Synchronicity of pyramidal neurones in the neocortex dominates isoflurane-induced burst suppression in mice

Abstract

Background: Anaesthesia-induced burst suppression signifies profound cerebral inactivation. Although considerable efforts have been directed towards elucidating the electroencephalographic manifestation of burst suppression, the neuronal underpinnings that give rise to isoflurane-induced burst suppression are unclear.

Methods: Electroencephalography combined with micro-endoscopic calcium imaging was used to investigate the neural mechanisms of isoflurane-induced burst suppression. Synchronous activities of pyramidal neurones in the auditory cortex and medial prefrontal cortex and inhibitory neurones in the auditory cortex (including parvalbumin [PV], somatostatin [SST], and vasoactive intestinal peptide [Vip]) and subcortical regions (including the medial geniculate body, locus coeruleus, and thalamic reticular nucleus) were recorded during isoflurane anaesthesia. In addition, the effects of chemogenetic manipulation inhibitory neurones in the auditory cortex on isoflurane-induced burst suppression were studied.

Results: Isoflurane-induced burst suppression was highly correlated with the synchronous activities of excitatory neurones in the cortex (∼65% positively and ∼20% negatively correlated neurones). Conversely, a minimal or absent correlation was observed with the neuronal synchrony of inhibitory interneurones and with neuronal activities within subcortical areas. Only activation or inhibition of PV neurones, but not SST or Vip neurones, decreased (P<0.0001) or increased (P<0.0001) isoflurane-induced neuronal synchrony.

Conclusions: Isoflurane-induced burst suppression might be primarily driven by the synchronous activities of excitatory pyramidal neurones in the cortex, which could be bidirectionally regulated by manipulating the activity of inhibitory PV interneurones. Our findings provide new insights into the neuronal mechanisms underlying burst suppression.

Keywords: auditory cortex; burst suppression; inhibitory neurones; isoflurane; micro-endoscopic calcium imaging; neuronal synchrony.

Fig 1

Neuronal synchrony in the auditory cortex during isoflurane-induced burst suppression. (a) Schematic of simultaneous micro-endoscopic calcium imaging and EEG recording in the AC under isoflurane anaesthesia. (b) The identification of isoflurane-induced burst suppression. (c) (top) Confocal image showing GCaMP expression in deep layers of the AC; (middle) field of view (FOV) with all regions of interest (ROIs) highlighted by red; and (bottom) CNMF-E of representative neurones. (d) (top) EEG envelope and corresponding burst events and (bottom) simultaneous Ca²⁺ signal traces for example positive correlation (neurones 1–10), negative correlation (neurones 15–18), and irrelevant (neurones 11–14) neurones from the same FOV shown in (c). (e) Spectrogram of EEG (top) and heatmap of z-scored Ca²⁺ traces (bottom) from the same data shown in (d). (f) Pearson's cross-correlation coefficients matrix of example neurones shown in (d); the correlation coefficients in the first line (row) correspond to the average population Ca²⁺ signal. (g) Two representative neurones showing a positive correlation (neurone 1) and a negative correlation (neurone 18) with isoflurane-induced burst suppression. (h) Area under the receiver operating characteristic curve (AUROC) of the neurones shown in (g). (i) Distribution of the AUROC of all imaged neurones in the AC (217 cells from seven mice). The dashed line indicates the AUROC thresholds for defining neurones with positive (0.65) and negative (0.35) correlations. (j) Proportions of positive, negative, and irrelevant neurones. (k) Distribution of the coefficient of variation (CV) of contributions during each burst event for burst event-related neurones.

Fig 2

Dependency of burst suppression-related neuronal synchrony on the concentration of isoflurane. (a) Examples of EEGs showing the burst suppression pattern (left) and corresponding spectrograms (right) under isoflurane anaesthesia at concentrations of 1.0 vol%, 1.5 vol%, and 2.0 vol%. (b) Isoflurane concentration drawn as a function of number of burst events (left) (n=6 for each test) and temporal proportion of burst events (right) (n=6 for each test). The red arrow indicates the isoflurane concentration corresponding to the maximum number of burst events (left) or the temporal proportion of burst events (right). (c) Isoflurane concentration drawn as a function of the loss of righting reflex (LORR) (n=18). The red arrow indicates the LORR level at the isoflurane concentration corresponding to the maximum number of burst events. (d) EEG spectrograms with bursts marked (top) and heatmaps of z-scored neuronal calcium traces (bottom) under different isoflurane concentrations. (e) Pearson's cross-correlation coefficients matrix of calcium activity for example neurones shown in (d). (f) Distribution of correlation coefficients between the population average of ΔF/F signals and the EEG envelope across mice. (g) Schematic illustration of the random forest regression model. (h) Comparison of the actual EEG envelope with the predicted EEG envelope by the random forest regression model. (i) R² of the random forest regression model across mice.

Fig 3

Burst suppression-related synchronised activity in the medial prefrontal cortex. (a) Schematic of EEG recordings in multiple locations. (b) Example EEG signals simultaneously recorded from medial prefrontal cortex (mPFC) (left), retrosplenial cortex (RSC) (middle), and auditory cortex (AC) (right) with 1.5 vol% isoflurane anaesthesia showing similar burst suppression patterns. (c) Average Pearson's cross-correlation coefficients matrix of EEG signals among the mPFC, RSC, and AC. (d–g) Similar to Fig. 1c–f, but for the mPFC. (h–j) Similar to Fig. 1i–k, but for the mPFC (256 cells from seven mice). (k) Comparison of the cumulative curves of neuronal synchrony indices between the AC and the mPFC.

Fig 4

Burst suppression-related synchronised activity is weak in parvalbumin (PV) inhibitory neurones and absent in both somatostatin (SST) and vasoactive intestinal peptide (Vip) neurones. (a) Burst suppression-related heatmap of z-scored calcium traces of PV-positive interneurones in the auditory cortex (AC). (b) Pearson's cross-correlation coefficients matrix of example neurones shown in (a). (c) Distribution of correlation coefficients between the population average of ΔF/F signals of recorded PV and the EEG envelope across mice (n=7). (d) Distribution of the area under the receiver operating characteristic curve (AUROC) of all imaged PV neurones in the AC (156 cells from seven mice). (e) Proportions of PV cells exhibiting positive, negative, and irrelevant burst suppression-related synchronised activity. (f–j) Similar to (a–e), but for SST neurones (155 cells from six mice). Note that there was no burst suppression-related synchronised activity in SST neurones. (k–o) Similar to (a–e), but for Vip neurones (110 cells from five mice). Note that there was also no burst suppression-related synchronised activity in Vip neurones.

Fig 5

Burst suppression-related synchronised activity is weak in both locus coeruleus (LC) and thalamic reticular nucleus (TRN) neurones and absent in medial geniculate body (MGB) neurones. (a) (top) Schematic of simultaneous micro-endoscopic calcium imaging and EEG recording in LC under isoflurane anaesthesia; (middle) an example field of view (FOV) with all ROIs highlighted in red; and (bottom) CNMF-E of example neurones. (b) Burst suppression-related heatmap of z-scored calcium traces of recorded neurones in the LC. (c) Pearson's cross-correlation coefficients matrix of example neurones shown in (b). (d) Distribution of correlation coefficients between the population average of ΔF/F signals of recorded LC neurones and the EEG envelope across mice (n=7). (e) Distribution of the area under the receiver operating characteristic curve (AUROC) of all imaged neurones in the LC (96 cells from seven mice). (f) Proportions of LC neurones exhibiting positive, negative, and irrelevant burst suppression-related synchronised activity. (g–l) Similar to (a–f), but for TRN recordings (68 cells from seven mice). (m–r) Similar to (a–f), but for MGB recordings. Note that there was no burst suppression-related synchronised activity in the MGB (167 cells from six mice). (s) Comparison of the proportion of neurones with burst suppression-related synchronised activity among different cell types in the cortex and various subcortical brain regions. Data are presented as mean (sd), ∗P<0.05, ∗∗P<0.01, ∗∗∗P<0.001, ∗∗∗∗P<0.0001, Kruskal–Wallis test with Dunn's multiple comparisons. (t) Similar to (s), but for neuronal synchrony indices. Data are presented as median (25th–75th percentiles). (u) Comparison of coefficient of variation for burst event-related neurones among different cell types in the cortex and various subcortical brain regions. ∗∗P<0.01, ∗∗∗∗P<0.0001, using the Kruskal–Wallis test with Dunn's multiple comparisons.

Fig 6

Isoflurane-induced neuronal synchrony could be regulated by parvalbumin (PV), but not by somatostatin (SST) or vasoactive intestinal peptide (Vip) neurones. (a) Schematic of viral injection strategy for chemogenetic inhibition (hM4Di) of PV interneurones and micro-endoscopic calcium imaging of pyramidal neurones in the auditory cortex (AC). (b) Confocal image showing expression of mCherry in PV and GCaMP8 in excitatory neurones within the AC region. (c) Raw trace showing a decrease of firing rate in a hM4Di-expessing PV neurone after clozapine N-oxide (CNO) application. Whole-cell current-clamp recording performed in a slice preparation. (d) Average firing rates before and after perfusion of CNO. Data are presented as mean (sd), Wilcoxon signed-rank test (∗∗∗P<0.001, Z=3.33), n=8 brain slices from three mice. (e) Isoflurane-induced synchronised activity of example excitatory neurones with 1.0 vol% isoflurane anaesthesia with and without chemogenetic inhibition of PV neurones. (f) Cumulative fraction of neuronal synchrony indices of excitatory neurones with and without chemogenetic inhibition of PV neurones. (g) Comparison of the proportion of neurones with isoflurane-induced synchronised activity between with and without chemogenetic inhibition of PV neurones. Data are presented as mean (sd), unpaired t-test (∗∗∗∗P<0.0001, t=–9.34), n=13. (h–n) Similar to (a–g), but for chemogenetic activation (hM3Dq) of PV interneurones. For (k), Wilcoxon signed-rank test (∗∗P=0.002, Z=–3.12), n=8 brain slices from three mice. For (n), Unpaired t-test (∗∗∗∗P<0.0001, t=11.39), n=13. (o) Cumulative fraction of neuronal synchrony indices of excitatory neurones with and without chemogenetic inhibition of SST neurones. (p) Comparison of the proportion of neurones with isoflurane-induced synchronised activity between with and without chemogenetic inhibition of SST neurones. Unpaired t-test (P=0.97, t=–0.04), n=6. (q–r) Similar to (o–p), but for chemogenetic activation of SST interneurones. For (r), unpaired t-test (P=0.59, t=0.55), n=6. (s–t) Similar to (o–p), but for chemogenetic inhibition of Vip interneurones. For (t), unpaired t-test (P=0.32, t=–1.06), n=5. (u–v) Similar to (o–p), but for chemogenetic activation of Vip interneurones. For (v), unpaired t-test (P=0.07, t=2.00), n=6.