Fluorescence imaging system
We used an Olympus MVX10 microscope and a Hamamatsu ORCA Flash 4.0 SCMOS (2048 × 2048, pixel size 6.5 μm, 16 bit) for fluorescence imaging. The images were acquired by HCImageLive software. A high-power mercury lamp (U-HGLGPS, Olympus, Japan; 130 W) served as the light source. The GFP filter cubes contained a 460–490 nm bandpass filter, a 520 nm high pass filter, and a 500 nm dichroic filter (Fig. 1A). Fluorescence images were acquired using a 200 ms exposure time with 1024 × 1024 pixels array, amplified at 12.6×, and were saved by 2 × 2 spatial binning. Direct current (DC) potentials were filtered at 0–100 Hz, amplified at 50× and acquired at a sampling rate of 200 Hz through a differential amplifier (Model 3000, A-M system, USA). Mice were trained to habituate to the head-fix setup with an adjustable running wheel, as shown in Fig. 1D. A near infrared camera was equipped to monitor mouse movement.
Temporal synchronization of all data streams (fluorescence imaging, DC recording, body tracking camera) was achieved by the LabVIEW (National Instruments, USA) control program.
Animal preparation
34 transgenic VGluT2-GCaMP6s mice weighing 27 ± 5 g (8–16 weeks old) were used in this study. Mice were acquired by hybridizing Vglut2-ires-Cre mice (Stock No. 028863) and Cre-dependent GCaMP6s mice (Ai96, Stock No. 028866) from the Jackson Laboratory. Mice were housed under a normal 12 h light/dark cycle with food and water provided ad libitum.
We used a Gradient Index (GRIN) lens (NEM-050-06-08-520-S-1.5p, Grintech, Germany; Single, 1×, NA 0.5, 0.5 mm in diameter) to collect the fluorescence signal (Fig. 1C). This GRIN lens is a cylindrical optical lens with a radial negative gradient of refractive index. It allows continuous refraction of light transmitted along the axial direction, achieving a smooth and continuous convergence of the light to a single point, which has the characteristics of focusing and imaging. The GIRIN lens would focus mercury lamp illumination near the top of the lens plane, relay the illumination into the brain, and relay neuronal signals of VPM to the lens plane above the cortex for image acquisition. Although glial activation around the implanted GRIN lens was inevitable like any other implanted entity such as electrodes, the optical penetration was up to ~ 650 μm, which could reach tissue lying beyond the glial activation layer for imaging [3].
Mice were anaesthetized with 2% isoflurane and placed in a stereotaxic apparatus during surgery. Body temperature was maintained at 37 ± 0.5 °C using a heating pad. After the skull was exposed and cleaned with saline, a craniotomy was performed (0.6 mm in diameter) at 1.7 mm posterior and 1.75 mm lateral to bregma. To prevent mechanical compression and damage, a cylindrical column of brain tissue above the VPM structure was removed by aspiration using a 27-gauge blunt needle [3]. The GRIN lens was fixed to a micromanipulator (MX7600L, Siskiyou, USA) and inserted into the exposed areas slowly to minimize bleeding and mechanical damage (150 μm/min). Two additional craniotomies were performed to implant electrodes. The recording electrode was implanted on the dura about 0.3–0.5 mm posterior to the GRIN lens. The recording electrode was made of silver wire (0.2 mm in diameter, CAS 7440-22-4) and attached to the dura. A steel screw (0.5 mm in diameter) was used as the reference electrode and implanted 1.5 mm deep into the nasal bone. In some experiments, electrodes were also implanted on the dura overlying the contralateral homologous cortex to simultaneously monitor electrophysiological signals in both hemispheres of the brain. We applied dental cement around the implants for stabilization and attach a metal head bar for head fixation, while a piece of flexible tube was affixed over the GRIN lens to protect it.
Mice were allowed to recover for 4 weeks before the experiment. Tolfedine (0.2 mg/kg, intraperitoneal) and penicillin (20,000 unit, intraperitoneal) were administered to minimize tissue swelling and inflammation for 3–7 days. Mice were habituated to the behaviour stage for several days before the CSD experiments.
Experiment procedure
we applied 1 μl KCl solution (1 mol/L) on the dura to generate CSD. The occurrence of CSD was assessed by continuous recording of the DC potential. Mice were anaesthetized with 2% isoflurane and a craniotomy (0.5 mm in diameter) was performed to expose the dura overlying the motor cortex (AP, 2.5 mm; L, 2 mm) or visual cortex (AP, − 3.5 mm; L, 3 mm) (Fig. 1B) for KCl application 1.5 h before the start of the experiment. The dura remained intact. We generally recorded resting-state fluorescence images for 2–5 min before each CSD implementation. We induced CSD in the ipsilateral (n = 50) and contralateral hemisphere (n = 15) to determine the effect of CSD on the bilateral VPM of the thalamus in awake mice. For the experiments investigating the effect of anaesthesia on VPM activation, two common anaesthetics were used. The concentration of isoflurane (0.6%, 0.9% and 1.2%, n = 10 respectively) was controlled by an anaesthesia machine (VIP 3000 Veterinary Vaporizer, Midmark, USA). Pentobarbital (n = 13) was administered intraperitoneally at a dosage of 50 mg/kg. To determine whether CSD caused the activation of the VPM, saline was used to replace KCl (n = 16), or 5 mM MK-801 (n = 8) was administered to the dura overlying the motor cortex ipsilateral to the GRIN lens 1 h before KCl application to block CSD induction. Fluorescence images were acquired at 5 Hz for 10 min at each CSD experiment, and mice were allowed to recover for more than 1.5 h before the next CSD stimulation to avoid the influence of previous CSD. No more than five CSDs were induced per mouse per day. In our experiments, it takes us at least one month waiting for a mouse to recover from the surgery of GRIN lens implantation. Furthermore, the mouse cannot be used if the inflammatory response due to cellular edema or necrosis blurs the fluorescent signal. These difficulties in the preparation of the animal models limit the total number of animals can be used in our study so that we have to carry out multiple KCl applications in the same animal. Under the condition of isoflurane with three different anesthetic level, the sequence of anesthetic level applied is randomized for each successive KCl application in the same mouse to reduce the carry-over effect by repeated measures.
Data processing
We performed data processing using code written in MATLAB R2018b. We derived the fluorescence intensity change (△F/F) by calculating \( \Delta F/F=\frac{F(t)-{F}_0}{F_0} \), where F(t) is the instant ROI fluorescence intensity and F0 is the average ROI fluorescence intensity among the resting-state recordings. Statistical analysis of experiments (involving ipsilateral and contralateral VPM in awake state, saline application, pentobarbital anesthetization and MK-801 treatment) was performed using the analysis of variance (ANOVA) between CSD stimulated state and resting state. The repeated measurements of 3 levels of isoflurane and awake state in each mouse were analyzed with the repeated measures ANOVA. A P value less than 0.05 was considered significant. Data are presented as the mean ± standard deviation (SD). Sample size was calculated by PASS (α = 0.05, power = 90%) with estimated mean value of 0.2 and SD value of 0.08 for ipsilateral fluorescence intensity change. Resting state fluorescence change were estimated with mean value of 0.03 and SD value of 0.08. The calculation results indicated that the sample size greater than 6 could be accepted. No data were excluded except when the micro-endoscope was inserted into the wrong region or inflammatory tissue obstructed imaging severely.
For mouse motion detection, the global difference was used to measure the amplitude of movement, which was defined as \( \varDelta {I}_m=\frac{1}{N_{count}}\sum \left({I}_{t+1}-{I}_t\right) \), where It and I(t + 1) are two contiguous images acquired by a body tracking camera and Ncount is the total number of pixels of each image, which was 640 × 480 in our experiment.
Confirmation of the implanted micro-endoscope locations
Mice were sacrificed and perfused transcardially with 4% paraformaldehyde solution (PFA). Coronal sections (100 μm thick) of the brain around the micro-endoscope implantation sites were prepared with a vibratome. Brain slice fluorescence images were acquired with a microscope (MVX10, Olympus, Japan). The edge contour of the hippocampus was extracted from each image of the brain slices and registered to the reference brain atlas by manual translation, scaling, rotation and skewing. This deformation operation was applied to the whole slice for overall registration, and the micro-endoscope insertion point was located (see Fig. 1E and Fig. 2).