Sleep-Stage-Specific Regulation of Cortical Excitation and Inhibition Niels Niethard, Masashi Hasegawa, Takahide Itokazu, Carlos N. Oyanedel, Jan Born, Takashi R. Sato  Current Biology  Volume 26, Issue 20, Pages 2739-2749 (October 2016) DOI: 10.1016/j.cub.2016.08.035 Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 1 Global Suppression of Cortical Activity during REM Sleep (A) Procedure for wide-field imaging experiments (n = 6 mice). A head-fixed mouse was placed on a treadmill (left). EEG and EMG were recorded simultaneously to distinguish brain states of wake, SWS, and REM sleep (right). (B) (Left) An imaging window was placed over the posterior cortex of one hemisphere. “x” indicates bregma. The 4 × 4 grid shows the 16 ROIs in which fluorescence signal (Fnorm) was analyzed. The star indicates the ROI for which the fluorescence signal is shown in (C) and (D). (C) An example recording of Fnorm and concurrent epochs of wake (green bars), SWS (red), and REM sleep (blue) during one imaging session. Dotted box corresponds to the time window shown enlarged in bottom trace. Note that Fnorm signal decreases shortly after transition into REM sleep (blue bar) and increases again at the end of the REM sleep epoch. (D) The distribution of Fnorm signal during wake (229,993 frames), SWS (140,213 frames), and REM sleep (21,932 frames) for the ROI shown in (B) and (C). (E) Mean Fnorm during wake, SWS, and REM sleep for six mice. In all ROIs, the Fnorm signal is highest during wake, intermediate during SWS, and lowest during REM sleep (p < 0.001). (F) Median Fnorm activity during REM sleep and a 40-s pre-REM sleep baseline after high pass filtering at 0.1 Hz. ∗∗p < 0.001. See also Figures S1, S2, and S5. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 2 Fluorescence Signal from Neurons in Superficial and Deep Cortical Layers (A) Experimental procedures. The virus was injected into superficial layers of the sensorimotor cortex (left), which led to the expression of GCaMP6f in neurons of both superficial and deep layers (right). (B) The signal from GCaMP6f in the neurons can be detected from the dorsal surface. Red circle indicates ROI for which fluorescence signal is indicated in (C). (C) Recording of fluorescence signal (Fnorm) and concurrent epochs of wake (green bars), SWS (red), and REM sleep (blue) during one imaging session. Dotted box corresponds to the time interval shown enlarged in bottom trace. Note decrease in Fnorm signal during REM sleep (blue bar). (D) Mean Fnorm during wake, SWS, and REM sleep in three different mice. (E–H) Panels correspond to (A)–(D) but with the virus injected into the deep layers of the sensorimotor cortex (three mice). Note that expression of GCaMP6f here was confined to the deep layers. See also Figure S5. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 3 In Vivo Two-Photon Calcium Imaging during Sleep (A) In vivo two-photon imaging was performed on a head-fixed mouse that repeatedly went through periods of wake, SWS, and REM sleep during one imaging session. Sleep stages were identified by EEG and EMG. (B) Representative in vivo two-photon images of neurons labeled with GCaMP6f (left) and tdTomato (right) in a PV-Cre mouse. GCaMP6f was expressed pan-neuronally, whereas tdTomato was expressed in a Cre-dependent manner. The scale bar indicates 100 μm. (C) Expression of GCaMP6f (left column) and tdTomato (middle column) was confirmed through histological sectioning. In PV-Cre mice, the expression of tdTomato was confined to PV+ interneurons (PV-INs). Arrows indicate colocalization of Alexa-Fluor-647-conjugated antibody and tdTomato signals. (D) Example recordings of ΔF/F from four neurons indicated in (C). Cells 3 and 4 are labeled with tdTomato, indicating that they are PV-INs. Note that these two PV-INs increased their activity during REM sleep (horizontal bar). (E) The activity (for 169-ms frames; see Experimental Procedures) of 110 unlabeled cells, including mainly putative pyramidal cells, and 17 PV-INs that were imaged simultaneously. The activity of unlabeled neurons, but not PV-INs, is substantially decreased during REM sleep. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 4 Cell-type-Specific Changes in Neural Activity during Wake, SWS, and REM Sleep Left panels: mean ± SEM activity of (A) unlabeled cells (n = 2,042), (B) PV-INs (n = 211), and (C) SOM-INs (n = 117) during wake, SWS, and REM sleep. Middle panels: the distributions of activity for the three cell-types are shown. Note that the distribution for PV-INs during REM sleep contains a long tail, indicating that only a subset of PV-INs contribute to the increased activity during REM. Tick marks at the top of each graph indicate the mean activity. Right panels: three-dimensional histograms indicating the distribution of activity for the cell-types across SWS and REM sleep are shown. x and y axes indicate active frames per seconds during SWS and REM, respectively. z axis indicates the proportion of cells relative to the total number of a cell type. The increased proportion of PV-INs only active during REM sleep, but not during SWS, is shown. See also Figures S3 and S4 and Table S1. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 5 Activity Patterns of Neurons with High versus Low Activity during Wake Left panels: relationship between activity during wake and the difference in activity between REM sleep and SWS (REM − SWS) for unlabeled cells (A), PV-Ins (B), and SOM-INs (C). These values were positively correlated for all three cell types. Light blue lines indicate the regression lines; the dashed lines indicate the border between the top 20% most wake active cells and the rest, as used in the right panels. The cells on the left to this border (with low activity during wake) are shown in light gray. Right panels: mean ± SEM activity during wake, SWS, and REM sleep is shown for a cell subset whose activity during wake was within the top 20% and top 5%, respectively (black), compared with activity of the remaining cells of the respective type (gray). Note that wake active cells display relatively enhanced activity also during REM sleep, which is most apparent for PV-INs. See also Figure S3 and Table S1. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions

Figure 6 Co-activation Patterns of Unlabeled Cells and Inhibitory Interneurons during Wake, SWS, and REM Sleep (A) Upper panels: example scatterplots showing the distribution for the proportion (with reference to total number cells) of unlabeled cells and the proportion of PV-INs that are active within the same imaging frame during a session, separately for periods of wake, SWS, and REM sleep. Lower panels: same, for the proportions of unlabeled cells versus SOM-INs active within the same imaging frame. Co-activation between unlabeled cells and PV-INs and SOM-INs, respectively, across the imaging session is also indicated by the Spearman rank correlation coefficient. (Frames showing neither unlabeled cell activity nor activity of the respective interneuron were omitted in these analyses.) ∗∗p < 0.01. (B) Co-activation between unlabeled cells and PV-INs (upper panel) and SOM-INs (lower panel) expressed as correlation (z-transformed) across imaging sessions (n = 12 and n = 7, respectively), separately for periods of wake, SWS, and REM sleep. ∗∗p < 0.01; ∗p < 0.05 for pairwise comparisons between brain states. Distinctly reduced co-activation between unlabeled cells and PV-INs as well as unlabeled cells and SOM-INs during REM sleep is shown. See also Table S1. Current Biology 2016 26, 2739-2749DOI: (10.1016/j.cub.2016.08.035) Copyright © 2016 Elsevier Ltd Terms and Conditions