A crucial role for the cortical amygdala in shaping social encounters – Nature

Mice

Wild-type Swiss-Webster (SW) mice (male and female, 12–15 weeks; Charles River) and Esr1-Cre mice (017911, B6N.129S6(Cg)-Esr1^{tm1.1(cre)And}/J; Jackson Laboratory) were crossed with SW females to obtain F₁ mice used as experimental mice. Intruders for the RI test were 8–12-week-old male or female C57/BL6J mice (Jackson Laboratory). All delivered mice were allowed one week of acclimation to the housing facility before any experimental protocol. At four weeks, mice were separated from their littermates and paired with a member of the opposite sex for sexual experience for two days at eight weeks of age. Females were paired with castrated SW males (8–12 weeks; Charles River) to prevent pregnancy. All mice were maintained on a 12 h–12 h light–dark cycle (07:00–19:00) with ad libitum access to food and water. Housing and experimental rooms were maintained at 20–22 °C and 40–60% humidity. Experiments were performed during the light phase. Animals were randomly assigned to groups, arena or treatments, and data analysis was done in a double-blind fashion. Sample sizes were chosen according to experience and historical studies. Procedures were performed in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals and were approved by the Icahn School of Medicine at Mount Sinai Institutional Animal Care and Use Committee.

RI test

Mice were screened using protocols adapted from previous studies^3,41,42. In brief, cage tops were removed and replaced with plexiglass covers to monitor trials. A novel C57BL/6J mouse matching the sex of the resident was introduced into each cage and mice were allowed to freely interact for 5 min. After this, intruder mice were returned to their home cages and, in the case of female RI trials, cohabiting male mice were returned to their home cages. For the male mating assay, SW female intruders were pre-screened for sexual receptivity. All assays lasted 10 min and intruder mice were returned to their home cages at the end of the session. All videos were recorded for later analysis. Resident behaviours were manually annotated using Observer XT 11.5 (Noldus Interactive Technologies).

Aggression self-administration test

Aggression self-administration testing was performed as previously described. In brief, SW resident mice were placed in standard Med Associates operant chambers and underwent eight trials per day for 16 days in which they could press a lever (FR1) to receive access to a subordinate same-sex mouse through a guillotine door into their operant chamber. A house light illuminated the chambers during trials, and an inactive lever was extended at all times. At the beginning of a trial, the house light was turned on, and a lever was extended 10 s later. If the resident pressed for a reward, a 2-s tone played and a guillotine door next to the active lever opened for 12 s. A same-sex C57BL/6J intruder manually pushed into the chamber, and the intruder was removed at the end of the trial. If residents did not press, the active lever retracted after 60 s. All chambers also had recessed food pellet receptacles with beam–break registered entry ports. We performed all progressive ratio (PR) tests using the same parameters as those in self-administration training, except for the trial design. After reward obtaining (number of required active lever presses for a given trial), the automatic guillotine door was opened for 10 s and the same-sex intruder (C57BL/6J) was manually pushed inside the main operant chamber. Immediately after the automatic door closed, the active–inactive levers were re-extended. The intruder mouse was removed from the chamber after an attack or after 30 s had elapsed. The PR session was terminated if no reinforced responses occurred after a total duration of 20 min had elapsed. Throughout the sessions, for each trial, the number of responses required to activate the guillotine door and gain access to the intruder was incremented following an exponential progression (R = (5 × e0.12P) − 5, where P is the previous ratio). The breakpoint value corresponds to the total number of rewards (number access to the intruder during the whole session). Esr1-cre mice transfected in the COApl with AAV9-DIO-hM4Di-mCherry or AAV9-DIO-mCherry (see below for titre, surgery information and coordinates) were administered either saline or CNO, counterbalanced over the two days of the PR. Acquisition of self-administration was monitored in one of two cohorts. The data in Fig. 3j are from cohort 2.

Hidden food test

Mice were habituated to and allowed to consume a palatable food (a Reese’s mini peanut butter cup) two days before being tested. Mice were then food-restricted the night before the test. Mice were tested under red light and placed in a clean cage. The food was placed 8 cm deep in the bedding of the cage and the time to find the food was recorded with a stopwatch.

Sex discrimination test

Male and female C57BL/6J mice were used as target mice and placed on opposite sides of an open-field enclosure in mesh cups. Test mice were placed in the open field and allowed to explore the arena for five minutes under red light. Mice were recorded with an overhead Ethovision camera and the time spent interacting with each mouse was scored offline manually. To calculate the discrimination ratio, the total time spent interacting with both mice was divided by the difference between the time spent with the female and male. A positive discrimination ratio indicates more time spent with the female, and a negative ratio indicates more time spent with the male.

Perfusion and brain-tissue processing

For iDISCO+, mice were injected with 10% chloral hydrate and perfused transcardially with ice-cold 1× phosphate-buffered saline (PBS) (pH 7.4), followed by fixation with cold 4% paraformaldehyde in 1× PBS. Brains were post-fixed for 12 h in the same fixative at 4 °C. For FISH, brains were rapidly removed and flash-frozen in −30 °C isopentane for 5–10 s, then kept at −80 °C until sectioning. Slices were sectioned at 15 μm using a cryostat (Leica). Mice injected with H129ΔTK-TT were perfused 48 or 70 h after injection.

Histology and imaging

For FISH, RNAscope Multiplex Fluorescent Kits (Advanced Cell Diagnostics) were used according to the manufacturer’s instructions. Fresh-frozen brains were mounted on slides at a thickness of 15 μm, fixed for 15 min in cold 4% PFA and dehydrated serially with 50%, 70% and 100% EtOH/H₂O for 2 min each, followed by 20 min Protease IV (RNAscope) treatment. Proprietary probes (Advanced Cell Diagnostics) for Fos (316921, accession no. NM_010234.2), Vglut1 (also known as Slc17a7) (SLC17A7-C3, accession no. NM_182993.2) and Esr1 (478201-C2, accession no. NM_007956.5) were hybridized at 40 °C for 2 h, then subjected to a series of amplification steps at 40 °C (1-FL, 30 min; 2-FL, 15 min; 3-FL, 30 min; 4-FL, 15 min). Reagent Alt-A was used for the fourth amplification step, with channel 1 at 488 nm, channel 2 at 550 nm and channel 3 at 647 nm. Finally, slides were treated for 30 s with DAPI and immediately coverslipped with Eco-Mount. All slices were imaged using a Zeiss LSM 780 confocal microscope. Cells from all ISH images (two per sample) were counted blindly across groups at 20×. Cells with at least five puncta for each probe were considered to be positive for the probe of interest. Cell nuclei were counted by creating a 16-bit greyscale image of the DAPI channel. A threshold was created and nuclei were counted using the ‘analyze particles’ plug-in. Two sections per sample were analysed for quantification of mRuby⁺ puncta at 40× using the ‘analyze particles’ function. Each section contained three or four optical sections at 4-μm intervals. Starter cells from two sections were manually counted.

iDISCO+ staining, imaging and ClearMap analysis

The iDISCO+ staining protocol was modified from http://www.idisco.info (ref. ⁴³). Fixed whole brains were incubated with the primary FOS antibody (226003, 1:1,000, Synaptic Systems) and secondary donkey anti-rabbit IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor 647 (A-31573, 1:1,000, Thermo Fisher Scientific) for seven days each. A LaVision light-sheet microscope with zoom body was used for half-brain sagittal scanning, with dynamic focus and a step size of 4 μm. Cleared brains were processed as previously described using ClearMap¹³. FOS⁺ cells were quantified using the cell detection module, with cell detection parameters optimized and validated according to the intensity and shape parameters of the signal. The autofluorescence channel was aligned to the Allen Institute’s Common Coordinate Framework using the Elastix toolbox. To compare cell counts between AGGs and NONs in both sexes and hM4Di versus mCherry, a negative binomial regression was applied using the glm.nb function from the MASS package in R. Group classifications were dummy coded (0 for the AGG group and 1 for the NON group). The maximum-likelihood coefficients α and β were determined through iterative reweighted least squares. A significant β means that group status is related to cell count number at the specified region of interest. z-scores correspond to this β coefficient, normalized by its sample standard deviation. P values were corrected for multiple comparisons using the Benjamini–Hochberg procedure to decrease the false discovery rate. q values below 0.05 were considered significant. For network analysis, the data were expanded to 480 brain regions. For glm analysis, the data were collapsed to 271 regions.

Stereotaxic surgery and viral-gene transfer

Esr1-Cre mice (8–10 weeks old) were anaesthetized with an intraperitoneal injection of ketamine HCl (100 mg kg^–1) and xylazine (10 mg kg^–1) and positioned on a stereotaxic instrument (David Kopf Instruments). In the lateral portion of the COAp (COApl) (bregma: AP −1.7 mm; ML ±2.8 mm; DV −5.9 mm), 0.3 μl of virus was bilaterally infused using 33-gauge Hamilton needles over 5 min, with needles left in place for 5 min after injection. For virus delivery, 0.3 μl of AAV8-hSyn-DIO-hM3D-mCherry (2.0 × 10¹² vg ml^–1, no. 44361-AAV8, Addgene), AAV9-hSyn-DIO-hM4D-mCherry (2.0 × 10¹² vg ml^–1, no. 44362-AAV9, Addgene) or AAV9-hSyn-DIO-mCherry (2.0 × 10¹² vg ml^–1, no. 50459-AAV9, Addgene) was injected into the COApl. We also injected 0.3 μl AAV9-hSyn-DIO-hM4D-mCherry (2.0 × 10¹² vg ml^–1, no. 44362-AAV9, Addgene) into the COApm (bregma: AP −2.4 mm; ML ±2.0 mm; DV −5.6 mm). For anterograde tracing, 0.3 μl of AAV1-hSyn-FLEx-mGFP-2A-Synaptophysin-mRuby (3 × 10¹² vg ml^–1, no. 7161-AAV1, Addgene) or 0.2 μl of H129ΔTK-TT (4.0 × 10⁹ vg ml^–1, Center for Neuroanatomy with Neurotropic Viruses) was injected unilaterally into the COApl. For optogenetic manipulations, AAV9-Ef1a-DIO eNpHR3.0-YFP (3.0 × 10¹² vg mL^–1, no. 26966-AAV9, Addgene) or AAV9-EF1a-DIO-YFP (3.0 × 10¹² vg ml^–1, no. 20298-AAV9, Addgene) was injected into the COApl. All AAV injections were performed three to four weeks before behavioural experiments. For optogenetic (NpHR) and FP experiments, cannulae (NpHR: MFC_200/240-0.22_3mm_MF1.25_FLT; FP: MFC_200/250-0.57_3mm_MF1.25_FLT) were implanted at the same time as viral delivery (for COApl local, fibres were implanted 0.2 mm above the injection site). For optogenetic (eNpHR3.0) experiments of COApl terminal stimulation, cannulae (MFC_200/240-0.22_MF1.25_FLT, 6 mm for VMHvl, 5 mm for CEA) were bilaterally implanted into the VMHvl (from bregma: AP −1.7 mm; ML ±2.5 mm; DV −5.5 mm, 20° angle), or CEA (from bregma: AP −1.5 mm; ML ±2.5 mm; DV −4 mm, 0° angle). For secure fixture of the optic fibre, dental cement (Grip cement; Dentsply) was added to the skull and around the fibres.

DREADD manipulation

CNO (1 mg kg^–1, Tocris) was given intraperitoneally 30 min before the RI test, open field, aggression self-administration, hidden food and sex discrimination tests.

Optogenetics manipulation

For yellow (eNpHR3.0) light stimulation, optical fibres (BFP(2)_200/220/900-0.22_4m_FCM-2xMF1.25, Doric Lenses) were connected to a 589-nm yellow laser diode (no. MGL-III-589-50mW, Opto Engine LLC) using a patch cord with a FC/PC adaptor (no. MFP_200/240/900-0.22_4m_FC-MF1.25, Doric Lenses). A function generator (no. 33220A, Agilent Technologies) was used to generate constant yellow light for eNpHR3.0 experiments during the five-minute RI test. For all optogenetics tests, experimental mice were habituated to patch cords for two days before testing in RI. For RI experiments, mice were tested over two days, counterbalanced under laser-on and laser-off conditions For closed-loop behavioural experiments, the laser was turned on for 10 s as the resident initiated approach of the intruder for every other approach bout in a counterbalanced manner. Attack behaviour was quantified during the period when the light was on. For mating experiments, the light was left on for the duration of the experiment.

Fibre photometry

Fibre photometry was performed according to the Neurophotometrics manual and published protocols⁵. A fibre-optic patch cord was attached to the implanted cannula with cubic zirconia sleeves covered with black tubing. The opposite end of the cable was coupled to a Neurophotometrics LED port. The open-source Bonsai programme was used to control the system; 470-nm and 415-nm LED lights were used for GCaMP6s signal and autofluorescence measurement. Light at the fibre tip ranged from 40 μW to 80 μW and was constant across trials over testing days. Simultaneous recording of 40 fps from both the 470-nm and the 415-nm channel was achieved phase-to-phase and visualized using Bonsai. Three weeks after virus injection and ferrule implantation, mice were tested in the RI test and were presented with different odours in a counterbalanced manner. Once connected to the apparatus, mice were allowed to rest and habituate for 3–5 min before starting. For the RI test and odour presentation, we recorded Ca²⁺ fluorescence during 2 min of baseline activity without an intruder, followed by 5 min of intruder or odour exposure. The 415-nm channel served as the control channel and was subtracted from the GCaMP6s channel to eliminate autofluorescence, bleaching and motion effects. Change in fluorescence (ΔF/F) was computed by subtracting the average value during the final minute of baseline recording and then dividing the resulting value by the averaged value during the final minute of baseline. This value was then z-scored by subtracting the average ΔF/F from each value and dividing by the standard deviation. Behavioural data were manually annotated in Ethovision and time stamps were aligned with fluorescence recording.

LFP recordings

Esr1-Cre × SW male mice (around three to four months old) were anaesthetized using a mix of ketamine and xylazine, and placed in a stereotaxic apparatus (David Kopf). After exposing the skull, custom-made platinum–iridium depth electrodes (Pinnacle Technology) were implanted in the CoApl (from bregma: AP −1.7 mm; ML ±2.8 mm; DV −5.7 mm) in the VMH (from bregma: AP −1.7 mm; ML ±0.7 mm; DV −5.8 mm) Before implantation, the electrodes were coated with Vybrant Dil cell-labelling solution (Invitrogen) to later check for the placement of the electrode. Two skull screws (0.25 cm, Pinnacle Technology) were implanted over the left and right cerebellum to serve as ground and common reference for recording LFPs. Electrodes and reference screws were connected to a 6-pin connector for mice (Pinnacle Technology) and sealed using dental cement (Metabond). After surgery, mice recovered for at least one week in standard housing conditions before behavioural testing. All recordings were performed in the mouse home cage and the behavioural sessions were videotaped with an overhead camera connected to Ethovision 15 (Noldus). A transistor–transistor logic signal was sent through Ethovision to the LFP acquisition software to synchronize behavioural and neurophysiological data. LFP signals were captured using a tethered data acquisition system (Model 8401HS, Pinnacle Technology) and acquired using Sirenia Acquisition software (Pinnacle Technology). Data were acquired at 2 kHz using a Pinnacle Technology preamplifier (gain ×100) and with high-pass (0.5 Hz) and low-pass (200 Hz) filters. Signals were segmented into single trial epochs aligned to stimulus-onset.

LFP analysis

Raw data files were loaded into R and analysed with the PSD package⁴⁴. The signal was low-pass-filtered up to 200 Hz using a Butterworth filter before analyses. Signal from the first two seconds of an attack or investigation bout that did not precede an attack was selected for analysis. We chose the first two seconds because all behaviours of interest lasted at least two seconds. A fast Fourier transform with multiple sine tapers⁴⁵ was performed on each segment. Data were then bandpass-filtered for the delta (1–4 Hz), theta (5–12 Hz) and gamma (40–100 Hz) frequency bands. Coherency (defined as the cross-spectral density divided by the auto-spectral density) values were computed from 0 Hz to 100 Hz using the pspectrum function for each event and averaged across all events for each value. Coherency values from a one-minute baseline period were subtracted from the coherency values obtained during social behaviours.

Ex vivo electrophysiology

For recording optically evoked excitatory postsynaptic currents (oEPSCs), AAV9-EF1a-DIO-ChR2-eYFP (0.5 μl, 3.0 × 10¹² vg ml^–1, Addgene, 35509-AAV9) was injected unilaterally into the COApl of eight-week-old male Esr1-Cre mice. At five to eight weeks after injection, coronal brain slices of the VMH and CEA were prepared using a Compresstome (no. VF-210-0Z, Precisionary Instruments) in cold (0–4 °C) sucrose-based artificial cerebrospinal fluid (aCSF) containing 87 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 4 mM MgCl₂, 23 mM NaHCO₃, 75 mM sucrose and 25 mM glucose. Slices were transferred to a recording chamber continuously perfused at 2–3 ml min^–1 with oxygenated aCSF. Patch pipettes (4–7 MΩ) were pulled from thin-walled borosilicate glass using a micropipette puller (P-97, Sutter Instruments) and filled with a potassium gluconate (KGlu)-based intrapipette solution containing 116 mM KGlu, 20 mM HEPES, 0.5 mM EGTA, 6 mM KCl, 2 mM NaCl, 4 mM ATP, 0.3 mM GTP and 2 mg ml^–1 biocytin (pH adjusted to 7.2). Cells were visualized using an upright microscope with an IR-DIC lens and illuminated with a white light source (Scientifica). Illumination with a 470-nm LED (pE-300ultra, CoolLED) through the microscope objective was used for visualizing eYFP⁺ cells. VMH and CEA neurons were recorded in voltage-clamp mode. COApl^Esr1 terminals were stimulated through the microscope ×40 objective (5 ms per pulse, 470 nm; pE-300ultra, CoolLED). oEPSCs were recorded at −70 mV in the presence of tetrodotoxin (TTX, 1 μM, Tocris) and 4-aminopyridine (100 μM, Tocris) to probe monosynaptic effects. oEPSCs were blocked by bath application of NBQX (no. SR-95531, 5 μM, Tocris) confirming the glutamatergic nature of the synaptic contact. Whole-cell recordings were performed using a patch-clamp amplifier (Axoclamp 200B, Molecular Devices) connected to a Digidata 1550 LowNoise acquisition system (Molecular Devices). Signals were low-pass-filtered (Bessel, 2 kHz) and collected at 10 kHz using the data acquisition software pCLAMP 11 (Molecular Devices). Electrophysiological recordings were extracted and analysed using Clampfit (Molecular Devices).

Weighted correlation network analysis

For network construction, we followed previously published guidelines⁴⁶. In brief, an n × n adjacency matrix was created that encodes the connection strengths between pairs of nodes (brain regions), using the power adjacency function: A_ij = |s_ij|^β. A_ij refers to the adjacency matrix and s_ij refers to the correlation between regions i and j raised to the power of β. The correlation is raised to the power of β to reduce the influence of potentially specious correlations. To measure the similarity between the nodes for clustering, the topological overlap measure (TOM) was used: \({{\omega }}_{i,j}=\frac{{l}_{ij}\,+\,{a}_{ij}}{\min \{{k}_{i,}{k}_{j}\}+1-{a}_{ij}}\). ω_i,j refers to the TOM matrix, \({l}_{ij}={\sum }_{u}{a}_{iu}{a}_{uj}\) and a_ij = |s_i|^β. In words, this is the sum of the product of the shared connections between regions i and j plus the direct connection between regions i and j. This value is divided by the denominator to achieve a value between 0 and 1. To compare the expression of these modules between sexes within a network, a singular value decomposition (SVD) was computed on each module. The SVD is defined as: \({X}^{n}=U\times S\times {V}^{{\rm{T}}}\). Xⁿ is the TOM matrix for the nth module in the network, U and V^T are matrices of orthonormal vectors and S is a diagonal matrix of eigenvalues that denote how much variation is accounted for by each column of U and V. To determine which modules differed between the AGG and NON networks, or whether the mCherry and hM4Di networks conserved module connectivity and density, we used the module preservation function. For comparisons between networks we focused on the following measures for connectivity: Z.cor KIM, Z.cor.KME, Z.cor.KMEall, Z.cor.cor and Z.cor.MAR. For density measures, we focused on Z.propVarExplained, Z.meanSignAwareCorDat, Z.meanAdj and Zmean.MAR. All z values were derived by randomly permuting the module labels in the test network and calculating the corresponding preservation metric. The average value of each statistic from the permutations is subtracted from the observed statistic and divided by the standard deviation of the statistic from the permutations. For a detailed explanation of the preservation measures, see Supplementary Table 2. All of the above steps were performed using the WGCNA package in R.

Statistical analysis

All statistical tests and associated information are reported in the figure legends. All t-tests, one- and two-way ANOVAs were performed using GraphPad Prism software. Two-way ANOVA analysis was followed by Šídák’s or Tukey’s multiple-comparisons test for post-hoc analysis. One-way mixed-effects ANOVA was followed by Tukey’s post-hoc test. For comparisons between groups for region by region iDISCO+ analysis, P values were corrected for multiple comparisons using the Benjamini–Hochberg procedure to reduce the false discovery rate. q values below 0.05 were considered significant.