Imaging microglial surveillance in the somatosensory cortex in freely moving mice.

(a-c) Experimental setup. The animal was head-mounted with a miniature two-photon microscope (mTPM) and EEG/EMG electrodes and behaved freely in a cylindrical chamber (a). Microglia expressing GFP in the somatosensory cortex were imaged through a cranial window using the mTPM (b) and the sleep/wake state of the animal was simultaneously monitored using an EEG/EMG recording system (b-c).

(d-e) Microglial morphological dynamics when the animal was awake. A representative image with a FOV of 220 × 220 μm2. (e1-e3) Expanded views of selected microglia in box from (d) at 10 min (e1), 20 min (e2) and 30 min (e3) of continuous recording. (e1’-e3’) Microglial process graphs digitally reconstruction for E1-E3 using Imaris software. In Figure e1′-e3′, the branch points of processes are represented in red and the number of different branches of the whole cell is represented as gradient colors.

(f-h) Quantitative analysis of changes in microglia surveillance area (f), process length (g), and process motility, indexed by the speed of the extension and retraction at end points of the processes (h). Note that the gross morphology of microglia remained largely unchanged over a 30-min time frame in the wake state, despite significant motility at the ends of the processes. Scale bars, 25 μm.

n = 12 cells from 3 mice. Scale bars, 30 μm. See supplementary Fig. 1-3 for more details.

Microglial surveillance is state-dependent in the sleep-wake cycle.

(a) Representative EEG/EMG recordings showing the sleep-wake stage switch. Top, EEG power spectrogram (1-20 Hz). Middle, EMG trace. Bottom, brain states classified as wake (color code: gray), REM (blue), and NREM (white).

(b-d) Representative microglial morphological changes during the sleep-wake cycle.

(b′-d′) Microglial morphology reconstructed from b-d using Imaris software.

(e-h) Microglial morphological parameters, length (e), area change (f), number of branch points (g), and process end point speed (h), all exhibited brain state-dependent dynamic change.

One-way ANOVA with Tukey’s post-hoc test in e; Friedman test with Dunn’s post-hoc test in f-h; n = 20 cells from 7 mice for each group (e-g), n = 15 cells from 6 mice for each group (h); *P < 0.05, **P < 0.01, ***P < 0.001.

(i-o) 3D multi-plane imaging and reconstruction of microglial morphology. A 3D ETL lens was used to acquire multi-plane imaging (220*220*40 μm3) at Z-intervals of 2 μm, at a rate of 7.5 stacks/5 min.

(i-k) 3D reconstructed microglial morphology in wake (i), REM (j), and NREM (k) states, with corresponding time stamps shown at the bottom.

(l-o) Quantitative analysis of microglial length (l), volume change (m), number of branch points (n), and process motility (o) based on multi-plane microglial imaging. Data from 3D imaging corroborated state-dependent changes of microglial morphology in the sleep-wake cycle. Scale bars, 30 μm.

One-way ANOVA with Tukey’s post-hoc test in l, o; Friedman test with Dunn’s post-hoc test in m, n; n = 17 cells from 7 mice for each group (l-n), n = 15 cells from 6 mice for each group (o); *P < 0.05, **P < 0.01, ***P < 0.001.

Changes of microglial surveillance in the state of sleep deprivation.

(a) Experimental setup for sleep deprivation. Sleep deprivation in mice was achieved by forcing them to exercise and interrupting their sleep with the rotation of a 46-cm rod (18 turns/min) in the chamber (diameter 50 cm).

(b-d) Microglial processes contracted after sleep deprivation (SD), baseline (b), SD 3h (c), and SD 6h (d). (b′-d′) Morphological changes of microglia reconstructed using Imaris software. b’-d’ correspond to b-d, respectively.

(e-h) Statistics for length (e), area (f), number of branch points (g), and process motility (h). Scale bar, 30 μm.

One-way ANOVA with Tukey’s post-hoc test in g; Friedman test with Dunn’s post-hoc test in e, f, h; n = 15 cells from 6 mice for each group; *P < 0.05, **P < 0.01,

***P < 0.001.

NE dynamics in mouse somatosensory cortex during the sleep-wake cycle.

(a) Schematic diagram depicting mTPM recording of extracellular NE indicated by the GRABNE2m sensor expressed in neurons.

(b) Representative traces of simultaneous recordings in the somatosensory cortex during the sleep-wake cycle in freely behaving mice. EEG and its power spectrogram (0-20 Hz); EMG (scale, 50 μV); NE signals reflected by the z-score of the GRABNE2m fluorescence (scale, 2 z-score). The brain states are color-coded (wake, gray; NREM, white; REM, blue; NE oscillatory during NREM sleep, green).

(c) Mean extracellular NE levels in different brain states. Data from the same recording are connected by lines. ***P < 0.001, one-way ANOVA with Tukey’s post-hoc test (n = 11 from 3 mice).

Microglial surveillance during natural sleep is controlled by LC-NE signal.

(a) Experimental setup: LC-selective neurotoxin DSP4 was used to destroy LC-NE neuronal axons.

(b-g) Lack of sleep/wake state-dependent microglial surveillance in LC-axon ablated animals. b-d: Representative mTPM images of microglia at different states. e-g: Statistics for microglial length (e), surveillance area (f), and number of branch points (g) in LC-axon ablated mice. One-way ANOVA with Tukey’s post-hoc test in e, f; Friedman test with Dunn’s post-hoc test in g; n = 17 cells from 6 mice for each group; ns, not significant.

(h) Schematic diagram for β2ARs on the plasma membrane of microglia in the cerebral cortex responding to NE released from axonal terminals projected from LC.

(i-k) State-dependent microglial surveillance during sleep-wake cycle in β2AR knockout mice. Representative microglial images (i-k) and statistics for microglial process length (l), surveillance area (m), and number of branch points (n) at different states in (CX3CR1-GFP+/-; Adrb2-/-) mice. Scale bar, 30 μm.

One-way ANOVA with Tukey’s post-hoc test in m; Friedman test with Dunn’s post-hoc test in l, n; n = 15 cells from 6 mice for each group; ns, not significant, **P < 0.01.