Acute recordings

For recordings under anesthesia, the mice were anesthetized with isoflurane and a craniotomy (~2 mm wide) was performed above the right mPFC. After removal of the dura, the craniotomy was sealed off with dura gel (Cambridge NeuroTech) and a custom 3D-printed head bar with a circular ring around the craniotomy site was fixed to the skull with dental cement. The animal was then immediately transferred to the recording station while anesthesia was maintained throughout the recording session with intraperitoneal injections of ketamine and xylazine (initial dose: 100 and 13 mg/kg body weight, respectively, topped up by 10–20% every 20–40 min). For awake recordings, a steel head plate was implanted on the skull and the animals were allowed to recover from head plate implantation for 3 days. For habituation to head fixation, the mice were briefly sedated with isoflurane and head-fixed such that they could comfortably stand on a circular Styrofoam weal. A virtual reality (circular track, length 1.5 m, visual cues placed outside the arena) was constructed with open-source 3D rendering software (Blender) and was projected on five computer screens surrounding the head-fixation setup (Schmidt-Hieber and Häusser, 2013). Over subsequent days, mice were accustomed to head fixation by daily increasing the time of head fixation until the animals appeared calm and traversed the circular maze reliably. Once the animals were habituated, a craniotomy was performed as described above. Carprofen was injected subcutaneously on the day of the surgery. A four-shank silicon probe (Cambridge NeuroTech) coated with fluorescent marker (DiD) was slowly (~5–10 µm/s) lowered to the mPFC (935–1758 µm below brain surface). Wide-band neural signals were recorded at 30 kHz sampling with a 64-channel amplifier (Intan Technologies) connected to a USB acquisition board (OpenEphys; Siegle et al., 2017). Laser light (473 nm, ~10 mW intensity at the fiber tip, 50 ms pulses at 0.1 Hz) was delivered through a 200 µm optical fiber glued to the silicon probe. Afterward, the silicon probe was slowly retracted and the animals were transcardially perfused with ~20 ml phosphate-buffered saline followed by ~30 ml of 4% paraformaldehyde. After post-fixation overnight, 100-µm-thick frontal sections of the mPFC were cut and stained with rabbit-anti-PV antibody (1:1000, Swant PV27) and DAPI. The location of the silicon probe and the immunostaining were visualized with a laser-scanning microscope (LSM 710, Zeiss). Recordings in the prelimbic and cingulate cortex were pooled for analysis.