Anesthetics fragment hippocampal network activity, alter spine dynamics, and affect memory consolidation

Abstract

General anesthesia is characterized by reversible loss of consciousness accompanied by transient amnesia. Yet, long-term memory impairment is an undesirable side effect. How different types of general anesthetics (GAs) affect the hippocampus, a brain region central to memory formation and consolidation, is poorly understood. Using extracellular recordings, chronic 2-photon imaging, and behavioral analysis, we monitor the effects of isoflurane (Iso), medetomidine/midazolam/fentanyl (MMF), and ketamine/xylazine (Keta/Xyl) on network activity and structural spine dynamics in the hippocampal CA1 area of adult mice. GAs robustly reduced spiking activity, decorrelated cellular ensembles, albeit with distinct activity signatures, and altered spine dynamics. CA1 network activity under all 3 anesthetics was different to natural sleep. Iso anesthesia most closely resembled unperturbed activity during wakefulness and sleep, and network alterations recovered more readily than with Keta/Xyl and MMF. Correspondingly, memory consolidation was impaired after exposure to Keta/Xyl and MMF, but not Iso. Thus, different anesthetics distinctly alter hippocampal network dynamics, synaptic connectivity, and memory consolidation, with implications for GA strategy appraisal in animal research and clinical settings.

Fig 1. LFP recordings in dorsal CA1 during wakefulness and anesthesia reveal distinct and complex alterations by Iso, Keta/Xyl, and MMF.

(A) Experimental setup. Extracellular electrical recordings in dorsal CA1 were performed in 4 head-fixed mice for 105 minutes, continuously. Each animal was recorded under all anesthetics as indicated in the scheme. Order of anesthetics was pseudo-randomized. (B) Characteristic LFP recordings during wakefulness and under 3 different anesthetics. (C) Color-coded heat maps depicting relative change (upper and middle panels) for LFP power and motion profiles (lower panels) for the 3 different anesthetic conditions. Upper panels display LFP power for 0–100 Hz frequency range, lower panels for 0–4 Hz. (D) Line plot displaying LFP power spectra for the 2 time periods indicated by horizontal black bars. For comparison, the 15-minute spectrum of the awake period before anesthesia induction is plotted in both graphs. Statistical differences are indicated in S1C Fig. (E) Line plot displaying the power-law decay exponent (1/f) of the LFP power spectrum for the 30–50 Hz range. Lines display mean ± SEM. (F) Line plot displaying the fraction of active periods compared to the preanesthetic wakeful state, in 15-minute bins throughout the entire recording duration. Lines display mean ± SEM. (G) Heat map displaying PAC for preanesthetic wakeful state (left) and for the indicated time periods during anesthesia. Different bin sizes (0.5 Hz and 1 Hz, separated by vertical black line) are used to resolve low- and high-frequency PAC. Vertical dashed lines in (C) and (E) indicate time points of anesthesia induction (Iso, MMF, and Keta/Xyl) and reversal (Iso and MMF only). Vertical dashed line in (F) indicates time point of anesthesia reversal (Iso and MMF only). Asterisks in (E) and (F) indicate significance of time periods indicated by black horizontal line compared to 15-minute period before anesthesia. Anesthetic conditions are color coded. Asterisks in (G) indicate significant differences compared to the corresponding frequency band during wakefulness. * p < 0.05, ** p < 0.01, *** p < 0.001, n = 4 mice. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure1_S1. FAB, flumazenil/atipamezole/buprenorphine; Iso, isoflurane; Keta/Xyl, ketamine/xylazine; LFP, local field potential; MMF, medetomidine/midazolam/fentanyl; PAC, phase-amplitude coupling.

Fig 2. SUA in dorsal CA1 is strongly reduced during anesthesia and remains significantly altered long after its termination.

(A) Raster plots of z-scored SUA for the 3 different anesthetic strategies in 4 mice. Units are sorted according to initial activity during wakefulness. (B) Line plot of SUA firing rate before, during, and after anesthesia induction. (C) Line plot displaying the fraction of active units compared to the preanesthetic wakeful state, for all 3 anesthetics in 15-minute bins throughout the entire recording duration. (D) Relative change of population firing rate power in the 0–0.5, 0.5–1, and 1–4 Hz frequency band. (E) Line plot displaying the normalized power spectra of population firing rate for the 2 time periods indicated by horizontal black bars. For comparison, the 15-minute spectrum for preanesthetic wakeful state is plotted in both graphs. (F) PPC at low frequencies in the same frequency bands as (D), for the indicated time points during anesthesia. White dots indicate median, and vertical thick and thin lines indicate first to third quartile and interquartile range, respectively. Colored lines in (B)–(D) display mean ± SEM. Vertical dashed lines in panels (A), (B), and (D) indicate time points of anesthesia induction (Iso, MMF, and Keta/Xyl) and reversal (Iso and MMF only). The vertical dashed line in (C) indicates the time point of anesthesia reversal (Iso and MMF only). Asterisks in (B)–(D) indicate significance of periods indicated by black horizontal line compared to period before anesthesia. Anesthetic conditions are color coded. Asterisks in (F) indicate significant differences to wakefulness. * p < 0.05, ** p < 0.01, *** p < 0.001, n = 4 mice. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure2_S2. FAB, flumazenil/atipamezole/buprenorphine; Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; PPC, pairwise phase consistency; SUA, single-unit activity; SUA PWR, power of SUA spike trains.

Fig 3. Repeated calcium imaging in dorsal CA1 reveals distinct activity profiles for Iso, MMF, and Keta/Xyl.

(A) Experimental strategy for chronic calcium imaging of cellular activity in dorsal CA1. For each condition, 7 mice were imaged 4 times for 5 minutes, as indicated by black fields in the scheme. The order of imaging conditions was pseudo-randomized. (B) Time-averaged, 2-photon images of the same FOV in CA1 aligned to the Iso condition. ROIs of automatically extracted, active neurons are overlaid for each condition. (C) Raster plots of z-scored calcium transients in the same animal under different conditions. Traces are sorted by similarity. (D) Violin plots quantifying the number (left), amplitude (middle), and decay (right) of detected calcium transients. White dots indicate median, and vertical thick and thin lines indicate first to third quartile and interquartile range, respectively. (E) tSNE plot summarizing the average calcium transients properties. Each data point represents 1 recording session. Asterisks in (D) indicate significant differences to wakefulness. *** p < 0.001. Note, to facilitate readability, only differences to wakefulness are indicated. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure3_S3. FOV, field of view; Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; ROI, region of interest; tSNE, t-distributed stochastic neighbor embedding.

Fig 4. Calcium activity profiles in neurons active during all conditions are similar between wakefulness and Iso.

(A) Two-photon time-averaged images of the same FOV in CA1, aligned to the Iso condition (same images as in Fig 3). ROIs show neurons active in each condition, allowing direct comparison of calcium transients in the same cells under different conditions. (B) Violin plots quantifying the number (left), amplitude (middle), and decay (right) of detected calcium transients. White dots indicate median, and vertical thick and thin lines indicate first to third quartile and interquartile range, respectively. (C) Heat maps displaying the relative change in the number (left), amplitude (middle), and decay (right) of calcium transients between neurons active in pairs of conditions (see also S6C Fig). (D) Schematic representation of long-term calcium imaging experiments to assess recovery from anesthesia. Black rectangles indicate imaging time points (up to 10 minutes duration each). Filled and open triangles indicate the start and end of the anesthesia period. (E) Line diagrams showing the relative change of the median number of calcium transients (left), their amplitude (middle), and decay constant (right) during anesthesia and recovery relative to the awake state before anesthesia induction. The black bar indicates the anesthesia period. Shaded, colored lines indicate 95% confidence interval. Note, Keta/Xyl anesthesia could not be terminated. The horizontal, colored lines indicate significant difference (p < 0.05) to awake time point (t = 0) for the respective condition. Asterisks in (B) and (C) indicate significant differences to wakefulness. *** p < 0.001. Note, to facilitate readability, only differences to wakefulness are indicated. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure4_S6. FOV, field of view; Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; ROI, region of interest.

Fig 5. Correlation analysis of CA1 calcium activity and SUA shows decorrelation under anesthesia.

(A) Heat maps displaying representative correlation matrices of calcium activity between pairs of neurons during wakefulness and the 3 different anesthetic conditions in the same animal. Matrices are sorted by similarity. (B) Left: Line plot displaying cumulative distribution of Fisher-corrected Pearson correlation coefficients between pairs of neurons (calcium imaging). Center: violin plot displaying the proportion of pairs found in the first (most negative) and fourth (most positive) quartile of the distribution. (C) Line plot displaying the absolute pairwise correlation coefficients over distance (calcium imaging, 25-micrometer bins). (D) Line plot displaying the cumulative distribution of population coupling (calcium imaging). (E) Quantification of correlation between pairs of extracellularly recorded single units using the STTC. Left: Schematic illustration of the STTC quantification. Center: cumulative distribution of the STTC with a 1,000-ms integration window. Right: violin plot quantifying the STTC. In violin plots, white dots indicate median, and vertical thick and thin lines indicate first to third quartile and interquartile range, respectively. Asterisks in (B) and (E) indicate significant differences to wakefulness. *** p < 0.001. Note, only differences to wakefulness are indicated. For comparison between conditions, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure5_S7. Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; STTC, spike time tiling coefficient; SUA, single-unit activity.

Fig 6. Calcium activity in CA1 is temporally and spatially fragmented during anesthesia.

(A) Left: violin plot quantifying the number of PCA clusters during wakefulness or anesthesia, as indicated. Middle: log-log line plot displaying the variance explained by the first 100 components for each condition. Right: violin plot quantifying the power-law slope of the variance explained by the first 100 components for each condition. (B) Left: tSNE plots of network events recorded in the same animal under the 4 indicated conditions. Right: Violin plot quantifying the number of tSNE clusters obtained from calcium recordings during the 4 different treatments. (C) Violin plot quantifying the number of clusters obtained by AP from calcium recordings during the 4 different treatments. (D) and (E) Line plots quantifying the number of detected communities and the modularity of the detected communities with the resolution parameter gamma ranging from 0 to 3. Horizontal lines in violin plots indicate median and first to third quartile. Asterisks in (A)–(C) indicate significant differences to wakefulness. ** p < 0.01, *** p < 0.001. Horizontal lines above plots in (D) and (E) indicate significant difference to wakefulness. Anesthetic conditions are color coded. Note, only differences to wakefulness are indicated. For comparison between conditions, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure6_S8. AP, affinity propagation; Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; PCA, principal component analysis; tSNE, t-distributed stochastic neighbor embedding.

Fig 7. Sleep alters CA1 activity in a similar way to anesthesia but with a lower magnitude.

(A) Classification of activity states during electrical recordings. (B) Characteristic LFP recordings during wakefulness, NREM, and REM sleep. (C) Line plot displaying LFP power spectra for the indicated activity states. (D) Violin plot displaying the power-law decay exponent (1/f) of the LFP power spectrum. (E) Raster plots of z-scored SUA for the 3 different activity states in 4 mice. Units are sorted according to initial activity during wakefulness. (F) Violin plot showing SUA firing rate. (G) Scatter plot showing modulation of SUA firing rate during NREM (light blue) and REM sleep (dark blue) with respect to activity during wakefulness. (H) Violin plot quantifying the STTC. (I) Classification of activity states during CA1 calcium imaging based on eye videography. (J) Raster plots of z-scored calcium transients in an example recording of 1 animal transiting between wakefulness and sleep. Traces are sorted by similarity. (K) Violin plots quantifying the number (left) and amplitude (right) of detected calcium transients. (L) Violin plots quantifying absolute pairwise correlation of all recorded neurons. White dots indicate median, and vertical thick and thin lines indicate first to third quartile and interquartile range, respectively. * p < 0.05, ** p < 0.01, *** p < 0.001 w.r.t. to wake state, n = 3–7 mice. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure7_S9. LFP, local field potential; NREM, non-rapid eye movement; REM, rapid eye movement; STTC, spike time tiling coefficient; SUA, single-unit activity.

Fig 8. Spine turnover at CA1 pyramidal neurons is distinctly altered by repeated application of Iso, MMF, and Keta/Xyl.

(A) Left: Schematic illustration of in vivo spine imaging strategy. In each animal, spines were imaged on basal dendrites located in S.O., oblique dendrites in S.R., and tuft dendrites in S.L.M. Right: Example showing an oblique dendrite in S.R. imaged chronically during all conditions. The order of anesthetic treatments was pseudo-randomized between mice (see S10A Fig). (B) Dot plots showing quantification of spine turnover (left), spine survival (middle), and spine density (right) under the 4 indicated treatments. Note that spines were imaged on the same dendrites across all conditions. Dots indicate mean ± SEM. Asterisks indicate significant differences to wakefulness in the left and middle panel. In the right panel, asterisks denote significant changes within each treatment compared to day 0. * p < 0.05, ** p < 0.01, *** p < 0.001. (C) Imaging of acute spine dynamics during 4 different conditions. Left: schematic of the experimental timeline. Right: example of dendrite imaged during wakefulness in 10-minute intervals (same dendrite as in A). (D) Dot plots showing quantification of acute spine turnover (left), spine survival (center), and spine density (right) under the 4 indicated treatments. Dots indicate mean ± SEM. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure8_S10. Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl; S.L.M., stratum lacunosum moleculare; S.O., stratum oriens; S.R., stratum radiatum.

Fig 9. Episodic memory consolidation is impaired by MMF and Keta/Xyl, but not by Iso.

(A) Experimental design to test episodic-like memory in a Morris water maze. On days 1 and 2, animals were trained to find the platform in position 1. Reversal learning was performed on day 3 where animals had to learn that the platform was moved to position 2. The training was followed 30 minutes later by a 1-hour period of one of the 4 indicated treatments per group. On day 4, consolidation of the memory for the platform in position 2 was tested. (B) Heat maps showing trajectories of all mice during the first probe trial before reversal learning on day 3 (left column), after reversal learning on day 3 (middle column), and after treatment on day 4 (right column). The position of the target zone is indicated by dashed circles. (C) Scatter plots showing quantification of time spent in the new target quadrant (top) and distance to the new platform (bottom) after reversal learning on day 3 and on day 4. Filled, colored circles indicate individual animals, and white circles indicate mean ± SEM. Asterisks in (C) indicate significant differences between days. * p < 0.05, ** p < 0.01. For full report of statistics, see S1 Table. All datasets of this figure can be found under https://github.com/mchini/Yang_Chini_et_al/tree/master/Stats_Dataset_(R)/datasets/Figure9_S11. Iso, isoflurane; Keta/Xyl, ketamine/xylazine; MMF, medetomidine/midazolam/fentanyl.