Two-photon Ca2+ imaging of the same neurons in the anesthetized and awake states.

(A) The primary visual cortex (V1) was first mapped by wide-field Ca²⁺ imaging (left), followed by two-photon cellular-level imaging (middle). On the right, higher-magnification images show jGCaMP7b-expressing neurons without (cell 1) and with tdTomato labeling (cell 2) (B) Representative visual responses of a tdTomato-negative excitatory (cell 1) and a tdTomato-positive inhibitory neuron (cell 2) in the awake (left), anesthetized (middle) and recovery (2nd awake, right) conditions (mean ± SEM). Responses (ΔF/F) were evoked by visual stimuli with various spatial frequencies (1-octave steps, shown on the left) and directions (45°-steps, shown at the bottom). Each gray box indicates the stimulus presentation period (2 s). Top, mean image of the recorded neurons in each condition.

Anesthetize-induced downward shift of the preferred spatial frequency.

(A) Preferred SF of individual neurons in the anesthetized condition (y-axis) is plotted against that in the awake condition (x-axis) for tdTomato-negative excitatory neurons (n = 610 cells, N = 9 mice). Linear mixed model; ***p < 0.001; N.S., not significant. The corner histogram shows the distribution of the pairwise difference in preferred SF between the two conditions. Bottom right: cumulative distributions of preferred SFs in each condition. Kolmogorov–Smirnov test; ***p < 0.001. (B) Preferred SF of excitatory neurons in the recovered condition from the anesthesia (2nd awake, y-axis) is plotted against that in the 1st awake condition (x-axis) (n = 376 cells, N = 5 mice). (C–D) Same plots for tdTomato-positive, inhibitory neurons (C, n = 111 cells, N = 9 mice; D, n = 80 cells, N = 5 mice). (E) Box and raincloud plots comparing the anesthesia-induced changes in preferred SFs for individual neurons between excitatory (Ext.) and inhibitory neurons (Inh.). ΔSF indicates the difference in preferred SF (from the awake state to the anesthetized state) in each neuron. Wilcoxon signed-rank test; ***p < 0.001. (F) Fractions of excitatory and inhibitory neurons exhibiting a significantly decreased (blue), non-significantly (N.S.) changed (light blue or light gray), or significantly increased (gray) preferred SF under anesthesia. Comparison of fractions between excitatory and inhibitory neurons. Chi-square test; *p < 0.05.

Effects of anesthesia on neuronal population coding.

(A–D) A decoder was trained using visual responses in excitatory neurons from the awake condition. (A) Decoding error for visual stimuli as a function of SF from visual responses in the anesthetized (black) and awake conditions (red). The error represents the difference between the decoded and actual orientations. The dotted line indicates the chance level. ***p < 0.001 (anesthesia vs. awake). (B) Distributions of decoder probabilities for the visual responses in the anesthetized (left) and awake (right, control) conditions. The value of 0 on the SF and orientation axes indicates that the decoded stimulus matches the true stimulus. (C–D) Comparison of decoder probability profiles for the anesthetized and awake conditions. The profiles were extracted from (B) along the SF (C) and orientation axes (D), respectively. Under anesthesia, the probability of decoding at higher SFs was increased compared to the awake state, while the orientation did not change.

SOM and PV neurons show downward shifts of the preferred spatial frequency after anesthesia.

(A) Comparison of preferred SFs between the anesthetized and awake states in the SOM neurons. Top, Representative visual responses. Bottom, Preferred SF of individual neurons in the anesthetized condition (y-axis) is plotted against that in the awake condition (x-axis) (n = 60 cells, N = 5 mice). Linear mixed model; ***p < 0.001. The corner histogram shows the distribution of the pairwise difference in preferred SF between the two conditions. Bottom right: cumulative distributions of preferred SFs in each condition. Kolmogorov–Smirnov test; ***p < 0.001. (B) Same plots for the PV neurons (n = 33 cells, N = 6 mice). (C) Box and raincloud plots comparing the anesthesia-induced changes in preferred SFs between the SOM and PV neurons. ΔSF indicates the difference in preferred SF (from the awake state to the anesthetized state) in each neuron. Wilcoxon signed-rank test; *p < 0.05. (D) Fractions of the SOM and PV neurons exhibiting a significantly decreased (blue), non-significantly (N.S.) changed (light blue or light gray), or significantly increased (gray) preferred SF under anesthesia. Comparison of fractions between the SOM and PV neurons. Chi-square test; N.S., not significant.

Manner of anesthesia-induced preferred SF shifts and accompanying changes in visual responsiveness.

(A–D) Population-averaged SF tuning curves for excitatory neurons (A), pan-inhibitory neurons (B), the SOM neurons (C) and the PV neurons (D). Colored lines represent the awake condition, and gray lines represent the anesthetized condition. The error band represents the SEM. (E–H) Box and raincloud plots comparing the bandwidth of SF tuning curve for individual neurons in the anesthetized and awake conditions. (E) The excitatory neurons. (F) The pan-inhibitory neurons. (G) The SOM neurons. (H) The PV neurons. Wilcoxon signed-rank test; *p < 0.05, **p < 0.01, ***p < 0.001; N.S., not significant. (I–L) Similar to E–H, but for response amplitude.

Comparison of visual responses between the anesthetic condition, and low-arousal and high-arousal states during wakefulness.

(A) Estimation of arousal level from pupil area. Top: Example raw pupil trace. Bottom: Normalized pupil area trace. Dotted lines represent the 75th and 25th percentiles, used as thresholds for high- (orange) and low-arousal trials (green), respectively. (B) Box and raincloud plots comparing the visual response amplitude across the anesthetized, low-arousal, and high-arousal conditions (n = 216 cells, N = 6 mice). Wilcoxon signed-rank test with Bonferroni correction; ***p < 0.001. (C) Preferred SF of individual neurons in the anesthetized condition (y-axis) is plotted against that in the high arousal condition (x-axis). Linear mixed model; ***p < 0.001; N.S., not significant. The corner histogram shows the distribution of the pairwise difference in preferred SF between the two conditions. Bottom right: cumulative distributions of preferred SFs in each condition. Kolmogorov–Smirnov test; ***p < 0.001, N.S., not significant. (D–E) Similar to (C), but for comparison between the anesthetized and low-arousal conditions (D) and between the low-arousal and high-arousal conditions (E). (F) Similar to (B), but for SF tuning bandwidth.