Amygdala neural ensemble mediates mouse social investigation behaviors

ABSTRACT Innate social investigation behaviors are critical for animal survival and are regulated by both neural circuits and neuroendocrine factors. Our understanding of how neuropeptides regulate social interest, however, is incomplete at the current stage. In this study, we identified the expression of secretin (SCT) in a subpopulation of excitatory neurons in the basolateral amygdala. With distinct molecular and physiological features, BLASCT+ cells projected to the medial prefrontal cortex and were necessary and sufficient for promoting social investigation behaviors, whilst other basolateral amygdala neurons were anxiogenic and antagonized social behaviors. Moreover, the exogenous application of secretin effectively promoted social interest in both healthy and autism spectrum disorder model mice. These results collectively demonstrate a previously unrecognized group of amygdala neurons for mediating social behaviors and suggest promising strategies for social deficits.


INTRODUCTION
Both humans and rodents display social interest toward conspecific individuals [1][2][3]. Such innate social behaviors are mediated by both neural circuits [4][5][6] and neuroendocrine factors [7][8][9]. Various brain regions, including the prefrontal cortex [6,[10][11][12][13] and subcortical nuclei such as the hippocampus [14,15], ventral tegmental area (VTA) [5,16] and hypothalamus [17,18], provide neural substrates for social behaviors. As a pivotal region integrating emotional information, the amygdala's nuclei have also been implicated in social interactions [19,20]. For example, the activation of the medial amygdala (MeA) suppresses social interest [21,22], and the negative regulation of social engagement [23][24][25] has been reported in the basolateral amygdala (BLA). However, one cannot exclude the effect of negative emotional status on social inhibition due to the widely accepted function of the BLA in stress response and anxiety coding [26]. Further dissection of the molecular and cellular composition of the BLA during social engagement is therefore required.
A recently identified computational model proposed the coding of social exploration by an activity-defined neuronal ensemble of the mouse BLA [27]. These results, in conjunction with clinical evidence supporting the role of the BLA in human social experiences [28,29], raise one plausible model, as distinct neuronal subpopulations of the BLA may regulate unique behavioral modules including social investigation. Specific molecular markers are expected for neuronal subtyping to address this question. Besides neurotransmitters, neuropeptides are frequently proposed to define distinct neuronal subpopulations with specific neuronal circuits and behavioral modules, such as the cholecystokinin (CCK)+ cells of the BLA in mediating depressive-like behaviors [30]. Here we focused on secretin (SCT), a classical gut-peptide hormone that was initially identified in duodenal tissues for stimulating pancreatic secretion [31]. Recent advances also revealed the expression of SCT across multiple brain regions, including the cerebellum, hippocampus, hypothalamus and amygdala [32]. The receptor of SCT has also been found across the brain stem, limbic system and hypothalamic nuclei [33]. SCT thus potentially acts as one neuropeptide hormone executing both physiological and behavioral modulatory functions. C The Author(s) 2022. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Previous findings showed the role of SCT in the central regulation of food [34] or water [35] intake, as well as in modulating spatial memory [36]. It is worth noting that the activation of supraoptic neurons by SCT facilitates social recognition in rodents [37], and the potential value of SCT in ameliorating social deficits in autistic spectrum disorder (ASD) patients has been debated [38][39][40]. We thus investigated the molecular and circuitry mechanisms of SCT in mediating social behaviors.
In the current study, by generating an SCT-Cre knock-in mouse line, we identified a distinct subpopulation of excitatory BLA neurons that express SCT. Those BLA SCT+ cells display the same transcriptomic profiles as neuroendocrine cells, and specifically facilitate social investigation behaviors toward an unfamiliar, same-sex mouse while other BLA cells are anxiogenic and can antagonize such social interest. The distinction between these two BLA neuronal subtypes also resides in the cell-type-specific projecting patterns, as BLA SCT+ neurons preferentially activate the medial prefrontal cortex (mPFC) for facilitating social affiliations. Moreover, this amygdalaprefrontal circuit can be activated by SCT, which helps to attenuate social deficits in an ASD model. Our results collectively demonstrate an amygdalaprefrontal pathway to elicit social behaviors, and suggest a promising intervention strategy targeting social deficits.

SCT is expressed in the BLA and mediates innate social behaviors
To fully characterize the spatial expression pattern of SCT, we generated an SCT-IRES-Cre knockin mouse line ( Fig. S1a and b). By crossing SCT-Cre mice with a second reporter line (Ai9, R26-tdTomato), the whole-brain expressional profile of SCT was revealed (Fig. S1c). In general, SCT was minimally expressed in cortical regions, but with relatively high levels across various subcortical nuclei in the limbic system, midbrain and hindbrain ( Fig. S1d and e). The specificity of SCT-Cre was further revealed by the localization of SCT transcript in tdTomato + cells (Fig. S1f). In particular, extensive distributions of SCT were found in the periaqueductal gray (PAG), ventral medial hypothalamus (VMH) and BLA, in addition to the cerebellum ( Fig. 1a and b). These SCT-expressing neurons in the BLA (BLA SCT+ ) lay across the whole amygdala nuclei and occupy 33.9% of all CaMKIIα+ neurons but not parvalbumin (PV) interneurons ( Fig. 1ce). These data collectively suggest the enrichment of SCT in a subpopulation of excitatory neurons in the BLA.
Having observed the existence of SCT in the BLA, we next investigated its behavioral implications. By stereotaxic injection of an adeno-associated virus (AAV) carrying Cre recombinase into male homozygous SCT-floxed mice (SCT fl/fl ), we performed a battery of behavioral screenings and found no significant change of anxiety behaviors or fear conditioning in those conditional SCT knock-out mice (Fig. S2a-e). These results suggest that SCT seems to be irrelevant to those well-known functions of the BLA. When we adopted a social exploration task in which the test mouse was allowed to visit an unfamiliar male mouse (4 weeks old) in the center of a novel field, however, the downregulation of SCT in the BLA (Fig. 1f-i) suppressed mouse social interest remarkably (Fig. 1j). Since the exploration behavior toward an empty cage was not affected by SCT knock-down (Fig. 1k), the decreased social interaction time and social preference ratio (= social time/cage exploration time) faithfully reflected impaired social investigation interest rather than the carry-on effect by potential anxiolysis (Fig. 1l and m). To further validate the social modulation of SCT, we recruited the classical 3-chamber assay and found that during phase II (with one novel mouse and one empty cage), SCT knock-down deprived normal social preference ( Fig. 1n-p). Moreover, in female mice with SCT knock-down, social investigation toward an unfamiliar female conspecific was also repressed in both an open field and 3-chamber apparatus ( Fig. S2f-k), illustrating the robust role of SCT in modulating social exploration behaviors across genders. Therefore, SCT in the BLA is necessary for innate social approaching behaviors.

BLA SCT+ neurons positively mediate social interaction
The behavioral phenotyping of SCT knock-down assays seems to be incompatible with current knowledge, which claims the suppressive effect of social interaction by BLA projecting neurons [23][24][25]. But the existence of SCT in only one third of excitatory neurons of the BLA (Fig. 1e) leaves one possible explanation: that SCT represents a distinct group of BLA cells that specifically facilitates social interest. To test this, we firstly performed a cell-ablation assay to selectively erase BLA SCT+        investigation behaviors, although the decrement of cage exploration time (Fig. S3d) also indicated the multifaceted function of BLA SCT+ neurons.
To better correlate the cellular activity of BLA SCT+ neurons with social behaviors, we quantified the immediate early gene (IEG) cFos expression after a brief social exploration session using SCT-Cre; Ai9 reporter lines (Fig. 2a). Compared with mice encountering an empty cage, sociable mice presented more activated BLA SCT+ cells ( Fig. 2b and c). An in vivo recording assay was subsequently performed by implanting an optic fiber into the BLA of a SCT-Cre mouse whose SCT+ cells were infected with genetically encoded calcium indicator GCaMP6m ( Fig. 2d and e). Using the fiber photometry apparatus, the population activity of BLA SCT+ cells was recorded when the animal faced different grades of social cues (empty cage, fake mouse, fake mouse with urine from a novel mouse, and novel animal; see Fig. 2f). In general, those cells showed significantly elevated calcium peaks when facing the urine odor of an unfamiliar mouse but not the fake mouse, and the strongest response occurred during the actual social interaction scenario ( Fig. 2g and h). Those results provide in vivo evidence supporting the coding of social cues by BLA SCT+ cells.
The role of BLA SCT+ neurons in social interaction was further tested by a cell-type-specific manipulation strategy in which channelrhodopsin-2 (ChR2) was expressed in those cells ( Fig. 3a and  b). Under the same social exploration test, cell activation strongly potentiated social sniffing behaviors ( Fig. 3c and d, Video S1). Similar phenotypes were observed in the 3-chamber assay, in which light activation potentiated mouse social preference, and the removal of optogenetic stimulus abolished such increment ( Fig. 3e and f). On the other hand, optogenetic inhibition of BLA SCT+ neurons using halorhodopsin (NpHR; Fig. 3g and h) suppressed social interest, as the light stimulus remarkably inhibited the social interest of the test mouse ( Fig. 3i and j). Three-chamber assays also displayed similar behavioral modulations, as inhibition of BLA SCT+ cells had diminished social interest, which was rapidly recovered after the removal of light stimulus ( Fig. 3k and l). To rule out the possible effects of place preference or social-related rewards, we repeated the optogenetic excitation assay but with the removal of the novel mouse. The absence of social cues did not facilitate sniffing toward the empty cage, given the existence of prior social experience (Fig. S4a). Moreover, the activation of BLA SCT+ neurons did not cause a preference toward the empty cage (Fig. S4b). These real-time neural manipulation assays agreed with prior in vivo recording results (Fig. 2), and support both the necessary and sufficient role of BLA SCT+ cells in modulating social investigation behaviors. As further evidence, we also performed chemogenetic assays by introducing designer receptors exclusively activated by designer drugs (DREADDs) hM3Dq or hM4Di into BLA SCT+ neurons ( Fig. S5a and b). The infusion of receptor ligand clozapine-N-oxide (CNO) rapidly potentiated social investigation behaviors in hM3Dq-infected animals ( Fig. S5c-f) or suppressed social interest toward the novel animal when hM4Di was expressed in BLA SCT+ neurons ( Fig. S5g-j). In sum, the BLA SCT+ neural ensemble is activated by social cues and facilitates innate social investigation behaviors.

SCT defines two distinct BLA neuronal subpopulations
The demonstration of BLA SCT+ cells in mediating social interactions cannot fully support the specificity of such a neural ensemble in behavioral modulation, as other excitatory neurons in the BLA (BLA SCT− ) may also participate in this process. To address this question, we designed a second set of AAV vectors carrying CaMKIIα-ChR2 under a double-floxed orientation (DO) frame, whose expression can be turned off only by the presence Natl Sci Rev, 2023, Vol. 10, nwac179 of Cre recombinase [41][42][43] (Fig. 4a). After expressing ChR2 in BLA SCT− cells, a set of behavioral tests found that the activation of BLA SCT− cells induced anxiety-like behaviors (Fig. 4b-e), agreeing with our current knowledge of the BLA in the anxiogenic process. We further propose that BLA SCT− cells may play different roles in social interaction in contrast with BLA SCT+ cells. To further validate the antagonistic effects of two BLA subpopulations, we co-infected ChR2 and a second excitatory optogenetic receptor, ChrimsonR, with different excitation spectra [44], into the BLA  of SCT-Cre mice, under DIO and DO flanked sequences, respectively (Fig. 4f). Such experimental design successfully implanted a dual-switch system in which BLA SCT+ neurons were activated by 473 nm excitation light, and BLA SCT− cells were excited under 594 nm wavelength stimulation (Fig. 4g). During the social approaching paradigm, the sequential activation of two neuronal subtypes produced opposite effects on social sniffing duration ( Fig. 4h and i, Video S2). Specifically, ChR2 stimulation on BLA SCT+ facilitated social interest, whilst Chrim-sonR excitation on BLA SCT− inhibited social behaviors. Such phenotypes are stable despite the change of stimulating sequence ( Fig. 4j and k), suggesting the cell specificity of these two BLA subpopulations for antagonistic modulation of social interest. The distinct behavioral modules further implied possibly unique molecular features between BLA SCT+ and BLA SCT− cells. To address this issue, we firstly studied the possible overlap between SCT and known molecular markers of BLA cells. The BLA SCT+ cells were mostly co-localized with Ppp1r1b (Fig. S6), which represents a group of BLA cells encoding positive valence [45]. However, as the BLA SCT+ cells only represent a fraction of Ppp1r1b + cells, we further characterized those neurons by infecting the SCT-Cre; Ai9 reporter line with AAV-CaMKIIα-DO-GFP, resulting in dual fluorescent labels on BLA SCT+ and BLA SCT− cells (Fig. S7a). A preliminary examination found minimal overlapping between GFP+ and tdTomato+ labeled cells, as BLA SCT+ cells mainly occupy the medial region whilst BLA SCT− cells are predominantly located in the lateral part (Fig. S7b).
The divergent spatial distribution further implies unique molecular, physiological and anatomic features of these two cell types. We thus performed a single-cell-based transcriptomic study after extracting cellular contents via a glass pipette (Fig. S7a). Based on data from a total of 27 neurons, the principal component analysis (PCA) showed that BLA SCT− and BLA SCT+ cells belonged to two distinct groups with unique transcriptomic profiles (Fig. S7c). Of note, a total of 2440 genes were differentially expressed (Figs S7d and S8). Gene enrichment analysis indicated that these genes were involved in neuropeptide ligand-receptor binding and intracellular pathways such as cAMP ( Fig. S7e and  f). Those molecular features further implied distinctions of cellular properties. Ex vivo electrophysiological recording showed that BLA SCT+ cells exhibited higher excitability, lower input resistance and lower rheobase compared to their counterparts without SCT expression (Fig. S7g-j), in conjunction with altered membrane properties ( Fig. S9a and b) but with their action potential (AP) kinetics left unchanged (Fig. S9c-h). Those comparative studies demonstrated that BLA SCT+ represented a unique group of excitatory neurons that specifically mediates social investigation behaviors.

BLA SCT neurons drive an amygdalaprefrontal circuit to regulate social behaviors
In addition to the molecular and electrophysiological features, features in neural circuitry architectures may further help to address the differential behavioral modules between BLA SCT+  and BLA SCT− cells. Using the AAV-CaMKIIα-DO-GFP vector-mediated infection on SCT-Cre mice (Fig. S10a), we found that BLA SCT− neurons prominently extend their axonal terminus into the ventral CA1, NAc and bed nucleus of the stria terminalis (BNST) in addition to the central amygdala (CeA) (Fig. S10b and c), which agrees with previous findings [30,[46][47][48]. However, when we examined the downstream target of BLA SCT+ neurons using an anterograde trans-synaptic herpes simplex virus (HSV) in conjunction with the Cre-mediated helper virus (Fig. 5a), those cells preferentially innerved the mPFC in addition to the proximal CeA, instead of subcortical projections ( Fig. 5b and c).
To validate the existence of this BLA→mPFC pathway, a modified rabies virus (RV) was infused into the prelimbic (PrL) of SCT-Cre; Ai9 reporter mice (Fig. 5d) to trace the input of those cells. In the BLA region, ∼48.3% of PrL-projecting BLA cells were colabeled with SCT ( Fig. 5e and f), highlighting the input of SCT+ cells in this circuit. The nature of this synaptic connection was further examined by patch-clamp recording of PrL neurons under ex vivo optogenetics stimulation, with prior ChR2 infection into BLA SCT+ neurons (Fig. 5g). The abolishment of excitatory postsynaptic current (EPSC) by tetrodotoxin (TTX) infusion, plus the rapid recovery by 4-aminopyridine (4-AP) all suggested the monosynaptic excitatory transmission between BLA SCT+ and PrL cells (Fig. 5h and i).
Since the mPFC has been recognized as playing a role in social behaviors [11][12][13], we next tested if the BLA SCT+ →mPFC pathway is relevant to the social interaction modulation. After infecting BLA SCT+ cells with Cre-dependent ChR2, we implanted an optic fiber into the mPFC region to stimulate the axonal terminus originating from the BLA (Fig. 6a). Using identical experimental protocols as those in somatic manipulation (Fig. 3), the activation of the BLA SCT+ →mPFC terminus elongated social sniffing durations in the social exploration assay (Fig. 6c and d) or in the 3-chamber assay ( Fig. 6e and f). Light-mediated inhibition of these fibers (Fig. 6g) also mimicked the effect of soma manipulation, suggesting the suppression of social interest (Fig. 6h-k). Combining all those data, it is likely that BLA SCT+ specifically facilitates innate social investigation behaviors primarily via their projections to the mPFC.

Exogenous SCT facilitates social behaviors in naïve and ASD mice
Since BLA SCT+ population potentiates social interest, further questions are raised as to whether SCT peptide plays a role in social modulations. To elaborate the molecular mechanism of social behaviors, we firstly identified the prominent expression of SCT receptor (SCTR) transcript in the BLA region (Fig. 7a), providing the molecular substrate for locally SCT-mediating effects. In a second study using acute BLA slices from SCT-Cre mice infected with AAV-dio-ChR2, a brief light-mediated excitation of BLA SCT+ cells remarkably increased SCT release into a culture medium (Fig. 7b and c). These results indicated possible autocrine and/or paracrine mechanisms in which SCT was released from activated BLA SCT+ neurons to potentiate social behaviors. As direct evidence of cellular functions, SCT perfusion induced higher spiking numbers plus lower input resistance or rheobase values, in addition to altered kinetics of AP under ex vivo patch-clamp recording ( Fig. 7d-g, Fig. S11), suggesting the  direct neuromodulatory role of SCT. Taken together, SCT acts as the central neuromodulator to potentiate BLA cells for mediating social investigation behaviors.
To further establish the role of SCT in social behaviors, we introduced a recombinant SCT peptide or an SCTR antagonist (SCT 5-27) into the bilateral BLA of C57 wild type male mice via a pre-implanted cannula (Fig. 7h). Social approaching assay found that SCT administration remarkably increased social exploration time and social preference, whilst receptor blockade by SCT 5-27 remarkably suppressed social interest but left the cage exploration time unchanged (Fig. 7i-l). To test whether the facilitated sniffing behavior was dependent on social cues, we repeated this assay using an aggressive CD1 mouse in the central cage. Behavioral phenotyping showed that, with prior exposure to CD1, SCT infusion did not enhance the social duration (Fig. S12), suggesting that SCT only enhanced 'amicable' social investigation but did not affect the avoidance of formidable social cues. Moreover, we employed a chronic SCT administration scenario in which SCT expression was persistently driven by an AAV vector infected into the BLA ( Fig. S13a and b). Similar to those phenotypes in SCT drug infusion, AAV-SCT infection remarkably increased social preference (Fig. S13c-e). On the other hand, the suppression of SCTR function using short hairpin RNA (shRNA) targeting SCTR also resulted in decreased social interest (Fig. S13f-i). Those data clearly suggested the indispensable role of the SCT-SCTR axis in innate social behaviors.
Lastly, we tested if these SCT-mediated social investigation behaviors can help to relieve relevant psychiatric disorders such as social dysfunctions in ASD. Different mouse ASD models converged to show synaptic defects in the prefrontal cortex [11,49], and we have proven the critical role of the BLA SCT+ →mPFC pathway in social behaviors. We thus investigated the amygdala-cortical pathway in a mouse ASD model developed by ablating the expression of the methyl-CpG binding protein 2 (MeCP2) gene in the mPFC region ( Fig. 8a  and b, Fig. S14). We identified decreased social durations and preference ratio in our test arena (Fig. 8d-f), which agreed with previous studies showing the neuropathology and ASD-related social deficits that occurred upon MeCP2 mutation [50,51]. However, when the BLA region of these MeCP2 knock-down animals was infected with AAV-SCT to elevate local expression of SCT (Fig. 8c), impaired social behaviors were recovered ( Fig. 8d-f). These results suggest that SCT may help to relieve social deficits. To provide a neural substrate for such behavioral modulations, we monitored the calcium activity of PrL cells using in vivo 2-photon microscopy (Fig. 8a). Results showed significantly reduced calcium activity under MeCP2 insufficiency, and re-potentiation by SCT over-expression in the BLA (Fig. 8g and h). In particular, SCT-mediated neuronal potentiation was mainly displayed by higher peak values of calcium spikes (Fig. 8i) rather than their frequencies (Fig. 8j), indicating stronger excitatory inputs. Such results were consistent with our findings showing the mono-synaptic excitatory connection between BLA SCT+ and mPFC cells (Fig. 5). In sum, our molecular, physiological, anatomical and behavioral evidence collectively identified two mutually exclusive BLA subpopulations, of which BLA SCT+ mediates social investigation behaviors via their mPFC projection, and BLA SCT− cells are mainly anxiogenic neurons that can antagonize social interest (Fig. 8k).

DISCUSSION
Our study defined a subpopulation of BLA neurons that can be activated upon social cues and elicit social interactions. These BLA SCT+ neurons might at least partially overlap with the 'social ensemble' [27]   due to their similar percentage of total BLA neurons (∼30%) and their active role in social interaction behaviors. With distinct molecular and transcriptomic profiles, BLA SCT+ neurons preferentially project to the mPFC to facilitate social engagement, providing a bottom-up model in which social cues are integrated in the BLA to drive social behaviors in cortical regions. Further elaboration also revealed the activation of such a circuit by SCT in both naïve and ASD model mice, implying the potential value for clinical intervention when treating social deficits. It has been suggested that amygdala neural circuits are involved in social behaviors [19], but are often correlated with negative valence coding [24] to suppress social engagement via the projection to midbrain nuclei such as the nucleus accumbens (NAc) [21,25]. Moreover, the top-down regulation from the prefrontal cortex towards amygdala nuclei affects social decision making [19,24], and GABAergic neurons in the MeA were recently reported to encode social reward via hypothalamic projections [22]. With regard to examining the positive control of control behaviors by BLA nuclei, the idea of amygdala representation of social exploration has recently been supported by the existence of the social ensemble in the BLA [27]. In juvenile mice, social approach was found to acutely activate BLA neurons [52], and our findings further identified these socialactivated BLA neurons as the SCT+ subpopulation. Moreover, the one-cage social exploration assay used in the current study may carry confounding factors related to anxiety status, as the novel mouse was placed in the center of the field. We thus performed the classical 3-chamber assay, which replicated key findings, as mice presented less social interest when SCT was knocked down in the BLA, or when BLA SCT+ neurons were deactivated. These observations largely exclude the effect from anxiety-like behavior and demonstrated that this BLA SCT+ neural ensemble specifically regulates social interest toward same-sex novel mice.
Classical views claimed that the BLA mainly consisted of CaMKIIα+ excitatory cells and PV+ inhibitory neurons [47,53]. However, such a divergent system seems to be over-simplified as a recent study reported the unique role of CCK+ cells of the BLA in mediating depressive-like behaviors [30]. SCT was also found to be expressed in the BLA in our study and participated in a different behavioral module. While the majority of identified BLA subpopulations encode negative valence such as anxiety or fear, our BLA SCT+ cells are involved in processing positive stimulus. Indeed, recent studies are suggesting the existence of certain 'reward-biased' cells in the BLA and their projections to the PrL [54], and a molecular dissection study has partially revealed two distinct groups of BLA cells to provide a neural substrate of positive and negative valence, respectively [45]. Our study, after excluding the possible confounding effect of place preference, supports the coding of social-related positive valence by BLA SCT+ cells. In future, the presynaptic input of those neurons can be further studied to better dissect both internal and environmental factors during social sensation and initiation.
The BLA SCT+ →mPFC pathway was shown here to affect social investigation behaviors. Currently available evidence suggests that the activation of the BLA→mPFC pathway suppresses social interaction in a resident-intruder test [24], but probably due to the concurrent anxiogenic effect [26]. Our study reported seemingly contradictory effects but can be explained by selectively manipulating BLA SCT+ cells, which consisted of <50% of the total number of mPFC-projecting cells. The existence of the BLA SCT− →mPFC pathway, plus our findings, which reveal the inhibition of social engagement upon BLA SCT− cell activation, suggest that paralleled pathways may exist from the BLA to mPFC to suppress social engagement. When we examined the BLA SCT+ →mPFC pathway, circuitry tracing and electrophysiological recording implied the excitatory nature of this synaptic connection. Moreover, in vivo calcium imaging of mPFC pyramidal neurons revealed the excitation of cortical neurons after BLA SCT+ cell activation, forming a bottom-up regulatory pathway. In the cortical region, since enhanced mPFC activity was closely correlated with social motivation and interaction [6,55,56], the BLA SCT+ →mPFC circuit thus provides an excitatory input into the cortical region for mediating social behaviors.
The presynaptic input for BLA SCT+ cells has not been studied yet but should be interesting for further exploration. Based on currently available anatomic studies, a lot of BLA inputs came from the frontal cortex, such as the cingulate cortex [57] and prefrontal cortex [58]. In terms of behavioral modulations, the mPFC inputs into the BLA are mainly involved in anxiety-like behaviors. Therefore, mPFC projection to the BLA may provide a top-down information flow reflecting mental status for affecting the social outcome, as suggested by other studies highlighting the PFC-driven activation of amygdala-cortical neurons [59]. Recently, a whole-brain study reported that BLA cells receive inputs from the entorhinal cortex, auditory thalamus and auditory cortex [60]. This anatomical evidence implies the possible role of BLA SCT+ cells as a hub for sensing multi-module environmental cues to initiate social behaviors, as supported by our in vivo calcium recording data (Fig. 2). Besides the performed in conjunction with circuitry recording or manipulation.
Statistical analysis: All data were firstly tested for normal distribution. Those fitting a normal distribution were analyzed by parametric approaches, whilst other data were compared using non-parametric methods.
Detailed materials and methods are available in the Supplementary Data.

DATA AVAILABILITY
All data are available within the article and the supplementary information or can be made available by the corresponding author (zhangli@jnu.edu.cn) upon reasonable request.