We recorded the depth profile of local field potentials (LFP) and extracellular multiunit activity in the mouse primary visual cortex and the subiculum using 16 channel linear probes (). We continuously recorded the change in neuronal activity from light to deep anesthesia after increasing isoflurane concentration from 0.7% to 2.5% (wash-in phase) until long periods (>30s) of cortical suppression appeared in the LFP (). Then, we recorded the return of neuronal activity to light anesthesia after decreasing isoflurane concentration back to 0.7% (wash-out phase). In each animal, this wash-in/wash-out cycle was repeated twice to measure ongoing activity and responses to visual and auditory stimulation.
In all mice, LFP activity evolved in the same typical sequence during one wash-in/wash-out cycle (, ). The increase of isoflurane concentration to 2.5 vol % transformed the LFP activity typical for light anesthesia (, first panel) into increasingly synchronized slow rhythmic patterns with gradually increasing peak amplitudes (, second to fifth panel). After several minutes, activity in V1 and the subiculum first synchronized in a common rhythm characterized by short periods of electrical silence, alternating with high amplitude activity (, third panel). Then, silent periods in V1 progressively increased (cortical suppression), leading to an irregular pattern of burst-suppression. In contrast, activity in the subiculum did not express prolonged silent periods, but continued with a regular rhythm of burst-suppression (, fourth panel). The remaining irregularly appearing bursts in V1 synchronized with the regular subicular bursts, and never appeared during the subicular silent periods without a burst in the subiculum. However, subicular bursts appeared also in the absence of cortical bursts.
Differences in ongoing LFP activity in V1 and subiculum.
After the decrease of isoflurane concentration to 0.7 vol %, this sequence of events appeared in reversed order: suppression was followed by burst-suppression, then slow rhythmic activity to finally return into the initial state of small amplitude LFP activity. The return to light anesthesia consistently occurred around five minutes after decrease of isoflurane concentration. Multiunit activity in V1 and the subiculum followed these changes in LFP activity with analogous changes in spiking patterns. LFP bursts in V1 or the subiculum were always accompanied with bursts of strong multiunit activity at the respective electrode sites. This also shows that the described LFPs reflect localized activity of the respective regions ().
Assessment of Anesthesia Depth with Cortical Suppression Ratio
To quantify anesthesia depth during the wash-in/wash-out cycle, we introduced a cortical suppression ratio (CSR), a measure based on the burst suppression ratio to quantify the proportion of suppression in cortical LFP activity (see methods
). The CSR consistently captured the effect of isoflurane concentration on ongoing cortical LFP activity and changed reproducibly in all mice (n
15, ). CSR increased monotonically from a mean of 0.36 (SD 0.02) during low isoflurane concentration at the beginning of the recording towards a mean of 0.84 (SD 0.06) at the end of the wash-in phase. CSR then consistently returned to a mean of 0.35 (SD 0.05) after the wash-out phase. Sensory stimulation did not significantly affect the level of the CSR (, red trace, Wilcoxon rank sum test, p>0.05, ). The time course of changes in the CSR was different between wash-in and wash-out. In both conditions, activity changes occurred faster and more uniform during wash-out than during wash-in (). In order to quantify these differences in the time course, we determined the duration for the change in CSR during wash-in to raise from 0.45 to above 0.6, and compared it to the duration for the CSR to fall from 0.6 below 0.45 during wash-out (). Duration of CSR change between 0.45 and 0.6 was significantly longer during wash-in with 174s (SD 100) than the mean duration of 60s (SD 36) during wash-out (Wilcoxon rank sum test, p<0.01).
Cortical suppression ratio (CSR) and definition of suppression levels.
Cortical suppression ratio during different phases of wash-in/wash-out of isoflurane.
Auditory Responses in Visual Cortex during Deep Isoflurane Anesthesia
In a second condition, alternating visual and auditory stimuli were presented every 5 seconds continuously during a full wash-in/wash-out cycle of isoflurane. To analyze the effect of the visual and auditory stimulation, we classified and pooled responses into one of five suppression levels with respect to their CSR values between light anesthesia/low suppression level (CSR <0.4) and deep anesthesia/high suppression level (CSR >0.7, ). That allowed to average the stimulation trials obtained in each suppression level, and thus to compare the effect of increasing isoflurane concentration on visual and auditory evoked responses in V1 and the subiculum (). The suppression level increases and decreases with anesthetic isoflurane concentration and thus generally corresponds to anesthesia depth, with a low suppression level during light isoflurane anesthesia and a high suppression level during deep isoflurane anesthesia.
Auditory evoked LFP responses in V1 during deep anesthesia.
In V1, visual stimulation evoked responses during all suppression levels. The response was characterized by a short latency response during light anesthesia, which changed into a two-component response with a short latency and a long latency component during deep anesthesia with high suppression level. The mean latency of the first component was significantly reduced from 74ms (SD 16ms) during light anesthesia to 56ms (SD 5ms) during deep anesthesia (direct comparison between Level 1 and 5, Wilcoxon rank sum test, p<0.01, , ). The second long latency component observed during deep anesthesia had a mean latency of 130ms (SD 16ms).
Onset latencies [ms] of sensory evoked activity during different anesthesia levels.
Auditory stimulation had no measurable effects in V1 during light anesthesia, but elicited long latency responses during deep anesthesia at suppression levels above CSR >0.5 (, level 3–5). The mean latencies of these auditory evoked responses increased from 164ms (SD 24ms) at level three to 199ms (SD 16ms) at level five. They were significantly longer than visual response latencies during the respective suppression level (, , Wilcoxon rank sum test, pairwise comparison, p<0.01). At the same time as auditory responses occurred in V1, both visual and auditory stimulation evoked responses in the subiculum with similar latencies (). The latency differences were not significant between stimulation modalities with a mean of 227ms (SD 74ms) for visual and 209ms (SD 33ms) for auditory stimulation during the highest suppression level during deep anesthesia (, Wilcoxon rank sum test, pairwise comparison, p>0.05, ).
In addition to LFP responses, increasing anesthesia depth had a similar effect on multiunit responses in V1 and the subiculum. The changes in excitability were well reflected in the depth profile of multiunit activity (). Visual responses in V1 were present during all anesthesia levels, whereas auditory stimulation first slightly suppressed activity in V1 during light anesthesia, and then elicited auditory multiunit activity with increase of anesthesia depth at the same suppression levels as LFP bursts appeared (CSR>0.5, ). At the same time, bursts could be evoked in the subiculum by visual and auditory stimulation ().
Anesthesia level alters response properties in V1 and the subiculum.
Visual multiunit responses were reliably present at electrode positions located in V1 during light anesthesia in all animals (CSR<0.5, , difference to pre-stimulus baseline activity, two-sided t-test, p<0.05). During deep anesthesia, the reliability of neuronal responses to sensory stimulation decreased and visual or auditory stimulation did not evoke significant activity in all of the animals. However, if visual or auditory responses were present, they evoked activity in all electrodes over the whole shank in an unspecific way, that did not reflect the border between V1 and the subciulum (). Thus, deep layer VI neurons and subicular neurons at the border to white matter seem to become active during synchronized burst events during deep anesthesia, which are inactive during light anesthesia.
We also compared individual depth profiles of multiunit activity of visual, auditory and spontaneous bursts in visual cortex (). Multiunit depth profiles of auditory and spontaneous bursts were generally similar, with earliest activity originating in lower cortical layers. In contrast, earliest onsets of visual evoked bursts were originating from upper and medium layers. The similarity of auditory and spontaneous bursts was also reflected in LFP depth profiles ().
Depth profiles of auditory and spontaneous bursts are similar.
Discrete Change of Response Properties between Trials
To identify the time point of the transition in excitability, we analyzed single trial responses during the wash-in/wash-out cycle. The responsiveness to sensory stimuli between light and deep anesthesia did not change gradually. Rather, the change in response properties appeared within two consecutive trials during wash-in, and disappeared with a similar discontinuity during wash-out (). The change towards auditory excitability in V1 happened in parallel to the change in excitability of the subiculum. This state change could be easily identified in most of the recordings and occurred consistently after CSR increased above 0.5 (mean: 0.58, SD: 0.04), and then disappeared when CSR dropped below 0.5 (mean: 0.53, SD: 0.04). A CSR >0.5 generally indicated the presence of distinct burst-suppression. Sensory stimuli sometimes ceased to elicit activity during prolonged high CSR levels, but auditory evoked burst responses often reappeared during wash-out, shortly preceding the return to the light state (). The return was marked by a sudden strong increase in ongoing multiunit activity most obvious acoustically on the audio monitor of the spike signals. From this point on, only visual evoked responses remained in visual cortex.
Discontinuous change in excitability in V1 and the subiculum between light and deep anesthesia.
Dichotomy of Rhythms between Light and Deep Anesthesia
The two response states could be characterized by the patterns of ongoing activity. Time-frequency analysis revealed a dichotomy in dominant frequencies and recurring patterns of the ongoing LFP (). The light anesthesia state was characterized by theta oscillations in the subiculum with a mean peak frequency of 4.8 Hz (SD 0.3 Hz) (, lower panel). In V1, the light state was marked by transient periods of gamma oscillations with mean peak frequency of 33.9 Hz (SD 2.1 Hz) (, upper panel) and by delta-oscillations with mean peak frequency of 2 Hz (SD 0.4 Hz) (low frequencies are not shown in for better visualization of ongoing gamma oscillations in the color-coded plot). To analyze the frequency of the recurring events of the bursts, we similarly applied time-frequency analysis to the envelope of the LFP signal during the wash-in/wash-out cycle (). The deep anesthesia state was characterized by regular slow rhythmic burst-suppression in the subiculum, with a mean peak frequency of 0.17 Hz (SD 0.05 Hz) (, lower panel). V1 activity did not reflect this rhythm, but exhibited the above mentioned, progressive lengthening of silent periods during increase of isoflurane, and a shortening of silent periods during decreasing levels of isoflurane. This concentration-dependent change of interval length was reflected in a frequency-modulated band in the time-frequency plot of LFPs in V1 (, upper panel). The continuous wash-in/wash-out cycle between high (2.5%) and low (0.7%) isoflurane concentration thus produced two general states with characteristic frequencies occurring on a macroscopic time scale. They could be distinguished by a slow rhythm in the deep state and fast rhythms in the light state.
Time-frequency analysis of ongoing LFP activity.
Build Up in Cortical Correlation Preceding Transition between States
We additionally computed correlations between electrode signals and determined the mean correlation for signals within V1, within subiculum and between V1 and subiculum (). Increase of isoflurane concentration induced a gradual increase of signal correlation in V1. Its maximum indicated the beginning of the burst-suppression phase (). Subiculo-cortical correlation increased similarly, also with a maximum at the onset of burst-suppression. The mean correlation decreased after the onset of burst-suppression. This happened because the signals during the silent periods of suppression mainly reflect uncorrelated background noise. Although signal correlation is high during the bursts, the combination with the interposed suppression periods yields a low mean correlation value. With increasing suppression periods the correlation then decreases accordingly. This behavior was reversed during wash-out, when correlation again first reached a maximum and then decreased towards its lowest values during light anesthesia at the end of the wash-out. In consequence, the peak in the correlation indicates the state transition between light and deep anesthesia, preceded by a gradual change of correlations during the light state.
Intracortical cross-correlation changes with anesthesia level.