The Cortical States of Wakefulness

Abstract

Cortical neurons process information on a background of spontaneous, ongoing activity with distinct spatiotemporal profiles defining different cortical states. During wakefulness, cortical states alter constantly in relation to behavioral context, attentional level or general motor activity. In this review article, we will discuss our current understanding of cortical states in awake rodents, how they are controlled, their impact on sensory processing, and highlight areas for future research. A common observation in awake rodents is the rapid change in spontaneous cortical activity from high-amplitude, low-frequency (LF) fluctuations, when animals are quiet, to faster and smaller fluctuations when animals are active. This transition is typically thought of as a change in global brain state but recent work has shown variation in cortical states across regions, indicating the presence of a fine spatial scale control system. In sensory areas, the cortical state change is mediated by at least two convergent inputs, one from the thalamus and the other from cholinergic inputs in the basal forebrain. Cortical states have a major impact on the balance of activity between specific subtypes of neurons, on the synchronization between nearby neurons, as well as the functional coupling between distant cortical areas. This reorganization of the activity of cortical networks strongly affects sensory processing. Thus cortical states provide a dynamic control system for the moment-by-moment regulation of cortical processing.

Keywords: acetylcholine; barrel cortex; brain states; sensory processing; synchrony.

Figure 1

State change during whisking in the mouse barrel cortex. (A) Example simultaneous recording of the membrane potential (Vm) of a layer 2/3 pyramidal neuron (Black trace) and local field potential (LFP; Blue trace, shows reversed polarity). The green trace shows the angular position of the contralateral whisker extracted from high-speed video filming (Top images). Adapted from Poulet and Petersen (2008) with permission from Springer Nature. (B) Example Vm recordings from layer 2/3 neurons in the barrel cortex of awake mice during transition from quiet wakefulness (QW) to whisking (whisker angle, green). From left to right: an excitatory neuron (black); a fast-spiking (FS) GABAergic interneuron (red); a non-FS (NFS) GABAergic interneuron (blue); and a somatostatin (SOM) expressing GABAergic interneuron (orange). Adapted from Gentet et al. (2010) with permission from Elsevier and Gentet et al. (2012) with permission from Springer Nature. (C) Left, schematic representation of a simplified local circuit in layer 2/3 barrel cortex. Vasointestinal peptide (VIP) expressing interneurons inhibit parvalbumin (PV) and SOM expressing interneurons. PV interneurons provide perisomatic inhibition onto excitatory pyramidal (PYR) cells, whereas SOM interneurons target preferentially their apical dendrites in layer 1. Right, table summarizing the main effects of the transition from quiet to active wakefulness (AW) on different cell types of the layer 2/3 in the barrel cortex: 1–5 Hz Vm fluctuations; Vm standard deviation (SD); mean Vm; and mean firing rate.

Figure 2

State change during AW across cortical areas. (A) State change during whisking (whisker angular position in green) observed in a layer 2/3 (top, red) and in a layer 5 (bottom, blue) pyramidal neuron recorded in the whisker primary motor cortex (M1) of awake head-fixed mice. Adapted from Sreenivasan et al. (2016) with permission from Elsevier. (B) Vm recording in the Au1 of a freely moving mouse reveal state change during locomotor activity (movement, red). From Schneider et al. (2014) with permission from Springer Nature. (C) Vm recording in the primary visual cortex (V1) of a head-fixed mouse shows similar state change during locomotion. From Polack et al. (2013) with permission from Springer Nature. (D) Left, examples multisite LFP recordings during QW and AW in head-fixed mice. The nuchal electromyogram (EMG, green traces) is used to monitor the overall motor activity. In the top example, LFPs were recorded from the dorsal CA1 region of the hippocampus (dCA1), the medial prefrontal cortex (mPFC), the M1, the secondary (S2) and primary (S1) somatosensory cortices. In the bottom example, LFPs were recorded from dCA1, mPFC, the parietal associative area (PtA), the V1, auditory (Au1) and somatosensory (S1) cortices. Right, the spectral analysis of the LFPs shows a general decrease in LF (1–10 Hz) activity during AW (green) compared to QW (black). Adapted from Fernandez et al. (2017) with permission from Oxford University Press.

Figure 3

Multiple cortical states during wakefulness. (A) Cortical activation/deactivation can occur independently across cortical areas. Example simultaneous LFP recordings (z-scored) from S1, S2, V1 and PtA. The ratio between the low-frequency (LF; 1–10 Hz) and high-frequency (HF; 30–90 Hz) activity of the LFP can be used to assess the level of cortical activation. The LF/HF ratio from the depicted LFPs (Bottom) indicates periods of deactivation in S1 and S2 while PtA and V1 are in an activated state (blue arrowhead) and periods of deactivation in V1 and PtA while S1 and S2 are activated (red arrow heads). Adapted from Fernandez et al. (2017) with permission from Oxford University Press. (B) Plotting the LF/HF ratio against the motor activity (EMG) reveals cortical state fluctuations during QW (low EMG activity). Examples from recordings in the primary (S1, Left) and secondary (S2, Right) somatosensory cortices of an awake mouse. While cortical activation largely dominates during high motor activity, both deactivated and activated (arrowhead) states can be observed during periods of QW. Adapted from Fernandez et al. (2017) with permission from Oxford University Press. (C) Pupil diameter fluctuates during wakefulness. Top, Pupils dilate during locomotion but pupil dilations are also observed during QW in the absence of locomotor activity. Bottom, Pupil dilation is associated to cortical activation. Left, example Vm recording in the V1 of an awake mouse together with the monitoring of the pupil diameter (top trace). The amplitude of the LF Vm fluctuations is shown below (2–10 Hz Hilbert amplitude). Right, mean amplitude of the LF Vm fluctuations plotted as function to the phase of the pupil diameter. Note the low amplitude of the LF fluctuations during the dilating phase compared to the constricting phase. Olive, recordings from wS1; blue, recordings from V1; mauve, recordings from V1 in FVB mice. Adapted from Reimer et al. (2014) with permission from Elsevier.

Figure 4

Cellular mechanisms of the state change in the barrel cortex. (A) The whisker primary somatosensory cortex (wS1) receives three main inputs coming from the thalamus, the cholinergic neurons in the basal forebrain and the whisker M1 (wM1). (B) Thalamic and cholinergic inputs increase activity during whisking. Top, example single-unit recording from a thalamic neuron (Thalamus APs, black) together with LFP recording in wS1 (Cortex LFP, blue, reversed polarity) and monitoring of the whisker position (Whisker angle, green). Adapted from Poulet et al. (2012) with permission from Springer Nature. Bottom, example 2-photon calcium imaging (GCaMP) of cholinergic axons in wS1. Adapted from Eggermann et al. (2014) with permission from Elsevier. (C) Both thalamic and cholinergic inputs contribute to the state change in wS1 during whisking. Top, example control recording of Vm in wS1 during quiet and whisking periods. Middle, example recording in wS1 following pharmacological inactivation of the thalamus. Bottom, example recording in wS1 following inactivation of the thalamus and local blockade of the cholinergic receptors. Note that the simultaneous blockade of the thalamic and cholinergic inputs abolishes the state change during whisking in wS1. Top and middle panels are adapted from Poulet et al. (2012) with permission from Springer Nature; Bottom panel is adapted from Eggermann et al. (2014) with permission from Elsevier.

Figure 5

Local and long-range synchrony. (A) Neuronal synchrony can be assessed by the cross-correlation of the Vm of simultaneously recorded neurons in wS1. Adapted from Poulet and Petersen (2008) with permission from Springer Nature. (B) Motor activity (whisking) decreases the synchronization between nearby neurons. From left to right, mean Vm cross-correlations during QW (black) and whisking (green): between L2/3 excitatory (PYR) neurons; between PYR and GABAergic interneurons (INTs); between INTs; and between PYR and SOM expressing interneurons (SOM) during QW only. Note the antiphase-correlation of the Vm between SOM and PYR. Adapted from Gentet et al. (2010) with permission from Elsevier and Gentet et al. (2012) with permission from Springer Nature. (C) Long-range synchrony can be assessed by the measurement of the coherence between LFPs recorded simultaneously from different cortical areas. Left, the interareal coherence is overall maximal in the LF range during AW and is strongly reduced during non-rapid eye movement (NREM) sleep in awake and naturally sleeping mice. Middle, change in coherence in the LF (0.5–10 Hz) range relative to AW across areas reveals three groups of cortical areas: areas showing a maintenance of, or an increase in, the coherence during NREM sleep compared to QW (red); areas showing the strongest decrease of coherence between QW and NREM sleep (yellow) and areas showing the strongest decrease from AW to QW (gray). Right, change in coherence in the LF range reveals a functional organization of the cortical areas: somatosensory and motor areas that are directly synaptically connected maintain a high coherence during NREM sleep; the other areas maintain coherence throughout wakefulness but not during NREM sleep; the coherence between the two groups drops already during QW compared to AW. Adapted from Fernandez et al. (2017) with permission from Oxford University Press.

Figure 6

Cortical states and sensory processing. (A) Sensory evoked responses recorded in a freely moving rat change as function of behavioral states. Left, example LFP recordings from wS1 during periods of NREM sleep (sleep), QW (awake immobility) and AW (awake exploration). Electrical stimulation of the whisker pad are applied in the different states. Right, sensory-evoked potentials recorded in wS1 in response to the stimulation of the whisker pad in different behavioral states. Note the decrease of the evoked response during exploration compared to immobility. Adapted from Castro-Alamancos (2004) with permission from Elsevier. (B) The early sensory response in wS1 is mostly modulated by behavioral states but not by the behavioral output. Wide field images of the activity of the dorsal cortex using voltage sensitive dye (VSD) imaging in mice performing a whisker-based sensory detection task. From top to bottom: average response for successful trials (Hit) during which the mouse was not whisking before the whisker stimulus; unsuccessful trials (Miss) during which the mouse was not whisking before the whisker stimulus; and unsuccessful trials (Miss) during which the mouse was whisking before the whisker stimulus. Note the strong reduction of the early sensory evoked response, both in wS1 and wM1, when the stimulus occurs when the mouse is whisking (Prestim Whisking), whereas the early response is very similar whether the mouse responded (Hit) or not (Miss) when the stimulus occurs during QW (Prestim Quiet). Adapted from Kyriakatos et al. (2017) with permission from SPIE Digital Library. (C) The sensory-evoked response is strongly modulated by task engagement. Sensory evoked responses are measured in the Au1 of a rat that is engaged in an auditory-discrimination task or is passively exposed to the same auditory stimulus. The sensory-evoked response in Au1 (Multiunit) is markedly reduced when the rat is engaged in the task compared to passive listening. Adapted from Otazu et al. (2009) with permission from Springer Nature.