C57BL/6J wild-type female and male mice of 8-18 weeks were used for this study (Janvier lab, France). Mice were housed in groups of three or four per cage with water and food ad libitum on a 12:12 h light cycle (lights on at 7 a.m.) in individually ventilated cages (IVC, Innovive, France). Stimulus pups (2-6 days old) were from donor C57BL/6J pairs. All procedures were conducted in compliance with the Swiss National Institutional Guidelines on animal experimentation and were approved by the canton of Vaud Cantonal Veterinary Office Committee for Animal Experimentation (Switzerland; License VD3171).
rAAV-DJ/8/2-hSyn1-eGFP-WPRE (titer, 9.4 x 1012 vg.ml-1), rAAV-DJ/8/2-hSyn1-GCaMp6f-WPRE (titer: titer: 4.5 x 1012 vg.ml-1), rAAV8-hsyn-hChR2(H134R)-EYFP (titer: 6.4 x 1012 vg.ml-1), rAAV1/2-hCMV-chl-Cre (titer: 1 × 1013 vg.ml-1), rAAV-DJ/8/2-hSyn1-dlox-EGFP(rev)-dlox (titer: 6.4 x 1012 vg.ml-1), rAAV-DJ/8/2-hEF1α-dlox-GCaMP6f-lox-WPRE (titer: 4.5 x 1012 vg.ml-1), ssAAV-DJ/2-shortCAG-loxFAS-GCaMP6s-loxFAS-WP1RE-hGH (titer: 4.5 x 1012 vg.ml-1) were purchase from the UZH Vector Facility (Zurich, Switzerland). rAAV8-hSyn1-JAWS-GFP (titer: 1.3 x 1013 vg.ml-1) was purchased from Addgene. rAAV2.5-CAG-Flex-EGFP-KASH (titer 1.045 x 1013 vg.ml-1) aliquots were gently provided by Magdalena Gotz and Riccardo Bocchi (Biomedical Center, Ludwig-Maximilians-Universitatet).
Mice were anaesthetized with ketamine (150 mg/kg)/xylazine (10 mg/kg) (Cantonal University Hospital, Lausanne, Switzerland). The ocular protector Viscotear was used to prevent eye damage. The surgery was performed on a heating pad to keep a stable body temperature. When the mouse was deeply asleep, we performed a local anesthesia (subcutaneous injection) with a mix of lidocaïne (6mg/kg) and bupivacaine (2.5mg/kg). We then unilaterally injected in the LHb (−1.4 mm AP, 0.45 ML, 3.1 mm DV) GCaMP6f or flex-GCAMP6f or creoff-GCAMP6s using a glass pipette on a stereotactic frame (Kopf, France). For optogenetic and cell counting experiments, we bilaterally injected in the LHb the inhibitory opsin JAWS or eGFP virus or a flex eGFP or a eGFP KASH. For these injections volumes ranged between 150 and 250 nl. An anterograde transynaptic AAV1ht-CRE virus or JAWS or ChR2 or eGFP were injected unilaterally or bilaterally in the BNST (+0.24 mm AP, 0.85 mm ML, 4.5 mm DV) with a volume ranging between 50 to 150 nl. All injections were performed at a rate of approximately 100-150 nl/min. The injection pipette was withdrawn from the brain 10 minutes after the infusion. Animals were allowed to recover for a minimum of two weeks before fiber or GRIN lenses implantation.
For fiber photometry experiments, a single fiber probe (200 μm, Chi Square Bioimaging) was placed and fixed (C and B Metabond, Parkell) 150 μm above the injection site in the LHb, lowered at a constant speed of 7 µm/s. For optogenetic manipulation a single fiber (200 μm, Thorlabs) was placed at the following coordinates from Bregma (AP: −1.4 mm, L: ± 0.1 mm, V: −2.6 mm from skull surface). Surgery was performed under isoflurane anesthesia (induction: 4%, maintenance: 1.8%–2%, Univentor).
For endoscope experiments, mice were anaesthetized (as described above) and implanted with a GRIN lens (6.1mm length, 0.5mm diameter; Inscopix, #100-000588). The lens was placed ∼150–200 μm above the injection site using the following coordinates: −1.40 mm posterior to bregma, 0.45 mm lateral from midline, and −2.85 to −2.9 mm ventral to skull surface (lowered at a speed of 1μm/s). For pain management, paracetamol (500mg/250ml; 200-300mg/kg/day) was added to the drinking water after the surgery. Two weeks after lens implantation, mice were again anaesthetized (isoflurane, as above) and a baseplate (Inscopix, #100- 000279) was secured above the lens. A doom micro-camera (Inscopix) was attached to the baseplate allowing efficient habituation of the mice to the micro-endoscope weight. The latter procedure ensure that mice display comparable behaviour with no implanted pairs during the recording session. Proper viral expression and fiber/GRIN lens placement in brain areas of interest were confirmed post hoc using histology for all experiments.
Histology, immunohistochemistry, and microscopy
For immunohistochemistry, mice were terminally anaesthetized with ketamine and xylazine and perfused transcardially with paraformaldehyde (PFA) 4% in 0.1 M phosphate buffered saline (PBS). Brains were collected and left overnight in 4% PFA at 4°C until slicing. Consecutive coronal slices (60 microns) of LHb and BNST were sectioned using a vibratome (Leica VT1200S). LHb slices were stained for NeuN, a neuron-specific nuclear marker, using a classical immunohistochemical protocol. Prior to immunostaining, slices were incubated in blocking buffer (10% Normal Goat Serum (NGS), 0.25% Triton in PBS) for 1 hour. All antibodies were diluted in PBS with 0.25% Triton and 3% NGS. Neurons were stained with a mouse anti-NeuN primary antibody, (Millipore, dilution: 1/500) incubated 48h at 4°C. Following extensive rinses with PBS, slices were left overnight at 4°C with a goat anti-mouse secondary antibody coupled with the fluorescent protein Alexa 555 (Invitrogen, dilution 1/500). Slices were mounted on glass slides with FluorSave reagent. Images of the LHb (6 slices/mouse) were acquired using a confocal microscope (TCS SP5 AOBS TANDEM, Leica) with a 20X objective using the same acquisition parameters between mice. For the cell counting, we selected for each mouse (n=4) 6 slices (60microns) containing the LHb, at 3 different coordinates in the antero-posterior axis (AP: -1.3, -1.4, -1.5 mm from Bregma). For each slice, the LHb was divided into two sub regions, the lateral and the medial LHb. The number of eGFP and NeuN-positive neurons was counted using ImageJ Cell Counter plugin. We then plotted the ratio of GFP positive cells on the total cells (NeuN positive cells).
To validate injection site in the BNST, we performed a staining for the Cre recombinase protein, using a similar protocol mentioned earlier or we co-inject fluorescent beads dye. Slices were incubated in the mouse anti-Cre primary antibobody (Millipore, dilution 1/500) 24h at 4°C, and in the goat anti-mouse secondary antibody coupled with the fluorescent protein Alexa 555 (Invitrogen, dilution 1/500) 2h at room temperature. We took images of the BNST slices with an epi-fluorescent microscope (Zeiss).
Pup retrieval. Animals were placed in a phenotyper box (Noldus, 58cm H, 30cm L, 30cm W). In addition of the top view camera provided by the phenotyper, an extra phenotyper lid with the embedded camera was placed on the side of the arena ensuring a close video-recording of the behaviour with a lateral point of view. The arena was filled with standard wood chip bedding (Safe, Germany) and divided in 2 main zones (see Fig. 1a): a nest zone provided by nest material (paper Kleenex) and a pup zone (in the opposite corner) where the pup was placed through an external sliding door. Video recording was controlled with EthovisionXT 15 (Noldus Information Technology).
Mice underwent a period of habituation to the arena of 15-20 min every day for 2 consecutive days. For the experiment, after 5-10 min of acclimatization, we placed the pup in the pup zone until it was retrieved to the nest (10 min maximum). If the pup was not retrieved after the 10 min, it was replaced by another pup. If it was retrieved within the 10 min, another pup was immediately placed in the pup zone.
Pup calls recording and pup motion tracking. Two microphones (UltraVox XT, Noldus) allowing ultrasound vocalization (USVs) recordings were placed 16 cm above the pup zone and the nest zone, respectively. A pup was placed in the pup zone and the recording was acquired for 5 minutes. After, a virgin female (previously selected for its high retrieval score) was introduced to the arena and retrieved the pup to the nest (pups were retrieved within 2 minutes). Ten minutes later, the female was gently removed from the phenotyper and USVs were recorded through the microphone placed above the nest zone. The number of USVs call (40-90 KHz) was analyzed with the Ultravox CTsound software and plotted for the same pup in the 2 different conditions. The order of the procedure was inverted for 4 pups. Pup movements were tracked using Ethovision XT15 as follows: the pup was segmented from the background using Ethovision’s standard detection method (Dynamic Subtraction), the resulting area in pixel and how it changes across frames (threshold: change > 60% of the area) was used to estimate body movement and crawling.
Conspecific, object exposure and airpuff. Females employed for the photometry and endoscope experiments were also exposed to sessions (5 minutes each) with a conspecific (male C57BL6/J of 4 weeks old), an object (a falcon tube cap) or an unpredicted series of airpuffs (5, 500ms duration, 1.5 bar, randomly presented in 5 min).
Video recording was synchronized with the photometric and endoscopic aquisition using hardware-time signals controlled with a I/O box (Noldus).
Automatic behavioural scoring was performed with Ethovision XT 15. For each mouse, speed, time spent in the zones, and mobility (displacement of the body center point > 2 cm/s) were collected from the video file. Manual behavioural annotation was performed on a frame-by-frame basis using the manual scoring module of EthovisionXT 15 or the video function of Spike2 (Cambridge Electronic Design). Grooming was defined as either a crouching posture over the pup together with head movements indicating licking of the pup, or olfactory inspections of the pup together with the forepaws of the experimental mouse being in touch with the pup.
Grooming score was quantified based on the time the female spent grooming the pup compared to the time spent in the pup zone in presence of the pup (express in percentage). Nest building was defined as nesting material collection and placement in order to reduce its dispersion across the arena. We calculate the time spent nesting for each mouse and we normalize it based on the total duration of the session. The quality of the nest was assessed at the end of the session following previously published score scale. Briefly, a score of 20% was given when the shredded paper remained scattered throughout the cage; 40% was assessed when some of the material was constructed into a nest; 60% when a noticeable nest was constructed, but several pieces were still scattered; 80% when almost all the material was used for the nest and 100% when all material was used to make an identifiable and organized nest. The nesting score was calculated as the average of the time spent nesting and the quality of the nest.
Retrieval episodes were split in 4 subsequent phases: Approach was defined as the first step toward the pup from the point farthest away from the pup. Interaction was defined as close contact to any parts of the body of the pup by the frontal end of the female. Retrieval onset was defined as the moment the female opened her jaw and made contact with the skin of the pup. Retrieval offset was scored as the moment when the female dropped the pup in the nest. Object interaction was defined as nose contact with any part of the object. Conspecific interaction was defined as close contact with any part of the conspecific. Regarding photometry and endoscope recordings, for object, conspecific, pup interaction (without retrieval), nest building and grooming only events separated by at least 10 seconds were included in the analysis. Most behavioural annotation was not done blindly. For a subset of videos, we compared the annotations done by an annotator blind to the experimental conditions and one that was not, founding high consistency.
Fiber photometry recordings
Fiber photometry measurements were carried out by the ChiSquare X2-200 system (ChiSquare Biomaging, Brookline, MA). Briefly, blue light from a 473-nm picosecond-pulsed laser (at 50 MHz; pulse width ∼80 ps FWHM) was delivered via a single mode fiber. Fluorescence emission from the tissue was collected by a multimode fiber with a sample frequency of 100Hz. The single mode and multimode fibers were arranged side by side in a ferrule that is connected to a detachable multimode fiber implant. The emitted photons collected through the multimode fiber pass through a bandpass filter (FF01-550/88, Semrock) to a single-photon detector. Photons were recorded by the time-correlated single photon counting (TCSPC) module (SPC-130EM, Becker and Hickl, GmbH, Berlin, Germany) in the ChiSquare X2-200 system.
Microendoscopic calcium imaging
All calcium imaging was recorded at 20 frames per second, 200-ms exposure time, and 10%–40% LED power (0.4-0.9mW at the objective, 475nm) using a miniature microscope from Inscopix (nVista). Calcium recording files were down-sampled (spatial binning factor of 4) to reduce processing time and file size, filtered, corrected for rigid brain movement and the ΔF/F0 was calculated using as F0 the average fluorescence for all the video (Inscopix, IDP). Individual component analysis and principal component analysis (ICA/PCA) applications were used to identify individual cells and to extract their respective calcium traces. In addition, to compare ROI detections and relative traces obtained with the PCA/ICA we also performed constrained non-negative matrix factorization for endoscopic data (CNMF-E) for a subset of data. Briefly, we denoised, deconvolved, and demixed calcium-imaging dynamics (https://www.github.com/zhoupc/cnmf_e). Calcium imaging frames were initially pre-processed in Mosaic (Inscopix) for motion correction. We use a Gaussian kernel width 4 μm, maximum soma diameter 16 μm, minimum local correlation 0.8, minimum peak-to-noise ratio 8 and merging threshold was set to 0.65 for optimal discrimination of temporal and spatial overlap.
To determine the putative cell location in the LHb we proceed as follow: we registered each field of view (FOV) of GCaMP6f-expressing neurons and aligned it to the GRIN lens borders visualized on post-hoc fixed brain sections (Fig. 2g and Extended Data Fig. 3a and 4a). We then annotated the active neurons within individual FOVs to build cell masks, and subsequently we overlapped them, GRIN lenses borders and LHb anatomy from all mice.
Analysis of calcium signal
Photometric signal as well as miniscope PCA/ICA derived traces were smoothed (constant time factor, 0.1 s) and further processed according to the trials using Spike2 software (Cambridge Electronic Design). We obtained an average peri-event time histogram (PETH) trace aligned to the events (2-7 s prior and 2-7 s after a given event). For the pup retrieval phases we z-scored each trials in reference to their baseline (7 s prior to the approach). For nesting, grooming events we used as reference the average and SD of the relative recording/movie. For conspecific, object and airpuff the z-score was calculated in relation to the 2s prior of the event.
Photometry: The duration of the fluorescent transient (Fig 1e) was calculated as follow: first we bin the z-score (10 Hz) around the retrieval episode (-10 to + 40 seconds). Then we consider the transient starting at the first bin higher than 2-score and terminated when we encounter more than 5 consecutive bins not fitting the criteria.
Single trial quantification was obtained, by the z-score of the 0.5 first seconds after the events except the retrieval offset (we plotted the last 0.5 s cause the slow decay of the GCaMP6).
Endoscope: for the single cell reliability analysis, we analyzed the fluorescence Ca2+ signals of individual trials after a given event, using 1 s time span. We consider a cell responding during a trial if the signal was higher than 2 z-score. The reliability account for the number of trial where the cell respond divided the total number of trials (express as percentage). Average amplitude calculated at the same time-span was obtained for each cell recorded.
The response was defined as “specific” if the activity was reaching 5 z-score for a mean duration of 0.3 sec during a window of 4sec centered onto a given event (Fig 4g).
Clustering and decoding
For clustering neurons based on their average responses around all the retrieval phases, grooming and nest building events we followed a similar general procedure as in Lecca et al. (2020). Briefly, we first calculated the average perievent time histogram (PETH) for each neuron around each action by averaging all trials. Due to the variability in duration of retrieval episode, and therefore to avoid signal contamination from subsequent phases, we considered the average signal in a time window from 0.0 s to 1.0 s for each episode. These time windows were treated as features of the response of a neuron. This feature space was then reduced in dimensionality using principal components analysis. The number of principal components to keep was decided based on the bend in the scree plot. A spectral clustering algorithm along with optimal selection of number of clusters using silhouette scores was used on the principal component scores to test for presence of clusters. The number of clusters was chosen by maximizing the silhouette score. Once cluster identities were assigned, all PETHs were recalculated using the activity from −7 s to +2 s for approach, -2 s to +2 s for interaction and retrieval onset, -2 s to +7 for retrieval offset and from -5 s to +5 s for nest building and grooming behaviours.
Decoding approach was used to create a model and attempt to reveal whether particular information is representative in calcium neuronal activity to predict category labels of behavioural events scored previously. A shuffle category was added by randomly permuting calcium neuronal activity from equally distributed number of trials of the different category of behavioural events. Based on PETHs of calcium activity from recorded neurons around a time window from -2sec to +2sec surrounding each event, multiple features were analyzed. To obtain these features, the probability distribution of amplitude of activity and duration (expressed in ratio of total time) of significant increased activity (>1.96 z-score) were computed and the following properties of each distribution were extracted: mean, median, coefficient of variation, skewness and kurtosis. The number of time bout as well as the total duration in significant increased activity was also computed. The features were normalized by using z-score and by rescaling the values. Then the dimensionality was reduced by using UMAP technique for visualization or a classifier was built by using k-nearest neighbor approach based on BNST-receiving LHb calcium neuronal activity with euclidean distance metric and a 5-fold cross-validation. By using this classifier model, the posterior probability that the neuronal activity is associated with each category of scored behaviour was then calculated. The decoding performance for each neuron during each behaviour was normalized by dividing it with the average posterior probability of the cluster 1 LHb neurons previously defined for corresponding behavioural category.
In vitro recordings
Virgin female mice were anaesthetized with ketamine (150 mg.kg-1) and xylazine (100 mg.kg-1) (Veterinary office University of Lausanne) and brains were rapidly extracted. Coronal slices (250 micrometers) of LHb and BNST were sectioned using a vibratome (Campden instruments), while immersed in an ice-cold solution, bubbled with 95% O2 and 5% CO2 and containing (in mM): choline chloride (110), glucose (25), NaHCO3 (25), MgCl2 (7), ascorbic acid (11.6), sodium pyruvate (3.1), KCl (2.5), NaH2PO4 (1.25), and CaCl2 (0.5). The low temperature and high level of magnesium limit the release of neurotransmitters and cell death. Slices were then incubated for 5 mins at 34°C in a solution of identical composition before being stored at room temperature in artificial cerebrospinal fluid (ACSF) containing (in mM): NaCl (124); NaHCO3 (26.2); glucose (11); KCl (2.5); CaCl2 (2.5); MgCl2 (1.3); and NaH2PO4 (1), for an hour prior recording. This solution allows us to preserve the integrity of the neurons throughout the day, by supplying oxygen and maintaining osmolarity and pH at biological levels.Whole cell voltage clamp recordings of LHb neurons were obtained using borosilicate glass pipettes (Phymep; impedance: 2.5-4 MΩ) filled with Cs-based intracellular solution containing (in mM): cesium methanesulfonate 120, CsCl 10, HEPES 10, EGTA 10, creatine phosphate 5; Na2ATP 4; Na3GTP 0.4. The bath solution (ACSF) was kept at 31°C with a flow rate of 2mL/min.
During the recordings, electrical signal was filtered (5 kHz) and digitized (10 kHz) using MultiClamp 200B (Molecular Devices). Data acquisition was performed with Igor Pro and NIDAQ tools (Wave Metrics). Access resistance was continuously monitored with a voltage step of −4 mV (0.1 Hz). Post synaptic currents were evoked with a LED coupled to Master-8 (AMPI) and an Olympus-BX51 microscope, delivering pulses of blue light (473 nm, 5 mW, 1-10 ms duration). Evoked excitatory and inhibitory post synaptic currents were recorded at -60mV and +5 mV, respectively. After the recordings, slices were kept overnight in PFA to verify the injection site (BNST) and the presence of fibers (LHb) using an epi-fluorescent microscope (Zeiss).
In vivo recordings
Mice, previously injected with ChR2 in the BNST were anaesthetized using isoflurane (Induction: 4%; maintenance: 1–1.5%) and placed in the stereotaxic apparatus (Kopf, Germany). Their body temperature was maintained at 36 ± 1°C using a feedback-controlled heating pad (CMA 450 Temperature controller, Phymep). The scalp was retracted, and one burr hole was drilled above the LHb (AP: −1.3 to −1.6 mm, L: 0.35–0.5 mm, V: −2.3 to −3.2 mm) for the placement of a recording optrode. Single-unit activity was recorded extracellularly using glass micropipettes filled with 2% Chicago sky blue dissolved in 0.5 M sodium acetate (impedance 5–15 MΩ). Signal was filtered (band-pass 500–5000 Hz), pre-amplified (DAM80, WPI, Germany), amplified (Neurolog System, Digitimer, UK) and displayed on a digital storage oscilloscope (OX 530, Metrix, USA). Experiments were sampled on- and offline by a computer connected to CED Power 1401 laboratory interface (Cambridge Electronic Design, Cambridge, UK) running the Spike2 software (Cambridge Electronic Design).
Single units were isolated and the spontaneous activity was recorded for a minimum of 3 min before assessing their response to BNST terminal optogenetic stimulation (15 trials, 40 Hz, 1 s, 473 nm, 8 mW). Peristimulus time histograms (PSTHs) and raster plots were built using 100 ms bin width. A cell was considered excited when the mean number of action potentials/bin in at least one of the five epochs (200 ms per epoch) after the light onset was higher than the baseline (the average number of action potentials/bin in the 2-s period before the light onset) plus two times the Standard Deviation (SD).
Inhibition was assessed if the mean count during the light period (1 s) dropped at least 35% below average baseline
Each BNST-excited cell was also tested for its response to repetitive (every 5 s) shocks (0.5 s, 1 mA) delivered to the hind paw contralateral to the recording side. PSTH was built using 10 ms bin width. Excitation or inhibition to Fs were assessed as previously reported (Congiu et al., 2019).
At the end of each experiment, mice were euthanized (overdose of isoflurane prior to killing) and the electrode placement was determined with an iontophoretic deposit of pontamine sky blue dye (1 mA, continuous current for 5 min). Brains were then rapidly removed and fixed in 4% paraformaldehyde solution. The position of the electrodes was identified with a microscope in coronal sections (60 μm). Only recordings in the correct area were considered for analysis.
After stereotaxic surgeries, the viruses were allowed to incubate for 4–7 weeks before behavioural testing. A ferrule patch cord was coupled to the ferrule fiber implanted in the subject mouse using a sleeve (Thorlabs). Optical fibers were connected to a 473 nm blue laser, (power 12 mW, Shangai Dream lasers technology) or a 638 nm red laser (power 4mW, MatchBox series, Integrated Optics, US). For the ChR2 experiment, the laser was remotely activated (473 nm, 40 Hz, 10 ms pulse duration) by the experimenter at the onset of interaction and turn off at the exit of the pup zone. This experiment was a “within animal” design: we alternate trial with light-paired with pup interaction in the pup zone with period (max 5 minutes) where the light was always off. For the JAWS experiments, for each trial the laser was activated (638 nm, continuously) and inactivated via the I/O box each time the females entered or quit the pup zone, respectively.
Nuclei preparation and FAC-sorting: Three virgin males, three virgin females and three mothers were sacrificed and brains were extracted in ice-cold oxygenated ACSF (containing (in mM): NaCl (124); NaHCO3 (26.2); glucose (11); KCl (2.5); CaCl2 (2.5); MgCl2 (1.3); and NaH2PO4 (1)). Each brain was then sectioned coronally at 1250 μm using a vibrating microtome (Campden Instruments). The lateral habenula was micro-dissected under a stereomicroscope, incubated for 5 min in 500 μl chilled 0.1X NP40 Lysis Buffer (Tris-HCL pH 7.4 10mM, NaCl 10mM, MgCL2 3mM, Tween-20 0.1%, NP40 0.1%, BSA 1%, DTT 1mM and RNase inhibitor), mechanically dissociated using a pellet pestle (15 goings and ongoings) and finally incubated on ice for 5 min. The suspension was then washed with 500 µL chilled wash buffer (Tris-HCL pH 7.4 10mM, NaCl 10mM, MgCL2 3mM, Tween-20 0.1%, BSA 1%, DTT 1mM, RNase inhibitor). Nuclei were then filtered through a 30-μm cell strainer, centrifuged (1000g, 10 minutes at 4°C) and resuspended in 500 µL of FACS buffer (Tris-HCL pH 7.4 10mM, NaCl 10mM, MgCL2 3mM, BSA 1%, DTT 1mM, RNase inhibitor). Positive GFP nuclei were sorted using a BD FACS Aria II flow cytometer (BD Biosciences) and collected in ice-cold FACS buffer.
Single-nuclei RNA capture and sequencing and quality controls: After sorting, nuclei were counted and immediately processed according to the 10X Chromium protocol. Briefly, an appropriate volume of each cell suspension containing 1000 cells from the three different conditions was combined with 10X Chromium reagent mix and samples were loaded into three separate lanes. Nuclei capture, lysis, mRNA reverse transcription, cDNA amplification and libraries were performed following 10X Genomics Chromium dual indexing Single Cell 3' V3.1 reagent kit instructions. Libraries were then multiplexed and sequenced according manufacture recommendations with paired-end reads using a HiSeq4000 platform (Illumina) with an expected depth of 250’000 reads per single nuclei. All the sequencing experiments were performed within the Genomics Core Facility of the University of Lausanne. Alignment of sequenced reads to the mouse genome (GRCm38) and filtered gene–barcode matrices were realized by running Cell Ranger Single-Cell Software Suite v5.0.1 (10X Genomics). The cell ranger count function was used to generate filtered gene/cell expression UMI corrected matrices by selecting probable nuclei and removing empty lipid droplets (219 nuclei for “virgin females” condition, 225 nuclei for “virgin males” condition and 223 nuclei for “mothers”). To filter only high-quality cells, we applied selection based on mitochondrial genes percentage (>10 %) and number of genes per cell (>500 genes). After applying these filters, 196 nuclei for “virgin females” condition, 205 nuclei for “virgin males” condition and 202 nuclei for “mothers” condition were kept for further analysis.
Single cell RNA sequencing Habenula database: For Wallace et al., dataset, count matrix was downloaded from Harvard database website https://doi.org/10.7910/DVN/2VFWF6, and metadata were kindly provided by the authors (11878 cells). For Hashikawa et al., datasets, count matrix was downloaded from GEO under accession number GSE137478 and metadata were also kindly provided by the authors (7506 cells).
Data integration and visualization: To generate reference habenula atlas, we applied data integration procedure from Seurat and identified shared sources of variation between Wallace and Hashikawa datasets. Briefly, each dataset was normalized and scale using SCTransform procedure from seurat. We then identified common features and used the “FindIntegrationAnchors” function with using for normalization method “SCT” and default parameters. We then performed the integration using the “IntegrateData” function and default parameters. For UMAP visualization, dimensionality reduction was performed using standard function in Seurat. To identify BNST-receiving LHb neurons clusters, we first adopted a graph-based clustering approach using “FindClusters” function from Seurat with a resolution of 0.4. We then use the Transfer Data function from Seurat to automatically annotate each cluster base on the reference habenula atlas annotation.
Differential expression analysis: Differentially expressed genes were identified based on their weight, resulting from differential pairwise expression analysis using Seurat “FindAllMarkers” function with default parameters (expect only.pos = TRUE, min.pct = 0.2, logfc.threshold = 0.5, and their p-value lower than 0.05). The identified gene candidates for each condition were interrogated for statistically significant gene ontologies using GSEA (http://software.broadinstitute.org/gsea/index.jsp). As background universe gene list for the Gene Ontology (GO) term analysis, we used a total of 32885 genes corresponding to all genes detected. For gene ontology enrichment, the top 20 “biological processes” or specific term enrichment such as “GABA” or “Hormone” were filtered using a false discovery rate (FDR)-corrected P value lower than 0.1 as a cutoff. Gene ontology filtered for “GABA” term analysis of gene candidates across biological processes. The string database platform (http://string-db.org) was used to determine the protein-protein interactions.
All statistical analyses were conducted using Prism (v.9, GraphPad). Statistical tests used in this study include paired and unpaired t-test, Wilcoxon matched-pairs tests, Chi Square test, one-way and two-way analysis of variance (ANOVA). When parametric tests were used, data normality was confirmed using the Shapiro–Walk normality test. P values were corrected for multiple comparisons when necessary. The bar plots show the mean ± s.e.m. In the box plots, the centre lines indicate the median, the box limits indicate the upper and lower quantiles. The significance threshold was held at α = 0.05. All behavioural, imaging and optogenetics experiments were replicated in multiple batches of animals with similar results. Sample sizes were not predetermined using statistical methods. Experiments were randomized whenever possible. Experimenters were not blind to the experimental group.