Kwik cast

Surgeries were performed under isoflurane anaesthesia (induction: 4% v/v; maintenance: 2% v/v in O₂). Before the start of the surgery, the depth of anaesthesia was verified by the absence of a reaction to an ear pinch. When the depth of anaesthesia was sufficient, the mouse was put on a heating pad to keep its body temperature stable at 37°C, which was verified with a rectal thermometer. The eyes were protected with eye ointment (Duratears, Alcon, Fort Worth, TX, USA). To prevent dehydration, each mouse received 1 ml of saline s.c. injection before the surgery commenced.
Mice received a magnetic pedestal for head fixation, attached to the skull above bregma using Optibond adhesive (Kerr Corporation, Orange, CA, USA). To place this pedestal, the surgical spot was shaved and lidocaine (4 mg/kg s.c.; Braun, Melsungen, Germany) was injected. After 1 min, a small incision was made in the skin so that the periosteum became visible. A few extra drops of 2% lidocaine were applied topically, after which the periosteum was removed and the skull was cleaned and dried.
Subsequently and while still under anaesthesia, a craniotomy was performed to expose the recording area (the lateral part of vermal lobules VI an VII and the adjacent hemispheric lobules crus 1 and crus 2 on the right side of the mouse). Four mice received a larger and more central craniotomy in order to expose both right and left vermis for optogenetic stimulation. The preparation was identical as for the pedestal placement, including shaving, lidocaine injection, and a small incision in the skin. With a dental drill, an opening in the skull was made, while the dura mater was preserved. The craniotomy was cleaned, and a small chamber was built with dental cement surrounding the craniotomy to contain fluids during the later recording session. At the end of the surgery, the craniotomy was covered with Kwik‐Cast (World Precision Instruments, Sarasota, FL, USA). Postsurgical pain was treated with carprofen (5 mg/kg s.c.; Pfizer, New York, NY, USA), buprenorphine (50 µg/kg s.c.; Indivior, Richmond, VA, USA) and bupivacaine (1 mg/kg s.c.; Actavis, Parsipanny‐Troy Hills, NJ, USA).
At the end of surgery, the mice were allowed to recover under a warming lamp for at least 30 min. The warming lamp covered only a part of the cage, so that the mice could chose to be under the lamp or not. When a mouse was behaving normally again, it was brought back to the stable.

Three-to-five-month-old Csf1r^{∆FIRE/∆FIRE} mice and their littermate controls were anesthetized with isoflurane (3% induction, 1% maintenance, 0.5 l min⁻¹ in pure oxygen) and mounted into a stereotaxic frame. Two steel screws (1-mm diameter) were placed in small boreholes drilled above the cerebellum and olfactory bulb. A custom-made head-ring for later head-restraint training and recording was attached to the skull and the screws were tightly fixed using bone cement (Refobacin, Biomet). After surgery, mice were injected with carprofen (120–130 µl subcutaneously, 20 mg kg⁻¹) and received a liquid supplement with 0.9% NaCl, 200–300 µl intraperitoneally. Mice were given at least 7 days to recover. Mice were then handled and habituated to head restraining on an air-cushioned styrofoam ball in a virtual reality (VR) system (JetBall, Phenosys). After habituation, mice were water-restricted (approximately 1.2 ml water per day) and trained to run in a VR linear treadmill for liquid rewards. When the animals could perform sufficient runs on the treadmill after 3–4 days, they were anesthetized and three small craniotomies were drilled, one above the PFC (anteroposterior: ≈1.9 mm; mediolateral: ≈0.5–1 mm) and two above the HPC in each hemisphere (anteroposterior: ≈−1.9 mm; mediolateral:≈ ±1.5 mm). The dura was removed. Craniotomies were sealed and protected using silicone (Kwik-Cast, World Precision Instruments). Animals were left to recover for at least 8 h before electrophysiological recordings started.
To perform the VR linear track treadmill task, mice were head-fixed on an air-cushioned styrofoam ball in a 270° TFT surround monitor system (JetBall, PhenoSys). Mice were trained to run to advance in a custom-made virtual linear corridor. The running path length was 50 cm for each trial. Once the animal reached the end of the virtual corridor in each trial, a 50-µl liquid reward (5% sucrose water) was delivered through a spout. After a 3-s inter-trial interval, animals were teleported to the start of the virtual corridor to initiate the subsequent trial. Each animal was trained once daily for 15–30 min for at least 3 days when sufficient amounts of liquid (≈1–1.2 ml) were obtained during the task. Trial signals in the tasks were recorded as transistor-transistor logic pulses by the acquisition device. The animal’s locomotion was detected by an XY-motion sensor at 50 Hz.
To perform recordings, mice were attached to the stereotaxic frame in the VR and were allowed to perform the linear treadmill task. Two silicone probes, each consisting of 128 channels on four parallel shanks (A4X32-Poly2–5mm-23s-200-177, NeuroNexus), were inserted into the PFC and the HPC, respectively through the craniotomies prepared. LFP signals were recorded using an Intan 128-channel head-stage from each probe and an Intan RHD recording controller. Signals were sampled at 20 kHz and digitized as 16-bit signed integers. All signal processing and data analysis were performed in MATLAB (MathWorks). The power-line interference (50, 100, 150 and 200 Hz) was removed from the LFP signals by a second-order band-stop filter. After eliminating further noise, LFP signals were downsampled to 2 kHz using a resample function that applies an FIR antialiasing low-pass filter. Then LFPs were filtered between 0.5 and 500 Hz using a second-order Butterworth band-pass filter. Segmentation of LFP data where the animal’s running speed was 5–10 cm s⁻¹ were extracted for analysis. A 5-s window, backwards from 0.5 s before the reward delivery time, was selected in each trial. The power spectrum density and coherence analysis were obtained using the pwelch (window: 2 s; overlap: 50%) and the mscohere function (window: 2 s; overlap: 50%) functions, respectively. A theta (6–10 Hz):delta (2–4 Hz) ratio greater than 4 of the mean power spectral density in the HPC was applied to select clear running-associated and alertness-associated theta epochs as described previously^{56 (link)–59 (link)}. Power was calculated as the sum of the power spectral density in each frequency band; coherence was calculated as the mean in each frequency band (theta: 4–12 Hz; beta: 15–25 Hz; gamma: 26–70 Hz).