acid. Coronal sections of 275 µm were cut on a Leica VT1200 S vibratome and incubated in artificial cerebral spinal fluid (ACSF) containing (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 1 MgCl₂, 2 CaCl₂ and 11 glucose at 34 °C for 30 min and then transferred to room temperature for cell recovery until the start of recording. Whole-cell voltage-clamp recordings were performed with micropipettes filled with intracellular solution containing (in mM): 135 CsMeSO₃, 10 HEPES, 1 EGTA, 3.3 QX-314 (chloride salt), 4 Mg-ATP, 0.3 Na-GTP and 8 sodium phosphocreatine (pH 7.3 adjusted with CsOH). Signals were recorded using MultiClamp 700B amplifier and digitized by DigiData1550B with sampling rate of 20 kHz (Molecular Devices). Data were analyzed using Clampfit (Molecular Devices) or MATLAB (Mathworks). To activate ChrimsonR-expressing axons, brief pulses of full-field illumination (pE-300 white; CoolLED, 605 nm, 1-ms duration, 10 repeats, with 6-s interval) were delivered onto the recorded cell. oESPSs and oIPSCs were recorded by holding the membrane potential of recorded neurons at −70 mV and 0 mV, respectively. ACSF, TTX (1 µM), TTX (1 µM) and 4-AP (100 µM) were sequentially used to test if optogenetically evoked responses are monosynaptic. All drugs were pre-applied for 5 min in the slice chamber before data acquisition. Latency was measured as the time difference when the current exceeded 1.5 folds of the standard deviation of baseline compared to the light onset.

Statistics

All statistical analyses were performed using MATLAB or GraphPad Prism (versions 8 and 9) software. Statistical analyses were two-tailed. Parametric tests, including paired and unpaired t-test and one-way ANOVA, were used if distributions passed the Kolmogorov–Smirnov normality test. Normality tests were not performed for one-way ANOVA with missing values, sample size ≤4 and two-way ANOVAs. If data were not normally distributed, non-parametric tests were used. One-sample t-test was performed to determine whether the group mean differs from a specific value. When multiple t-tests were performed, P values were adjusted using the original false discovery rate (FDR) method of the Benjamini–Hochberg process. For comparisons across more than two groups, one-way ANOVA or repeated-measures one-way ANOVA was performed for normally distributed data, followed by Tukey’s multiple comparison tests; Friedman or Kruskal–Wallis tests were used for non-normally distributed data, followed by a two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli multiple comparison tests. Two-way ANOVA was performed for differences between groups with two independent variables, followed by Sidak’s multiple comparison tests. All significant P values < 0.05 are indicated in the figures. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. For detailed statistical analysis, see the Source Data associated with each figure.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Raw values associated with each figure panel can be found in the Source Data files. Fiber photometry recording data, raw representative images and behavior annotations can be downloaded from https://doi.org/10.5281/zenodo.8357846 as of the publication date. Behavior videos, additional histology images and information required to re-analyze the data are available from the corresponding authors upon reasonable request. Source data are provided with this paper.

Code availability

Custom-written MATLAB code used in the study can be downloaded from https://doi.org/10.5281/zenodo.8357846.

References

Wei, D., Talwar, V. & Lin, D. Neural circuits of social behaviors: innate yet flexible. Neuron 109, 1600–1620 (2021).
CAS PubMed PubMed Central Google Scholar
Newman, S. W. The medial extended amygdala in male reproductive behavior. A node in the mammalian social behavior network. Ann. N Y Acad. Sci. 877, 242–257 (1999).
CAS PubMed Google Scholar
Lischinsky, J. E. & Lin, D. Neural mechanisms of aggression across species. Nat. Neurosci. 23, 1317–1328 (2020).
CAS PubMed Google Scholar
Mucignat-Caretta, C. The rodent accessory olfactory system. J. Comp. Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 196, 767–777 (2010).
PubMed Google Scholar
Keshavarzi, S., Power, J. M., Albers, E. H., Sullivan, R. K. & Sah, P. Dendritic organization of olfactory inputs to medial amygdala neurons. J. Neurosci. 35, 13020–13028 (2015).
CAS PubMed PubMed Central Google Scholar
Bergan, J. F., Ben-Shaul, Y. & Dulac, C. Sex-specific processing of social cues in the medial amygdala. eLife 3, e02743 (2014).
PubMed PubMed Central Google Scholar
Choi, G. B. et al. Lhx6 delineates a pathway mediating innate reproductive behaviors from the amygdala to the hypothalamus. Neuron 46, 647–660 (2005).
CAS PubMed Google Scholar
Meredith, M. & Westberry, J. M. Distinctive responses in the medial amygdala to same-species and different-species pheromones. J. Neurosci. 24, 5719–5725 (2004).
CAS PubMed PubMed Central Google Scholar
Lischinsky, J. E. et al. Embryonic transcription factor expression in mice predicts medial amygdala neuronal identity and sex-specific responses to innate behavioral cues. eLife 6, e21012 (2017).
Li, Y. et al. Neuronal representation of social information in the medial amygdala of awake behaving mice. Cell 171, 1176–1190 (2017).
CAS PubMed PubMed Central Google Scholar
Kondo, Y. Lesions of the medial amygdala produce severe impairment of copulatory behavior in sexually inexperienced male rats. Physiol. Behav. 51, 939–943 (1992).
CAS PubMed Google Scholar
Kemble, E. D., Blanchard, D. C., Blanchard, R. J. & Takushi, R. Taming in wild rats following medial amygdaloid lesions. Physiol. Behav. 32, 131–134 (1984).
CAS PubMed Google Scholar
Fleming, A. S., Vaccarino, F. & Luebke, C. Amygdaloid inhibition of maternal behavior in the nulliparous female rat. Physiol. Behav. 25, 731–743 (1980).
CAS PubMed Google Scholar
Numan, M., Numan, M. J. & English, J. B. Excitotoxic amino acid injections into the medial amygdala facilitate maternal behavior in virgin female rats. Horm. Behav. 27, 56–81 (1993).
CAS PubMed Google Scholar
Takahashi, L. K. & Gladstone, C. D. Medial amygdaloid lesions and the regulation of sociosexual behavioral patterns across the estrous cycle in female golden hamsters. Behav. Neurosci. 102, 268–275 (1988).
CAS PubMed Google Scholar
Hong, W., Kim, D. W. & Anderson, D. J. Antagonistic control of social versus repetitive self-grooming behaviors by separable amygdala neuronal subsets. Cell 158, 1348–1361 (2014).
CAS PubMed PubMed Central Google Scholar
Unger, E. K. et al. Medial amygdalar aromatase neurons regulate aggression in both sexes. Cell Rep. 10, 453–462 (2015).
CAS PubMed PubMed Central Google Scholar
Padilla, S. L. et al. Agouti-related peptide neural circuits mediate adaptive behaviors in the starved state. Nat. Neurosci. 19, 734–741 (2016).
CAS PubMed PubMed Central Google Scholar
Miller, S. M., Marcotulli, D., Shen, A. & Zweifel, L. S. Divergent medial amygdala projections regulate approach–avoidance conflict behavior. Nat. Neurosci. 22, 565–575 (2019).
CAS PubMed PubMed Central Google Scholar
Nordman, J. C. et al. Potentiation of divergent medial amygdala pathways drives experience-dependent aggression escalation. J. Neurosci. 40, 4858–4880 (2020).
CAS PubMed PubMed Central Google Scholar
Wu, Y. E. et al. Neural control of affiliative touch in prosocial interaction. Nature 599, 262–267 (2021).
CAS PubMed PubMed Central Google Scholar
Chen, P. B. et al. Sexually dimorphic control of parenting behavior by the medial amygdala. Cell 176, 1206–1221 (2019).
CAS PubMed PubMed Central Google Scholar
Hirata, T. et al. Identification of distinct telencephalic progenitor pools for neuronal diversity in the amygdala. Nat. Neurosci. 12, 141–149 (2009).
CAS PubMed PubMed Central Google Scholar
Bielle, F. et al. Multiple origins of Cajal–Retzius cells at the borders of the developing pallium. Nat. Neurosci. 8, 1002–1012 (2005).
CAS PubMed Google Scholar
Madisen, L. et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci 13, 133–140 (2010).
CAS PubMed Google Scholar
Rousso, D. L. et al. Two pairs of ON and OFF retinal ganglion cells are defined by intersectional patterns of transcription factor expression. Cell Rep. 15, 1930–1944 (2016).
CAS PubMed PubMed Central Google Scholar
He, M. et al. Strategies and tools for combinatorial targeting of GABAergic neurons in mouse cerebral cortex. Neuron 91, 1228–1243 (2016).
CAS PubMed PubMed Central Google Scholar
Bell, M. R. Comparing postnatal development of gonadal hormones and associated social behaviors in rats, mice, and humans. Endocrinology 159, 2596–2613 (2018).
CAS PubMed PubMed Central Google Scholar
Schulz, K. M., Molenda-Figueira, H. A. & Sisk, C. L. Back to the future: the organizational–activational hypothesis adapted to puberty and adolescence. Horm. Behav. 55, 597–604 (2009).
CAS PubMed PubMed Central Google Scholar
Barkley, M. S. & Goldman, B. D. A quantitative study of serum testosterone, sex accessory organ growth, and the development of intermale aggression in the mouse. Horm. Behav. 8, 208–218 (1977).
CAS PubMed Google Scholar
Scalia, F. & Winans, S. S. The differential projections of the olfactory bulb and accessory olfactory bulb in mammals. J. Comp. Neurol. 161, 31–55 (1975).
CAS PubMed Google Scholar
Mohedano-Moriano, A. et al. Segregated pathways to the vomeronasal amygdala: differential projections from the anterior and posterior divisions of the accessory olfactory bulb. Eur. J. Neurosci. 25, 2065–2080 (2007).
PubMed Google Scholar
Dominguez, J. M. & Hull, E. M. Dopamine, the medial preoptic area, and male sexual behavior. Physiol. Behav. 86, 356–368 (2005).
CAS PubMed Google Scholar
Jennings, K. J. & de Lecea, L. Neural and hormonal control of sexual behavior. Endocrinology 161, bqaa150 (2020).
Matos, H. Y. et al. Sex differences in biophysical signatures across molecularly defined medial amygdala neuronal subpopulations. eNeuro 7, ENEURO.0035-20.2020 (2020).
Carvalho, V. M. et al. Lack of spatial segregation in the representation of pheromones and kairomones in the mouse medial amygdala. Front. Neurosci. 9, 283 (2015).
PubMed PubMed Central Google Scholar
Aggarwal, S. et al. Medial amygdala Kiss1 neurons mediate female pheromone stimulation of luteinizing hormone in male mice. Neuroendocrinology 108, 172–189 (2019).
CAS PubMed Google Scholar
Guo, Z. et al. Neural dynamics in the limbic system during male social behaviors. Neuron https://doi.org/10.1016/j.neuron.2023.07.011 (2023).
Article PubMed Google Scholar
Bayless, D. W. et al. Limbic neurons shape sex recognition and social behavior in sexually naive males. Cell 176, 1190–1205 (2019).
CAS PubMed PubMed Central Google Scholar
Zhou, X. et al. Hyperexcited limbic neurons represent sexual satiety and reduce mating motivation. Science 379, 820–825 (2023).
CAS PubMed Google Scholar
Hashikawa, Y., Hashikawa, K., Falkner, A. L. & Lin, D. Ventromedial hypothalamus and the generation of aggression. Front. Syst. Neurosci. 11, 94 (2017).
PubMed PubMed Central Google Scholar
Lin, D. et al. Functional identification of an aggression locus in the mouse hypothalamus. Nature 470, 221 (2011).
CAS PubMed PubMed Central Google Scholar
Yang, C. F. et al. Sexually dimorphic neurons in the ventromedial hypothalamus govern mating in both sexes and aggression in males. Cell 153, 896–909 (2013).
CAS PubMed PubMed Central Google Scholar
Remedios, R. et al. Social behaviour shapes hypothalamic neural ensemble representations of conspecific sex. Nature 550, 388–392 (2017).
CAS PubMed PubMed Central Google Scholar
Arnold, A. P. The organizational-activational hypothesis as the foundation for a unified theory of sexual differentiation of all mammalian tissues. Horm. Behav. 55, 570–578 (2009).
CAS PubMed PubMed Central Google Scholar
Phoenix, C. H., Goy, R. W., Gerall, A. A. & Young, W. C. Organizing action of prenatally administered testosterone propionate on the tissues mediating mating behavior in the female guinea pig. Endocrinology 65, 369–382 (1959).
CAS PubMed Google Scholar
McCarthy, M. M., Wright, C. L. & Schwarz, J. M. New tricks by an old dogma: mechanisms of the Organizational/Activational Hypothesis of steroid-mediated sexual differentiation of brain and behavior. Horm. Behav. 55, 655–665 (2009).
CAS PubMed PubMed Central Google Scholar
Dulac, C. & Torello, A. T. Molecular detection of pheromone signals in mammals: from genes to behaviour. Nat. Rev. Neurosci. 4, 551 (2003).
CAS PubMed Google Scholar
Yamaguchi, T. et al. Posterior amygdala regulates sexual and aggressive behaviors in male mice. Nat. Neurosci. 23, 1111–1124 (2020).
CAS PubMed PubMed Central Google Scholar
Zha, X. et al. VMHvl-projecting Vglut1⁺ neurons in the posterior amygdala gate territorial aggression. Cell Rep. 31, 107517 (2020).
CAS PubMed Google Scholar
Stagkourakis, S. et al. A neural network for intermale aggression to establish social hierarchy. Nat. Neurosci. 21, 834–842 (2018).
CAS PubMed Google Scholar
Hashikawa, K. et al. Esr1⁺ cells in the ventromedial hypothalamus control female aggression. Nat. Neurosci. 20, 1580–1590 (2017).
CAS PubMed PubMed Central Google Scholar
Yin, L. et al. VMHvll^Cckar cells dynamically control female sexual behaviors over the reproductive cycle. Neuron 110, 3000–3017 (2022).
CAS PubMed PubMed Central Google Scholar
Osborne, J. E. & Dudman, J. T. RIVETS: a mechanical system for in vivo and in vitro electrophysiology and imaging. PLoS ONE 9, e89007 (2014).
PubMed PubMed Central Google Scholar
Fang, Y.-Y., Yamaguchi, T., Song, S. C., Tritsch, N. X. & Lin, D. A hypothalamic midbrain pathway essential for driving maternal behaviors. Neuron 98, 192–207 (2018).
CAS PubMed PubMed Central Google Scholar
Feng, J. et al. A genetically encoded fluorescent sensor for rapid and specific in vivo detection of norepinephrine. Neuron 102, 745–761 (2019).
CAS PubMed PubMed Central Google Scholar
Wang, F. et al. RNAscope: a novel in situ RNA analysis platform for formalin-fixed, paraffin-embedded tissues. J. Mol. Diagn. 14, 22–29 (2012).
CAS PubMed PubMed Central Google Scholar
Yang, T. et al. Social control of hypothalamus-mediated male aggression. Neuron 95, 955–970. e954 (2017).
CAS PubMed PubMed Central Google Scholar
Lee, H. et al. Scalable control of mounting and attack by Esr1⁺ neurons in the ventromedial hypothalamus. Nature 509, 627–632 (2014).
CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank all members of the Lin laboratory and R. Sullivan, J. Dasen and M. Long for their inputs during the course of this study. We thank Y. Jiang and the Genotyping Core Laboratory of NYU Langone Health for genotyping of the mice for this study. We thank L. Mei for kindly providing some schematics that were included in this study. We also thank N. Uchida for kindly providing the AAV8-fDIO-GCaMP6f virus. We thank A. Pierani and L. Vigier for providing the Dbx1^cre mouse line and for primer sequences for genotyping. We thank R. Palmiter for providing the Foxp2^cre mice. This research was supported by a Leon Levy Neuroscience Fellowship and National Institute of Mental Health K99MH127295 (to J.E.L.); National Institutes of Health grants R01MH101377, 1R01HD092596 and U19NS107616 (to D.L.); the Mathers Foundation (to D.L.); Dean’s Undergraduate Research Funds (to J.B., G.S. and M.G); the Collegiate Research Initiative (to J.B.); R01DA020140 and R21MH129995; and the Caroline Fredricka Holdship Charitable Trusts through the PNC Charitable Trust Grant Review Commission (to J.G.C.).

Author information

Authors and Affiliations

Neuroscience Institute, New York University School of Medicine, New York, NY, USA
Julieta E. Lischinsky, Luping Yin, Chenxi Shi, Jared Burke, Govind Shekaran, Maria Grba & Dayu Lin
Hunter College, New York, NY, USA
Chenxi Shi
Center for Neuroscience Research, Children’s National Hospital, Washington, DC, USA
Nandkishore Prakash & Joshua G. Corbin
Center for Neural Science, New York University, New York, NY, USA
Jared Burke, Govind Shekaran, Maria Grba & Dayu Lin
Department of Psychiatry, New York University School of Medicine, New York, NY, USA
Dayu Lin

Authors

Julieta E. Lischinsky
View author publications
You can also search for this author in PubMed Google Scholar
Luping Yin
View author publications
You can also search for this author in PubMed Google Scholar
Chenxi Shi
View author publications
You can also search for this author in PubMed Google Scholar
Nandkishore Prakash
View author publications
You can also search for this author in PubMed Google Scholar
Jared Burke
View author publications
You can also search for this author in PubMed Google Scholar
Govind Shekaran
View author publications
You can also search for this author in PubMed Google Scholar
Maria Grba
View author publications
You can also search for this author in PubMed Google Scholar
Joshua G. Corbin
View author publications
You can also search for this author in PubMed Google Scholar
Dayu Lin
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

J.E.L., D.L. and J.G.C. conceived the project. J.E.L. and D.L. designed experiments, analyzed the data and co-wrote the manuscript. D.L. supervised the project. J.E.L. conducted most experiments. L.Y. performed in vitro electrophysiology experiments. C.S. assisted with chemogenetic and fiber photometry experiments. J.B., G.S. and M.G. assisted with histology and behavior annotation. N.P. worked on the preliminary characterization of axonal projections. J.G.C. provided feedback throughout the study and supervised N.P.

Corresponding authors

Correspondence to Julieta E. Lischinsky or Dayu Lin.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Neuroscience thanks the anonymous reviewers for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Characterization of MeA^Foxp2 and MeA^Dbx1 cell distribution in the MeApd, related to Fig. 1.

(a) Representation of a histological MeAp image. The open circles and the dashed line indicate the optic tract used to determine the medial/lateral location of MeApd^Foxp2 and MeApd^Dbx1 cells. (b-f) Distribution of MeApd^Foxp2 (magenta) and MeApd^Dbx1 (green) cell distance to the optic tract at Bregma −1.5 mm (b), −1.6 mm (c), −1.7 mm (d), −1.8 mm (e) and −1.9 mm (f). (g) Representation of a histological MeAp image. The open circles and the dashed line indicate the MeA dorsal edge used to determine the dorsal/ventral location of MeApd^Foxp2 and MeApd^Dbx1 cells. (h-l) Distribution of MeA^Foxp2 (magenta) and MeA^Dbx1 (green) cell distance from the dorsal edge at Bregma −1.5 mm (h), −1.6 mm (i), −1.7 mm (j), −1.8 mm (k) and −1.9 mm (l). Magenta and green dashed lines denote the median medial-lateral (b-f) and ventral-lateral (h-l) locations of the MeApd^Foxp2 and MeApd^Dbx1 cells, respectively. White arrows indicate how the distance to the optic tract (a) or the dorsal edge of the MeApd (g) is calculated. n = 133-568 from 3 animals for each group (where n is the number used for statistical analysis). (b-f, h-l) Two-tailed Mann Whitney tests; (b) p = 0.0018, (c) p = 0.0002, (e, i-l) p < 0.0001. **p < 0.01; ***p < 0.001; ***p < 0.0001. Otherwise, p > 0.05. See Source Data Extended Data Fig. 1 for more detailed statistics.

Source data

Extended Data Fig. 2 Additional characterization of MeA^Foxp2 and MeA^Dbx1 cell responses in experienced male mice, related to Fig. 3.

(a) Average PETHs of MeA^Foxp2 Ca²⁺ signal aligned to the onset (left) and offset (right) of investigation only (blue) and investigation-followed-by-attack (purple) trials. Open circles denote the period with significantly different responses (p < 0.05). (b) Average PETHs of MeA^Dbx1 cell responses aligned to the onset (left) and offset (right) of investigation only (blue) and investigation-followed-by-attack (purple) trials. The two traces do not differ significantly at any time point. (a and b) Two-tailed one-sample t-test at each time point corrected for repeated testing with FDR 0.05. Parentheses denote the number of animals. Data are mean ± SEM. See Source Data Extended Data Fig. 2 for more detailed statistics.

Source data

Extended Data Fig. 3 MeA^Foxp2 cell responses to male and female mice from multiple strains, related to Fig. 3.

(a-d) Representative Ca²⁺ traces of MeA^Foxp2 cells during the presentation of a BALB/c male (a), a C57BL/6 male (b), a C57BL/6 female (c), and a 129S4/SvJae female (d) mouse. (e) Average PETHs of MeA^Foxp2 cell responses aligned to investigation onset of mice from different strains. The dashed black line represents the behavior onset at time zero. (f) Average Fz scores of MeA^Foxp2 cells during investigation of male and female mice from different strains; p = 0.0052. Friedman test followed by multiple comparison tests with correction for FDR of 0.05. n = number of animals. Data are mean ± SEM; *p < 0.05, **p < 0.01. Otherwise, p > 0.05. See Source Data Extended Data Fig. 3 for more detailed statistics.

Source data

Extended Data Fig. 4 Quantification of behaviors towards different intruders in Foxp2^GCaMP and Dbx1^GCaMP male mice, related to Fig. 3.

(a) Percent of time naïve Foxp2^GCaMP, socially experienced Foxp2^GCaMP and socially experienced Dbx1^GCaMP male mice spent investigating a male intruder. (b) The fraction of naïve Foxp2^GCaMP, socially experienced Foxp2^GCaMP and socially experienced Dbx1^GCaMP male mice that attacked a male intruder. (c-e) Percent of time naïve Foxp2^GCaMP, socially experienced Foxp2^GCaMP and socially experienced Dbx1^GCaMP male mice spent attacking a male intruder (c), as well as the latency to first attack (d) and total number of attack events (e). (f-l) Percent of time Foxp2^GCaMP and Dbx1^GCaMP male mice spent investigating (f), mounting (g), and intromitting (j) a receptive female intruder, as well as the total number of mounting events per recording session (h), latency to first mount (i), total number of intromission events (k) and latency to ejaculation (l). (m) Percent of time Foxp2^GCaMP and Dbx1^GCaMP male mice spent investigating a pup intruder. (a, c-m) Two-tailed unpaired t-tests (if pass normality test) or two-tailed Mann Whitney tests (if not pass normality test) between naïve Foxp2^GCaMP and socially experienced Foxp2^GCaMP animals, and between socially experienced Foxp2^GCaMP and Dbx1^GCaMP animals; (d) p = 0.0148, (f) p < 0.0001, (h) p = 0.0039 between naïve Foxp2^GCaMP and socially experienced Foxp2^GCaMP animals, (j) p = 0.0045 between naïve Foxp2^GCaMP and socially experienced Foxp2^GCaMP animals, (k) p = 0.0016 between naïve Foxp2^GCaMP and socially experienced Foxp2^GCaMP animals, (m) p = 0.0025 between socially experienced Foxp2^GCaMP and socially experienced Dbx1^GCaMP male mice. (b) Two-tailed Chi-square tests; p = 0.0023 between naïve Foxp2^GCaMP and socially experienced Foxp2^GCaMP animals. n = number of animals. Data are mean ± S.E.M.; *p < 0.05, **p < 0.01, ***p < 0.001. Otherwise, p > 0.05. See Source Data Extended Data Fig. 4 for more detailed statistics.

Source data

Extended Data Fig. 5 GCaMP viral expression and investigation time in Foxp2^cre mice at P25, related to Fig. 5.

(a) Percent of time P25 Foxp2^cre mice spent investigating an anaesthetized male, anaesthetized female or pup intruder after fiber implantation surgery at P24 or without surgery. Two-way ANOVA followed by Sidak’s multiple comparison tests. All p values > 0.05. Parentheses denote the number of animals. Data are mean ± SEM. (b) Schematic of virus injection and a representative histology image indicating GCaMP6f expression (green), Foxp2 antibody (red), and DAPI (blue) staining in Foxp2^cre male mice at P25. White dotted lines mark the fiber end. See Source Data Extended Data Fig. 5 for more detailed statistics.

Source data

Extended Data Fig. 6 Additional behavioral assays after chemogenetic activation and inactivation of MeA^Foxp2 cells, related to Fig. 7.

(a) In Foxp2^hM3Dq and Foxp2^mCherry male mice, the latency to attack a male intruder after CNO injection did not differ from that after saline injection. Only animals that showed attack after both saline and CNO injections were included in this analysis. (b) Percent of time Foxp2^hM3Dq male mice spent attacking a male intruder during three consecutive days of testing after 1^st day i.p. saline injection (SAL), 2^nd day CNO injection (CNO), and 3^rd day saline injection (SAL2). (c) No changes in velocity (pixels/frame) in Foxp2^hM3Dq or Foxp2^mCherry male mice were observed in a 5 min period 30 min after CNO or saline injection when the test animal was alone in its cage. (d) Number of Foxp2^hM3Dq or Foxp2^mCherry male mice that attacked pups vs. those that did not after saline or CNO injection. Each circle represents one mouse. (e) Percentage of time Foxp2^hM3Dq or Foxp2^mCherry male mice spent investigating the pup after saline or CNO injection. (f) Representative raster plots showing the behaviors towards an adult female intruder of 5 Foxp2^hM3Dq and 5 Foxp2^mCherry mice after i.p. injection of saline or CNO. (g-l) Between CNO-injected and saline-injected days, there is no difference in any parameters related to male sexual behaviors in Foxp2^mCherry and Foxp2^hM3Dq male mice. (m) Representative raster plots showing the behaviors towards an adult female intruder of 5 Foxp2^hM4Di and 5 Foxp2^mCherry mice after i.p. injection of saline or CNO. (n-s) Between CNO-injected and saline-injected days, there is no difference in any parameters related to male sexual behaviors in Foxp2^mCherry and Foxp2^hM4Di male mice. (b) Friedman tests followed by multiple comparison tests with FDR correction; p < 0.0001. (c, e, g-l, n-s) Two-way repeated measures ANOVA followed by Sidak’s multiple comparison tests; (d) Two-tailed McNemar’s test. n = number of animals. Data are mean ± SEM. **p < 0.01, ***p < 0.001. Otherwise, p > 0.05. See Source Data Extended Data Fig. 6 for more detailed statistics.

Source data

Extended Data Fig. 7 The aggression level of Foxp2^cre and Dbx1^cre;LSL-FlpO naïve male mice, related to Fig. 7.

(a) Percent of time Foxp2^cre and Dbx1^cre;LSL-FlpO male mice spent investigating a male intruder. (b) Percent of Foxp2^cre and Dbx1^cre;LSL-FlpO male mice that attacked a male intruder. (c) Percent of time Foxp2^cre and Dbx1^cre;LSL-FlpO male mice spent attacking a male intruder. (d) Attack latency of Foxp2^cre and Dbx1^cre;LSL-FlpO male mice that attacked the male intruder. (e) Number of attacks of Foxp2^cre and Dbx1^cre;LSL-FlpO male mice towards a male intruder. Data are the same as those in Fig. 7e–k after day 1 saline injection. (a,d) Two-tailed unpaired t-test. (b) Two-tailed Chi-Square test; p = 0.0455 (c,e) Two-tailed Mann-Whitney test; (c) p = 0.0465. n = number of animals. Data are mean ± SEM; *p < 0.05. Otherwise, p > 0.05. See Source Data Extended Data Fig. 7 for more detailed statistics.

Source data

Extended Data Fig. 8 Additional behavioral assays during chemogenetic activation and inactivation of MeA^Dbx1 cells, related to Fig. 7.

(a) In Dbx1^hM3Dq and Dbx1^mCherry male mice, the latency to attack a male intruder after CNO injection did not differ from that after saline injection. Only animals that showed attack after both saline and CNO injections were included in this analysis. (b) Velocity (pixels/frame) of Dbx1^hM3Dq or Dbx1^mCherry male mice in a 5 min period after 30 min CNO or saline injection when the test animal was alone in its home cage. (c) Number of Dbx1^hM3Dq and Dbx1^mCherry male mice that attacked pups vs. those that did not after saline or CNO injection. Each circle represents one mouse. (d) Percentage of time Dbx1^hM3Dq and Dbx1^mCherry male mice spent investigating the pup after saline or CNO injection. (e) Representative raster plots showing the behaviors towards an adult female intruder of 5 Dbx1^hM3Dq and 5 Dbx1^mCherry mice after i.p. injection of saline or CNO. (f-k) No difference in male sexual behaviors after CNO injection in comparison to saline injection in Dbx1^mCherry or Dbx1^hM3Dq male mice. (l) Representative raster plots showing the behaviors towards an adult female intruder of 5 Dbx1^hM4Di and 5 Dbx1^mCherry mice after i.p. injection of saline or CNO. (m-r) No difference in male sexual behaviors after CNO injection in comparison to saline injection in Dbx1^mCherry or Dbx1^hM4Di male mice. (a, b, d, f-k, m-r) Two-way repeated measures ANOVA followed by Sidak’s multiple comparison tests. (c) Two-tailed McNemar’s test. n = number of animals. Data are mean ± SEM. All p > 0.05. See Source Data Extended Data Fig. 8 for more detailed statistics.

Source data

Extended Data Fig. 9 Brain regions downstream of MeA^Foxp2 and MeA^Dbx1 cells, related to Fig. 8.

(a-b) Representative images of 6 brain regions showing the GFP fibers originating from MeA^Foxp2 (a) and MeA^Dbx1 (b) cells.

Extended Data Table 1 List of brain regions and abbreviations

Full size table

Supplementary information

Reporting Summary

Source data

Source Data Fig. 1

Statistical Source Data.

Source Data Fig. 2

Statistical Source Data.

Source Data Fig. 3

Statistical Source Data.

Source Data Fig. 4

Statistical Source Data.

Source Data Fig. 5

Statistical Source Data.

Source Data Fig. 6

Statistical Source Data.

Source Data Fig. 7

Statistical Source Data.

Source Data Fig. 8

Statistical Source Data.

Source Data Extended Data Fig. 1

Statistical Source Data.

Source Data Extended Data Fig. 2

Statistical Source Data.

Source Data Extended Data Fig. 3

Statistical Source Data.

Source Data Extended Data Fig. 4

Statistical Source Data.

Source Data Extended Data Fig. 5

Statistical Source Data.

Source Data Extended Data Fig. 6

Statistical Source Data.

Source Data Extended Data Fig. 7

Statistical Source Data.

Source Data Extended Data Fig. 8

Statistical Source Data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Lischinsky, J.E., Yin, L., Shi, C. et al. Transcriptionally defined amygdala subpopulations play distinct roles in innate social behaviors. Nat Neurosci 26, 2131–2146 (2023). https://doi.org/10.1038/s41593-023-01475-5

Download citation

Received: 16 July 2022
Accepted: 29 September 2023
Published: 09 November 2023
Issue Date: December 2023
DOI: https://doi.org/10.1038/s41593-023-01475-5

This article is cited by

A molecularly defined amygdala-independent tetra-synaptic forebrain-to-hindbrain pathway for odor-driven innate fear and anxiety
- Hao Wang
- Qin Wang
- Xiao-Ming Li
Nature Neuroscience (2024)
Substance P in the medial amygdala regulates aggressive behaviors in male mice
- Zi-Xuan He
- Mei-Hui Yue
- Xiao-Juan Zhu
Neuropsychopharmacology (2024)

Subjects

Abstract

Similar content being viewed by others

Main

Results

Distribution of MeADbx1 and MeAFoxp2 cells in male mice

MeADbx1 and MeAFoxp2 cells in male mice are predominantly GABAergic

Distinct MeAFoxp2 and MeADbx1 cell responses to social sensory cues

Distinct MeADbx1 and MeAFoxp2 cell responses during social behaviors

Refinement of MeAFoxp2 cell responses with adult social experience

The male-specific response of MeAFoxp2 cells exists before puberty

Differential inputs to MeAFoxp2 and MeADbx1 cells

Activating MeAFoxp2 cells is sufficient for aggression in naive mice

Inhibiting MeAFoxp2 cells reduces aggression in experienced animals

Differential outputs of MeADbx1 and MeAFoxp2 cells

Discussion

MeAFoxp2 and MeADbx1 cell activity and function in social behaviors

Developmentally wired versus experientially wired

Social behavior circuits beyond MeA

Methods

Mice

Viruses and stereotaxic surgery

Behavioral assays and analysis

Fiber photometry

Chemogenetic-mediated activation and silencing

Animal body tracking

Distribution of MeAFoxp2 and MeADbx1 cells across the medial-lateral and dorsal-ventral axis

Triple in situ hybridization RNAscope

Immunohistochemistry and imaging analysis

Monosynaptic retrograde rabies input mapping

Output axonal projection mapping

Brain slice electrophysiology

Statistics

Reporting summary

Data availability

Code availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Extended Data Fig. 1 Characterization of MeAFoxp2 and MeADbx1 cell distribution in the MeApd, related to Fig. 1.

Extended Data Fig. 2 Additional characterization of MeAFoxp2 and MeADbx1 cell responses in experienced male mice, related to Fig. 3.

Extended Data Fig. 3 MeAFoxp2 cell responses to male and female mice from multiple strains, related to Fig. 3.

Extended Data Fig. 4 Quantification of behaviors towards different intruders in Foxp2GCaMP and Dbx1GCaMP male mice, related to Fig. 3.

Extended Data Fig. 5 GCaMP viral expression and investigation time in Foxp2cre mice at P25, related to Fig. 5.

Extended Data Fig. 6 Additional behavioral assays after chemogenetic activation and inactivation of MeAFoxp2 cells, related to Fig. 7.

Extended Data Fig. 7 The aggression level of Foxp2cre and Dbx1cre;LSL-FlpO naïve male mice, related to Fig. 7.

Extended Data Fig. 8 Additional behavioral assays during chemogenetic activation and inactivation of MeADbx1 cells, related to Fig. 7.

Extended Data Fig. 9 Brain regions downstream of MeAFoxp2 and MeADbx1 cells, related to Fig. 8.

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Search

Quick links

Distribution of MeA^Dbx1 and MeA^Foxp2 cells in male mice

MeA^Dbx1 and MeA^Foxp2 cells in male mice are predominantly GABAergic

Distinct MeA^Foxp2 and MeA^Dbx1 cell responses to social sensory cues

Distinct MeA^Dbx1 and MeA^Foxp2 cell responses during social behaviors

Refinement of MeA^Foxp2 cell responses with adult social experience

The male-specific response of MeA^Foxp2 cells exists before puberty

Differential inputs to MeA^Foxp2 and MeA^Dbx1 cells

Activating MeA^Foxp2 cells is sufficient for aggression in naive mice

Inhibiting MeA^Foxp2 cells reduces aggression in experienced animals

Differential outputs of MeA^Dbx1 and MeA^Foxp2 cells

MeA^Foxp2 and MeA^Dbx1 cell activity and function in social behaviors

Distribution of MeA^Foxp2 and MeA^Dbx1 cells across the medial-lateral and dorsal-ventral axis

Extended Data Fig. 1 Characterization of MeA^Foxp2 and MeA^Dbx1 cell distribution in the MeApd, related to Fig. 1.

Extended Data Fig. 2 Additional characterization of MeA^Foxp2 and MeA^Dbx1 cell responses in experienced male mice, related to Fig. 3.

Extended Data Fig. 3 MeA^Foxp2 cell responses to male and female mice from multiple strains, related to Fig. 3.

Extended Data Fig. 4 Quantification of behaviors towards different intruders in Foxp2^GCaMP and Dbx1^GCaMP male mice, related to Fig. 3.

Extended Data Fig. 5 GCaMP viral expression and investigation time in Foxp2^cre mice at P25, related to Fig. 5.

Extended Data Fig. 6 Additional behavioral assays after chemogenetic activation and inactivation of MeA^Foxp2 cells, related to Fig. 7.

Extended Data Fig. 7 The aggression level of Foxp2^cre and Dbx1^cre;LSL-FlpO naïve male mice, related to Fig. 7.

Extended Data Fig. 8 Additional behavioral assays during chemogenetic activation and inactivation of MeA^Dbx1 cells, related to Fig. 7.

Extended Data Fig. 9 Brain regions downstream of MeA^Foxp2 and MeA^Dbx1 cells, related to Fig. 8.