Strange IndiaStrange India


All procedures were approved under NYU School of Medicine IACUC protocols.

Data reporting

No statistical methods were used to predetermine sample size. The experiments were not randomized. The behavioural raters were blinded to allocation during co-housing and outcome assessment.

Behaviour

Co-housing

Pup-naive C57Bl/6 virgin female mice were bred and raised at NYU School of Medicine and kept isolated from dams and pups until used for these studies when approximately eight weeks old. For experiments where viral injections were performed, we first allowed two weeks for viral expression before animals were used in experiments. Dams were initially pre-screened to ensure they behaved maternally, meaning that they retrieved pups and built nests; about 1% of dams did not retrieve pups and these animals were not used for co-housing. Naive virgins were initially pre-screened for retrieval or pup mauling before co-housing; around 5% of the naive virgins retrieved at least one pup or mauled pups during pre-screening and these mice were excluded from subsequent behavioural studies.

Co-housing of a virgin female with a mother and litter was conducted for 4–6 consecutive days in 80 × 40 × 50 cm plastic home cages. The floor was covered with abundant bedding material, food pellets and a pack of hydrogel for hydration placed in a corner of the bin and refreshed daily. Nesting material was also placed in the cage. We first placed the dam and her postnatal day 1 (P1) litter in the cage. After the dam was acclimatized for ~30 min, we introduced the virgin female with a tail mark for identification. Well-being of the adult mice and pups was monitored at least twice a day. A surveillance infrared camera system (Blackrock Microsystems) was positioned ~100 cm above the home cage to capture the entire surface. An ultrasonic microphone (Avisoft) was placed in the corner of the cage, ~10 cm above the nest. Two initial cages had a second camera placed on the side but these videos were not analysed for these experiments. For studies of spontaneous pup retrieval by dams and the influence of co-housing, some dams were singly housed with their litter but not with other adults.

In cases where co-housing was done only between a virgin female and pups (Fig. 1c), the pups were returned to the donor mother every 12 h (for at least 48 h) and immediately replaced with new pups. This was done to ensure that they stay alive and healthy despite not being fed during co-housing with the virgin. The procedure was repeated throughout the duration of the co-housing27.

Pup retrieval testing

This test was used for the initial screening of dams and virgin female mice. In addition, outside of the spontaneous home cage behaviours, we specifically monitored pup retrieval every 24 h by the virgin females. We placed the female mouse to be tested in a behavioural arena (38 × 30 × 15 cm) containing bedding and nesting material; the female was alone, without contact with other animals. Each animal was given 20 min to acclimatize before each testing session began. The entire litter (ranging from 3 to 7 P1–4 pups) were grouped in a corner of the arena and covered with nesting material, and the adult female given an additional 2 min of acclimatization (pup group size did not affect retrieval behaviour; Extended Data Fig. 2c). One pup was removed from the nest and placed in an opposite corner of the arena. The experimental female was given 2 min per trial to retrieve the displaced pup and return it back to the nest; if the displaced pup was not retrieved within 2 min, the pup was returned to the nest and the trial was scored as a failure. If the pup was successfully retrieved, the time to retrieval was recorded and the trial was scored as a success. Another pup was then taken out of the nest, placed away from the nest (varying the position of the isolated pup relative to the nest from trial to trial), and the next trial was begun. After ten trials, pups were placed back into their home cage with their dam. We used an ultrasonic microphone (Avisoft) to verify that isolated pups vocalized during testing.

We reported probability of retrieving out of ten trials. Reliable retrieval was defined as having at least two out of ten successful trials. We used two-way ANOVA and Sidak’s multiple-comparison test corrections to compare probability of retrieving in each group over days, and Student’s t-test to compare the day of retrieval onset for each group.

Video and audio analysis

Video and audio recordings were synchronized with the neuronal recordings, and then analysed with Adobe Audition and Avisoft. For video recordings we used the BORIS suite for scoring of behavioural observations. Three separate teams of independent scorers (two scorers from the Sullivan laboratory, three scorers from the Carcea laboratory and four scorers from the Froemke laboratory) were trained in a similar way on how to identify relevant individual and social behaviours during co-housing, and then scored the videos blind to the conditions. The results from each raster were compared and compiled, and results from each lab were cross-validated. Nest entry was considered the moment when the head of the animal entered the nest. Nest exit was considered the time when the rear of the animal left the nest. We used two-way ANOVA and Sidak’s multiple-comparison test to compare pup retrieval rates and time in nest across days for each group.

Any event in which the dam chased the virgin towards the nest was identified as a shepherding event (that is, where distance from start to nest was greater than distance from end to nest). To determine the distance from nest during shepherding, we measured the distance from the bottom left corner of the cage to the position of the snout of the mouse, and to the position of the nest center. We then calculated distance from the virgin to nest. In cases of physical contact, start of shepherding was considered to be the moment when the dam made contact with the virgin, and the end of shepherding was the moment when the virgin stopped running. In some cases (especially later into co-housing), we noticed that virgins started running as soon as they noticed the dam approaching; in those cases, the start of shepherding was considered to be the moment when the virgins started running after the dam’s approach. For Fig. 1i, we used paired t-tests to compare distance from start of shepherding to nest with the distance from end of shepherding to nest. For Fig. 1j, we used one-sample Student’s t-tests to determine if the daily frequency of shepherding was higher than 0.2 events per h (which was the average rate of dam–virgin chases in absence of pups). Audio recordings were processed in Adobe Audition, and isolation or distress calls were distinguished from adult calls and wriggling calls on the basis of the characteristic statistics (bout rate of 4–8 Hz and frequencies of 40–90 kHz).

Observation of experienced retrievers

We first confirmed that virgins did not retrieve and dams retrieved at 100% at baseline. The exposures were done in standard behavioural arena (38 × 30 × 15 cm). The virgin and dam were acclimatized for 20 min, then the nest with pups was transferred to this arena. After another 5–10 min, we manually isolated one pup at a time so that the dam would retrieve the pup back into the nest. We repeated this for ten times per session. In the experiments where either a transparent or an opaque divided the cage, the two adult animals were acclimatized on opposite sides of the barriers. After exposure, the adult animals were separated and the virgins were tested for pup retrieval 30 min later, as described above. As the preparation for testing and the acclimatization to the testing cage also took 30 min, this amounted to a total 60-min interval between virgin observation and testing of responses to isolated pups. The exposure was repeated for four sessions (one per day). A virgin that retrieved at least once during the four days of observation was considered as having acquired pup retrieval behaviour. We used chi-square exact tests to compare retrieval between conditions: wild-type mice with no barrier, wild-type mice with transparent barrier, wild-type mice with opaque barrier, and OXTR-KO virgins with transparent barrier.

Surgery

Viral injections

To test the effects of DREADDs, stereotaxic viral injections were performed in Oxy-IRES-Cre mice15. Mice were anaesthetized with 0.7–2.5% isoflurane (adjusted on the basis of scored reflexes and breathing rate during surgery), placed into a stereotaxic apparatus (Kopf), and bilateral craniotomies performed over PVN (from bregma: 0.72 mm posterior, 0.12 mm lateral). Injections were performed at a depth of 5.0 mm with a 5 µl Hamilton syringe and a 33 gauge needle. Cre-inducible AAV2 hSyn::DIO-hM4D(Gi)-mCherry (University of North Carolina Viral Core) virus was injected into PVN at 0.1 µl min−1 for a final injection volume of 1.2–1.5 µl. The craniotomy was sealed with a silicone elastomer (World Precision Instruments), the skin sutured. Animals were used for experiments after two weeks to allow for viral expression.

For monosynaptic retrograde tracing of inputs on OT-PVN neurons, we injected 0.5 µl of a helper virus, AAV8-DIO-TVA-2A-oG (Salk Institute), in the left PVN of Oxy-IRES-Cre mice. Two weeks later, we injected 0.2 µl of a pseudotyped rabies virus, EnvA G-deleted rabies-mCherry (Salk Institute) at the same location. After another two weeks, we perfused the animals and collected the brains for histology.

For optogenetic stimulation of superficial superior colliculus (sSC)→PVN projections, wild-type virgins were injected in the left sSC (from bregma: −3.08 mm posterior, −0.27 mm lateral, −1.9 mm ventral) with 0.5 µl of either AAV9-pACAGW-ChR2-Venus or control AAV9-CAG-Venus virus. We then implanted a 400 µm optic fibre in the left PVN.

For fibre photometry from auditory cortex, we performed viral injections into the left auditory cortex (~1.7 mm anterior from the occipital suture, 0.5 mm lateral from the temporal ridge, 1 mm ventral from pia), using a similar procedure. These coordinates corresponded with the ones previously published from our laboratory15: 2.9 posterior, 4.0 lateral (left) from bregma. We also oriented using local markers consisting in the branching patterns of the rhinal vein and of the middle cerebral artery, to target the auditory cortex as previously described28. We injected 1 µl of AAV1 Syn::GCaMP6s (Addgene) at a titre of 1 × 1013 vg ml−1 in the auditory cortex of wild-type mice. Following the injection, we implanted a 400 µm optical fibre (ThorLabs) just above the auditory cortex, or inserted in the superficial layers of the cortex (200–300 µm below the pial surface). For photometry from PVN→auditory cortex projection neurons, we injected 50 nl of AAVrg-hSyn1-GCaMP6s-P2A-nls-dTomato virus (Addgene) at three locations within the left auditory cortex. The virus titre was 1.31 × 1012 vg ml−1, and we injected at 10 nl min−1.

For optogenetically manipulating OT-PVN→auditory cortex projections, we injected the left PVN of Oxy-IRES-Cre virgins with 0.5 µl of either AAV9-Ef1a-DIO-ChETA-EYFP or of AAV9-pCAG-FLEX-EYFP-WPRE. We then implanted a 400-µm optic fibre in the left auditory cortex.

Microdrive implantations

For in vivo single-unit electrophysiology we implanted microdrives either in the left PVN (Figs. 2–4) or the left auditory cortex (Fig. 4). We built microdrives using the parts and instructions for 4-tetrode Versadrives (Neuralynx), adapting these instructions for two bundles each made up of eight 12.5-µm Nichrome wires. The day of the implantation, the wires were cleaned and gold plated to achieve impedances <500 kΩ. After the virgin female mice were anaesthetized with isoflurane, a craniotomy (1.5–2 mm in diameter) was performed above the target structure, and two additional small craniotomies were performed in the occipital bone and the right parietal bone for insertion of bone screws. The ground and reference wires of the microdrives were soldered separately to these two bone screws. The dura was removed at the desired implantation site and the electrode bundles were slowly lowered to ~500 µm above the target brain structure (4 mm ventral from pia for PVN). For recordings from the auditory cortex, we first acutely recorded multiunit activity with tungsten electrodes to localize auditory cortex during implantation procedure. Auditory cortex was identified with pure tones (60 dB SPL, 7–79 kHz, 50 ms, 1 ms cosine on–off ramps) delivered in pseudo-random sequence at 0.5–1 Hz. The craniotomy was covered with mineral oil and silicone elastomer, and the microdrive was secured to the skull using dental cement (C&B Metabond).

For optically identified recordings from OT-PVN units, we used Oxt-IRES-Cre mice bred into a C57BL/6 background. Prior to the microdrive implantation, we injected the PVN of these animals with 1–2 µl of AAV1-CAG::DIO-ChR2 at a titre of 1 × 1013 vg ml−1 (in one mouse, we injected AAV2-EEF1::DIO-ChETA at a titre of 1 × 1013 vg ml−1). Then, after implanting and cementing the microdrive, we rotated the head of the animal at a 45º−50º angle around the anterior–posterior axis, and another craniotomy was done at 0.72 mm posterior and 3.5 mm left from bregma. A 400 µm optical fibre for delivering blue light was slowly lowered at this position for 4.5 mm below the brain surface. For dual recordings, we implanted a microdrive in left PVN, injected the GCaMP virus (as described above) and implanted an optic fibre in the left auditory cortex during the same surgery.

Optical fibre implantation

For fibre photometry we used 400-μm-diameter 0.48NA optic fibres inserted into 2.5-mm-diameter ceramic ferrules (Doric Lenses). For recordings of PVN→auditory cortex, we used 1.25-mm metal ferrules, with fibres 1 mm long (below ferrule) for recordings in the auditory cortex and 5 mm long for recordings in PVN. They were implanted during the same surgery as the viral injections, and cemented to the skull as described above.

Recordings

Single-unit recordings during behaviour

Neuronal recordings were performed at 30 kHz sampling rate using the Cereplex µ headstage, a digital hub and a neuronal signal processor (Blackrock Microsystems). The recordings were synchronized with a video recording system (Neuromotive, Blackrock). Before the start of the experiment the electrode bundles were lowered into the target structure and then advanced daily by ~70 µm. For PVN recordings, optical tagging tests were performed at the end of each co-housing day prior to lowering the electrodes for unit identification on the following day. Optical identification was performed in a separate clean cage in absence of pups or dam. The optical fibre was connected to a blue laser triggered using an analogue output from the recording system. The laser pulses were 5 ms in duration and delivered at either 2 or 5 Hz. Light intensity was controlled by adjusting the output of the laser, and at least three different intensities were tested in each mouse. The optical identification procedure was <5 min each day, and we used lower frequency (2 Hz) stimulation to minimize the likelihood of functional changes induced by optogenetic stimulation. To allow free movement of the implanted mice, we used a pulley combined with a fibre-optic and electrical rotary joint–commutator (Doric Lenses).

Spike sorting

Neurons were isolated offline using Cerebrus and BOSS software (Blackrock Microsystems), as previously described29. A notch filter was applied to eliminate the line noise, then a high-pass filter of 250 Hz was applied, and a threshold multiplier was used to extract spike from signal energy between 1–5 kHz. We then used principal component analysis to extract spike features and to manually cluster them in different units. We eliminated spikes that violated a refractory period of 2–3 ms, and spikes that were simultaneously recorded in 10 or more channels (as potential artifacts). For tetrode recordings, we used a similar procedure in BOSS (Blackrock) that allowed triangulation of waveforms from the four tetrode channels. For identifying oxytocin neurons in optically tagged recordings, we aligned neuronal activity to the onset of the blue light pulse30. Neurons that reliably fired (≥70% of trials) at short latency (≤4 ms) after the onset of blue light were selected as being oxytocin neurons.

Spike train analysis

To analyse changes in firing patterns, we aligned the spike trains to the onset of the behavioural episode and calculated the firing rate for the duration of the behaviour. In the case of ‘nest entry’, we used a 40-s cut-off, in order to minimize the impact from other behaviours that occur in the nest over longer time periods (for example, grooming, sleeping and nest building); this cut-off was chosen as most nest dwell times were ≤40 s (Extended Data Fig. 7b). For each behavioural episode we also calculated the corresponding baseline firing rate, either for an interval similar in duration to the behavioural episode (in the case of nest entry), or for 10 s before the start of the behavioural episode (for shepherding and dam retrievals). The change in firing rate was calculated as: (Spikingbehaviour − Spikingbaseline) × 100/(Spikingbehaviour + Spikingbaseline), where Spikingbehaviour was firing rate during the behavioural event, and Spikingbaseline was firing rate during baseline. To determine whether these changes in firing rate were statistically significant, we shifted the behaviour intervals by a random value between −500 and 500 s, and then calculated new firing rates and modulation percentages for 1,000 random shifts. All cells for which the modulation of activity in the unshifted data was greater than modulation in 950 shuffles were considered to be potentially significantly activated31. Cells for which the change in firing rate was smaller than the change in 950 shuffles were considered to be potentially significantly suppressed. For each unit, we computed the P-value from the permutation analysis and then applied a false-discovery rate correction for the multiple comparisons performed on each data subset (all OT-PVN or all PVN neurons recorded in a day), to identify neurons with significant changes in firing rate. The threshold for q-value was set 0.05, that is, 5% accepted false-discovery rate. To determine if the fraction of activated neurons is meaningful, we then applied a one-sided Fisher’s exact test (with the null hypothesis of zero neurons being activated) with the threshold for alpha set at 0.05 (ref. 32). To compare fractions of OT-PVN and PVN neurons activated by a specific behaviour, we applied a two-sided Fisher’s exact test, and calculated the effect size as the relative risk: the probability of neurons being activated in the OT-PVN group/the probability of neurons being activated in the PVN group, using the Koopman asymptotic score for calculating 95% confidence intervals. To investigate how PVN activity correlated with activity in the auditory cortex, a population average change in PVN firing was calculated for each maternal-retrieval trial of the observational learning procedure.

To quantify unit activity during behavioural episodes, we aligned the neuronal data to the onset of the behaviour, and binned the spiking data in 250-ms bins for an interval between −20 s and +40 s (−10 s to +20 s in the case of shepherding episodes). We then calculated the z-score value for each bin in the chosen time window; for display purposes, we used bounds of −3 (minimum z-score) and +3 (maximum z-score). For responses to pup call playback, spike trains from single-unit recordings in PVN or auditory cortex were aligned to the onset of the played-back vocalization. The average firing rate during the pup call (1-s interval) was normalized to the firing rate during baseline (1 s preceding pup call onset).

To measure the cross-correlation for simultaneously recorded OT-PVN and PVN single-units during co-housing (Extended Data Fig. 7d, e), we calculated the normalized zero-lag correlation of spike counts (in 10-ms bins) across all pairs of units33, separately for each of the three behavioural conditions (shepherding, nest entry and maternal retrievals). We then averaged z-transformed r values across all events of each behaviour, to obtain a single cross-correlation value for each pair and each behaviour. Higher correlation r values reflect higher synchrony within the pair.

Fibre photometry

To perform fibre photometry, we connected the optical fibre implanted in the auditory cortex and housed in a ceramic ferrule to a custom-built photometry rig34. A 400-Hz sinusoidal blue light was delivered via the optical fibre from an LED (30 µW) for GCaMP6s excitation. We collected the emitted (green) light via the same optical fibre, used a dichroic mirror and appropriate filters to direct emitted light to a femtowatt silicon photoreceiver (Newport), and recorded using a real-time processor (RX8, TDT). The analogue readout was then low-pass filtered at 20 Hz. The intensity of blue light excitation was adjusted to produce similar baseline fluorescence levels across sessions in the same mouse. The sound processor used for delivering the pup calls (RZ6, TDT) was synchronized with the fibre photometry system.

For investigating changes in neuronal activity of the left auditory cortex throughout co-housing, we recorded responses to six different played-back pup calls every 24 h. To avoid possible changes in the magnitude of calcium transients with changes in the position of the mouse relative to the speaker, in some recordings we restrained the mouse in a mash cup, at the same distance from the speaker. Data from restrained or unrestrained mice showed similar trends, and were thus pooled together. To determine the evoked response recorded with fibre photometry, we calculated the ΔF/F for each pup call interval.

Perturbations

Chemogenetic suppression of oxytocin neurons during co-housing

Oxy-IRES-Cre virgin female mice expressing in OT-PVN cells either DREADD coupled to inhibitory G-protein (DREADDi, also known as hM4D(Gi)) and fused with mCherry, or mCherry alone (as control), were co-housed with mothers and pups as described above. In both experimental (Oxy-IRES-Cre virgins expressing DREADDi-mCherry) and control (Oxy-IRES-Cre virgins expressing mCherry) cohorts, clozapine-N-oxide (CNO) was administered in the drinking water at a concentration of 25 mg l−1 (ref. 35). Five sugar pellets were added in each bottle to obscure the taste of CNO. At this concentration, for the average mouse body weight (~30 g) and average daily water consumption in mice (~6 ml), the average CNO consumption was ~5 mg kg−1 per day.

Three out of seven mice in the mCherry+ CNO cohort had DREADDi virus expressed outside of the PVN due to mis-targeting. These mice were also treated with CNO; as their behaviour was similar to the mCherry-expressing mice treated with CNO—we combined data from these groups together in Fig. 2a. For each animal, the retrieval probability for the first two days of co-housing was averaged. For Extended Data Fig. 4b, retrieval testing was done in animals that stably retrieved at baseline and as such were not subjected to the co-housing procedure.

Optogenetic stimulation

For optogenetics, we connected the optical fibre implanted either in PVN or in the left auditory cortex of the virgin mouse to a blue light laser (OptoEngine). Light pulses (5 ms) were delivered at 20 Hz (for Extended Data Fig 9; ref. 21) or 30 Hz (for Fig. 4; ref. 15), while the dam performed 10 retrieval trials across an opaque barrier from the virgin. To minimize potential adaptation or synaptic depression, stimulation was performed for ≤5 min, enough time for dams to complete 10 retrievals. After 30 min, the non-co-housed virgin was tested for retrieval. This procedure was repeated once a day for four consecutive days.

Drug infusion

To determine the role of cortical oxytocin receptor activation during observation of pup retrieval, we implanted cannulas in the left auditory cortex of naive virgin females15. After mice recovered from surgery, we placed them in the observational chamber, and then infused either saline or the oxytocin receptor antagonist OTA (1 µM solution in saline) through the cannula. We injected a volume of 1.5 µl at a rate of 1 µl min−1. We removed the internal from the cannula five minutes after the end of injection to allow the solution to diffuse, and then allowed the virgins to acclimatize to the observational chamber for another 10 min. After this, we exposed virgins to maternal retrievals across the transparent barrier as described above. Drug conditions were compared with a survival log-rank analysis to determine effects on retrieval onset across groups.

Immunohistology

To verify viral expression at the end of the experiments, animals were fixed with transcardiac perfusions of 4% paraformaldehyde. Brains were removed and further preserved in a paraformaldehyde solution for 1–2 h at 4 °C. Afterwards, brains were sequentially cryopreserved in 15% and then 30% sucrose solution, embedded in OCT solution and sectioned (30 µm) with a cryostat then mounted on positively charged slides. Immunohistochemistry was performed on the mounted sections as previously described29,36. Sections were blocked in 5% goat serum solution for 1 h at room temperature or for 14 h at 4 °C. A solution of the appropriate primary antibodies was diluted in 1% goat serum and 0.01% Triton X solution and then applied for 24 h at 4 °C. We used a rabbit anti-oxytocin antibody (EMD Millipore, 1:500), mouse anti-oxytocin antibody (a gift from H. Gainer at national Institutes for Health), a chicken anti-GFP (Aves, 1:500) and a chicken anti-mCherry (Abcam, 1:1,000) antibody. The sections were washed in PBS solution, and a solution of fluorophore-conjugated secondary antibodies applied for 1.5 h at room temperature. All secondary antibodies were from Jackson Immunoresearch and used at 1:200. Slides were examined and imaged using a Carl Zeiss LSM 700 confocal microscope with four solid-state lasers (405/444, 488, 555 and 639 nm) and appropriate filter sets. For imaging sections co-stained with multiple antibodies, we used short-pass 555 nm (Alexa Fluor 488), short-pass 640 nm (Alexa Fluor 555), and long-pass 640 nm (Alexa Fluor 647) photomultiplier tubes.

Micro-computed tomography imaging

The localization of the implanted electrodes was assessed in vivo using μCT scans in post-implanted mice, followed by co-registration with an online digital MRI mouse brain atlas. The µCT datasets were acquired using the μCT module of a MultiModality hybrid micro-Positron Emission Tomography–µCT Inveon Scanner (Siemens Medical Solutions). The Inveon scanner is equipped with a 165 mm × 165 mm x-ray camera and a variable-focus tungsten anode x-ray source operating with a focal spot size of less than 50 μm. The scan consisted of a 20-min whole-head acquisition over an axial field of view of 22 mm and a transaxial of 88 mm with a resolution of 21.7 μm pixels binned to 43.4 μm. 440 projections were acquired using a 1-mm aluminum filter, a voltage of 80 kV, and a current of 500 μA. The datasets were reconstructed using the Feldkamp algorithm37.

The hybrid scanner was equipped with a M2M Biovet (Cleveland) module used to monitor continuously vital signs. All mice were monitored continuously throughout the scanning session via a respiration sensor pad (SIMS Graseby). The imaging scan consisted of initially placing each mouse in an induction chamber using 3–5% isofluorane exposure for 2–3-min until the onset of anesthesia. The mouse was then subsequently positioned laterally along the bed palate over a thermistor heating pad in which 1.0% to 1.5% isofluorane was administered via a 90º angled nose cone throughout the scan. The head of each subject was judiciously oriented perpendicular to the axis of the mouse body so that the extracranial part of the implanted electrode could be easily kept away from the field of view of the µCT image acquisition. Importantly, the large extracranial metal components and dental cement of the implant can cause beam hardening that can appear as cupping, streaks, dark bands or flare in the µCT38,39,40. To this effect, the head positioning helped reduce the risks of image artifacts that could be induced by the implant along the path of the X-ray beam.

Unlike MRI, μ-CT imaging can be performed on subjects with metal implants. However, lack of the soft tissue contrast of the µCT limited its usefulness to provide the needed brain anatomical detail in order to verify the electrode correct localization. Our approach combined registration of post-implant µCT with an existing online MRI brain atlas for adult C57Bl/6 mice from the Mouse Imaging Centre: (https://wiki.mouseimaging.ca/display/MICePub/Mouse+Brain+Atlases).

The three-dimensional MRI mouse brain atlas was established by acquiring 40 individual ex vivo mice using T2-weighted sequence on a 7-Tesla scanner. All the data were averaged and resulted into a 40 µm isotropic resolution dataset detailed in Dorr et al. (ref. 41). The hypothalamic PVN was manually segmented and color-coded, with the guidance of the P56 coronal Allen mouse histology brain atlas42. This region was set as the target of reference. A rigid co-registration between µCT and the modified MRI atlas images was systematically performed using a commercial software Amira (Thermo Fisher Scientific). Both datasets were then overlaid for visual analysis and to determine the sub-millimetric localization of the electrode tip.

Reporting summary

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



Source link

By AUTHOR

Leave a Reply

Your email address will not be published. Required fields are marked *