All animal experiments were performed in accordance with the guidelines of RIKEN’s Animal Experiment Committee and conducted in accordance with the National Institutes of Health guidelines (NIH publication no. 85–23, revised 1985).

BALB/c mice were originally purchased from the Jackson Laboratory in Japan and raised in our breeding colony. Mice were maintained under controlled conditions [12 h:12 h light/dark cycle (lights on at 08.00 am) with food and water ad libitum]. BLAB/c mice of both sexes were housed in groups of four or five after weaning at 4 weeks. All the mice were 3–7 months old at the beginning of the experiments.

The social isolation and reunion test was performed as described previously (Fig. 2A; [4]). Briefly, four group-housed BALB/c adult female mice were placed in a separation apparatus in the middle of the test cage consisting of a replaceable sham partition with two large holes and were accustomed to the cage for 7 days. Then, one out of four mice was separated into a no-nest chamber by replacing the sham partition with a real one with two barrier windows (somatic isolation). Three cage mates remained in the same chamber as their nest (3-together). Two days after somatic isolation, the real partition was replaced by a sham partition, and the pre-isolated mouse was reunited with its cage mates. Behavioral changes after somatic isolation or reunion were tracked for > 2 h and scored during either 15-s or 5-min intervals based on a previously established ethogram [4]. In addition to direct physical contact, the following behaviors were coded: (1) general movement (still/movement); (2) exploratory behavior (partition-sniffing, rearing, and digging); (3) contact-seeking behavior (partition-biting); (4) nonsocial behaviors (self-grooming, eating, and panic); (5) social behavior (crawling-under, peer-sniffing, allogrooming, mounting, peer-biting, and chasing). All behavioral data was quantified in an experimenter-blind setting.

Detailed protocols for in situ hybridization (ISH) and immunohistochemistry (IHC) are described elsewhere [4, 10]. For ISH combined with Calcr IHC, GAD67 mRNA (1064–2045 bp, NM_008077) was detected by ISH and Calcr-ir (immunoreactive) cells were detected by 3,3'-diaminobenzidine (DAB) reaction without nickel.

For Calcr or amylin mapping experiments, amygdala sections were stained with anti-Calcr or anti-amylin antibodies; positive cells were detected with a DAB reaction with nickel. Following the initial staining, NeuN-ir cells were counterstained with ImmPACT Vector Red (Vector Laboratories). For double labeling of Calcr and ERα, Calcr-ir neurons were detected with a DAB reaction without nickel. Following the first staining, ERα+ cells were detected by a DAB reaction with nickel.

To test the knockdown efficacy of Calcr shRNA, Calcr-ir neurons were detected by DAB reaction with nickel. Following the first staining, shRNA-viral infected and EGFP-ir cells were detected with ImmPACT Vector Red.

The ICH antibodies were as follows: rabbit anti-Calcr (1:4,000, PAb188/10, Welcome receptor antibodies); rabbit anti-amylin (1:20,000, cat#H-017-11, Phoenix Pharmaceuticals, Burlingame, CA, USA); rabbit anti-GFP antibody (1:5,000, 598, MBL); rabbit anti-ERα (1:20,000, 06–935, Merck Millipore), biotin-conjugated horse anti-rabbit (1:2000, BA-1100, Vector Laboratories), or anti-mouse (1:2000, BA-2000, Vector Laboratories) secondary antibody.

Bright-field images were acquired using a NanoZoomer Digital Pathology (Hamamatsu Photonics) with a 20 × objective.

For Calcr mapping in the amygdala, conservative contours were set on brain coronal sections based on NeuN counterstaining as previously described [4]. The number of labeled somata of each conservative contour was manually counted using the ImageJ plugin “Cell Counter.” A single brain section from each mouse was used for histological analyses related to Fig. 1. Histological analyses were not blinded due to the high complexity of these tasks.

Sexual dimorphic Calcr expression and gene profiles of Calcr + cells in the medial amygdala. A Representative coronal brain sections of both sexes, showing the distribution of Calcr immunoreactivity (Calcr-ir, black) and NeuN counterstaining (red) in subregions of the amygdala in female or male mice. MeAad anteroventral part of the medial amygdala, MeApd posterodorsal part of the medial amygdala, MeApv posteroventral part of the medial amygdala, CeM medial part of the central amygdala, CeL lateral part of the central amygdala, CeC capsular part of the central amygdala, AA anterior amygdaloid area, LA lateral amygdala, BLA basolateral amygdala, BMA basomedial amygdala. Scale bars, 500 μm. B Number of Calcr-ir neurons in each subregion of female or male mice (n = 4 mice per group). C Sections stained with Gad67 mRNA (blue) by ISH together with anti-Calcr (brown) antibody. Arrowheads indicate double-labeled cells. Scale bars, 500 μm (left) and 50 μm (right). D Sections stained with anti-ERα (black) by IHC together with anti-Calcr (brown) antibody. Scale bars, 500 μm (left) and 50 μm (right). E Sections stained with anti-amylin (black) by IHC together with anti-NeuN (red) antibody. Scale bars, 500 μm (left) and 50 μm (right). Asterisks in (B) denote significant differences between two groups (Welch’s unpaired t-test, *P < 0.05). Graphs represent mean ± SEM. See Additional file 1 for detailed statics

Stereotactic AAV injections were performed as previously [4, 6, 10]. Stereotactic coordinates for the MeApd were obtained from the mouse brain atlas [11]. The coordinates used were as follows: MeApd (AP − 1.6, ML ± 2.4, DV − 5.15).

The AAVs AAV5-Calcr shRNA-CAG-EGFP and AAV5-Cont shRNA-CAG-EGFP (1.0E11-1.0E12 vg/mL) were those used in our previous publication [4], and provided by Thomas McHugh’s laboratory. A total of 100 nL of each AAV was injected bilaterally into the MeApd of one in four female mice. At 2 weeks after AAV injection, viral injected mice were subjected to a somatic isolation test and a reunion test. After behavioral testing, mice were evaluated for proper viral injections: briefly, the brains were coronally sectioned at 40 μm, and every third section was taken out as one series. In one series, nine consecutive sections covering the MeApd were doubly stained for Calcr and EGFP immunohistochemistry. Mice without any expression of EGFP in the MeApd were excluded from the analyses (1 out of 16 females injected with Control shRNA and 4 out of 16 females). We confirmed no Calcr immunoreactivity in the EGFP + neurons as in our previous study [4]. Then, all the MeApd Calcr + neurons in the nine coronal sections were manually counted to yield Fig. 2C.

Targeted Calcr knockdown in the MeApd does not affect behavioral responses to social isolation. A Timeline of behavioral experiments. At 2 weeks before testing, female mice were administered an AAV expressing shRNA-EGFP for specific Calcr knockdown or with a shRNA virus carrying scrambled shRNA (Cont shRNA) into the MeApd region. An AAV-injected adult female mouse was subjected to a somatic isolation test and reunion tests. B Representative coronal sections of shRNA-EGFP viral expression in the MeApd showing the efficacy of Calcr shRNAs. Scale bars, 2.5 mm (Left), 500 μm (Middle and Right). A shRNA virus carrying scrambled shRNA was used as control. C Quantification of fold changes in the number of Calcr + cells distributed in the MeApd (Cont shRNA: n = 15 mice and Calcr shRNA: n = 12 mice). D Time course of biting responses (Cont shRNA: n = 15 mice and Calcr shRNA: n = 12 mice). E Quantification of the total number of biting responses over 30 min (Cont shRNA: n = 15 mice and Calcr shRNA: n = 12 mice). F Number of various behaviors per min observed under somatic isolation for 30 min (Cont shRNA: n = 15 mice and Calcr shRNA: n = 12 mice). Time bins 15 s (D–F). Asterisks denote significant differences between the two groups (C, E, F: Welch’s unpaired t-test, D: 2-way repeated measure ANOVA tests followed by Sidak's post hoc tests; ***P < 0.001). Graphs represent mean ± SEM. See Addi for detailed statics

Statistical analyses were conducted by unpaired t-test in Microsoft Excel for single comparisons and repeated 2-way ANOVA with multiple comparison test using GraphPad Prism8 software (GraphPad) with a 95% confidence limit. In addition to constructing a Q-Q plot, we used the D'Agostino-Pearson normality test to assess the normality of distribution. Data are indicated as mean ± SEM. All data visualization was conducted using GraphPad Prism8 software.