Attractor competition enriches cortical dynamics during awakening from anesthesia

Neural models for detection and classification of brain states and transitions

Exploring natural or pharmacologically induced brain dynamics, such as sleep, wakefulness, or anesthesia, provides rich functional models for studying brain states. These models allow detailed examination of unique spatiotemporal neural activity patterns that reveal brain function. However, assessing transitions between brain states remains computationally challenging. Here we introduce a pipeline to detect brain states and their transitions in the cerebral cortex using a dual-model Convolutional Neural Network (CNN) and a self-supervised autoencoder-based multimodal clustering algorithm. This approach distinguishes brain states such as slow oscillations, microarousals, and wakefulness with high confidence. Using chronic local field potential recordings from rats, our method achieved a global accuracy of 91%, with up to 96% accuracy for certain states. For the transitions, we report an average accuracy of 74%. Our models were trained using a leave-one-out methodology, allowing for broad applicability across subjects and pre-trained models for deployments. It also features a confidence parameter, ensuring that only highly certain cases are automatically classified, leaving ambiguous cases for the multimodal unsupervised classifier or further expert review. Our approach presents a reliable and efficient tool for brain state labeling and analysis, with applications in basic and clinical neuroscience.

H-current modulation of cortical Up and Down states

Understanding the link between cellular processes and brain function remains a key challenge in neuroscience. One crucial aspect is the interplay between specific ion channels and network dynamics. This work reveals a role for h-current, a hyperpolarization-activated cationic current, in shaping cortical slow oscillations. Cortical slow oscillations are generated not only during slow wave sleep and deep anaesthesia, but also in association with disorders of consciousness and brain lesions. Cortical slow oscillations exhibit rhythmic periods of activity (Up states) alternating with silent periods (Down states). By progressively reducing h-current in both cortical slices and in a computational model, we observed Up states transformed into prolonged plateaus of sustained firing, while Down states were also significantly extended. This transformation led to a fivefold reduction in oscillation frequency. In a biophysical recurrent network model, we identified the cellular mechanisms underlying this transformation of network dynamics: an increased neuronal input resistance and membrane time constant, increasing neuronal responsiveness to even weak inputs. A partial block of h-current therefore resulted in a change in brain state. HCN (hyperpolarization-activated cyclic nucleotide-gated) channels, which generate h-current, are known targets for neuromodulation, suggesting potential pathways for dynamic control of brain rhythms. KEY POINTS: We investigated the role of h-current in shaping emergent cortical slow oscillation dynamics, specifically Up and Down states, in cortical slices. Blocking h-current transformed Up states into prolonged plateaus of sustained firing, lasting up to 4 s. Down states were also significantly elongated and the oscillatory frequency decreased. A biophysical model of the cortical network replicated these findings and allowed us to explore the underlying mechanisms. An increase in cellular input resistance and time constant led to a rise in network excitability, synaptic responsiveness and firing rates. Our results highlight the significant role of h-current in controlling cortical slow rhythmic patterns, making it a relevant target for neuromodulators regulating brain states.

Norepinephrine-mediated slow vasomotion drives glymphatic clearance during sleep

As the brain transitions from wakefulness to sleep, processing of external information diminishes while restorative processes, such as glymphatic removal of waste products, are activated. Yet, it is not known what drives brain clearance during sleep. We here employed an array of technologies and identified tightly synchronized oscillations in norepinephrine, cerebral blood volume, and cerebrospinal fluid (CSF) as the strongest predictors of glymphatic clearance during NREM sleep. Optogenetic stimulation of the locus coeruleus induced anti-correlated changes in vasomotion and CSF signal. Furthermore, stimulation of arterial oscillations enhanced CSF inflow, demonstrating that vasomotion acts as a pump driving CSF into the brain. On the contrary, the sleep aid zolpidem suppressed norepinephrine oscillations and glymphatic flow, highlighting the critical role of norepinephrine-driven vascular dynamics in brain clearance. Thus, the micro-architectural organization of NREM sleep, driven by norepinephrine fluctuations and vascular dynamics, is a key determinant for glymphatic clearance.

Control of cortical slow oscillations and epileptiform discharges with photoswitchable type 1 muscarinic ligands

Acetylcholine and the cholinergic system are crucial to brain function, including functions such as consciousness and cognition. Dysregulation of this system is implicated in the pathophysiology of neurological conditions such as Alzheimer's disease. For this reason, cholinergic neuromodulation is relevant in both basic neuroscience and clinical neurology. In this study, we used photopharmacology to modulate neuronal activity using the novel selective type-1 muscarinic (M1) photoswitchable drugs: the agonist benzyl quinolone carboxylic acid-azo-iperoxo (BAI) and the antagonist cryptozepine-2. Our aim was to investigate the control over these cholinergic receptors using light and to investigate the effects of these drugs on physiological spontaneous slow waves and on epileptic activity in the cerebral cortex. First, we used transfected HEK cell cultures and demonstrated BAI's preferential activation of M1 muscarinic acetylcholine receptors (mAChRs) compared with M2 mAChRs. Next, we found that white-light illumination of BAI increased the frequency of spontaneous slow-wave activity in brain cortical networks of both active slices and anesthetized mice, through M1-mAChRs activation. Illumination of cryptozepine-2 with UV light effectively suppressed not only the muscarinic-induced increase in slow-wave frequency, but also muscarinic-induced epileptiform discharges. These findings not only shed light on the role of M1 acetylcholine receptors in the cortical network dynamics but also lay the groundwork for developing advanced light-based pharmacological therapies. Photopharmacology offers the potential for high-precision spatiotemporal control of brain networks with high pharmacological specificity in both healthy and pathological conditions.

The temporal asymmetry of cortical dynamics as a signature of brain states

The brain is a complex non-equilibrium system capable of expressing many different dynamics as well as the transitions between them. We hypothesized that the level of non-equilibrium can serve as a signature of a given brain state, which was quantified using the arrow of time (the level of irreversibility). Using this thermodynamic framework, the irreversibility of emergent cortical activity was quantified from local field potential recordings in male Lister-hooded rats at different anesthesia levels and during the sleep-wake cycle. This measure was carried out on five distinct brain states: slow-wave sleep, awake, deep anesthesia-slow waves, light anesthesia-slow waves, and microarousals. Low levels of irreversibility were associated with synchronous activity found both in deep anesthesia and slow-wave sleep states, suggesting that slow waves were the state closest to the thermodynamic equilibrium (maximum symmetry), thus requiring minimum energy. Higher levels of irreversibility were found when brain dynamics became more asynchronous, for example, in wakefulness. These changes were also reflected in the hierarchy of cortical dynamics across different cortical areas. The neural dynamics associated with different brain states were characterized by different degrees of irreversibility and hierarchy, also acting as markers of brain state transitions. This could open new routes to monitoring, controlling, and even changing brain states in health and disease.

Mapping brain state-dependent sensory responses across the mouse cortex

Sensory information must be integrated across a distributed brain network for stimulus processing and perception. Recent studies have revealed specific spatiotemporal patterns of cortical activation for the early and late components of sensory-evoked responses, which are associated with stimulus features and perception, respectively. Here, we investigated how the brain state influences the sensory-evoked activation across the mouse cortex. We utilized isoflurane to modulate the brain state and conducted wide-field calcium imaging of Thy1-GCaMP6f mice to monitor distributed activation evoked by multi-whisker stimulation. Our findings reveal that the level of anesthesia strongly shapes the spatiotemporal features and the functional connectivity of the sensory-activated network. As anesthesia levels decrease, we observe increasingly complex responses, accompanied by the emergence of the late component within the sensory-evoked response. The persistence of the late component under anesthesia raises new questions regarding the potential existence of perception during unconscious states.

Chronic full-band recordings with graphene microtransistors as neural interfaces for discrimination of brain states

Brain states such as sleep, anesthesia, wakefulness, or coma are characterized by specific patterns of cortical activity dynamics, from local circuits to full-brain emergent properties. We previously demonstrated that full-spectrum signals, including the infraslow component (DC, direct current-coupled), can be recorded acutely in multiple sites using flexible arrays of graphene solution-gated field-effect transistors (gSGFETs). Here, we performed chronic implantation of 16-channel gSGFET arrays over the rat cerebral cortex and recorded full-band neuronal activity with two objectives: (1) to test the long-term stability of implanted devices; and (2) to investigate full-band activity during the transition across different levels of anesthesia. First, we demonstrate it is possible to record full-band signals with stability, fidelity, and spatiotemporal resolution for up to 5.5 months using chronic epicortical gSGFET implants. Second, brain states generated by progressive variation of levels of anesthesia could be identified as traditionally using the high-pass filtered (AC, alternating current-coupled) spectrogram: from synchronous slow oscillations in deep anesthesia through to asynchronous activity in the awake state. However, the DC signal introduced a highly significant improvement for brain-state discrimination: the DC band provided an almost linear information prediction of the depth of anesthesia, with about 85% precision, using a trained algorithm. This prediction rose to about 95% precision when the full-band (AC + DC) spectrogram was taken into account. We conclude that recording infraslow activity using gSGFET interfaces is superior for the identification of brain states, and further supports the preclinical and clinical use of graphene neural interfaces for long-term recordings of cortical activity.

Beyond spiking networks: The computational advantages of dendritic amplification and input segregation

The brain can efficiently learn a wide range of tasks, motivating the search for biologically inspired learning rules for improving current artificial intelligence technology. Most biological models are composed of point neurons and cannot achieve state-of-the-art performance in machine learning. Recent works have proposed that input segregation (neurons receive sensory information and higher-order feedback in segregated compartments), and nonlinear dendritic computation would support error backpropagation in biological neurons. However, these approaches require propagating errors with a fine spatiotemporal structure to all the neurons, which is unlikely to be feasible in a biological network. To relax this assumption, we suggest that bursts and dendritic input segregation provide a natural support for target-based learning, which propagates targets rather than errors. A coincidence mechanism between the basal and the apical compartments allows for generating high-frequency bursts of spikes. This architecture supports a burst-dependent learning rule, based on the comparison between the target bursting activity triggered by the teaching signal and the one caused by the recurrent connections, providing support for target-based learning. We show that this framework can be used to efficiently solve spatiotemporal tasks, such as context-dependent store and recall of three-dimensional trajectories, and navigation tasks. Finally, we suggest that this neuronal architecture naturally allows for orchestrating "hierarchical imitation learning", enabling the decomposition of challenging long-horizon decision-making tasks into simpler subtasks. We show a possible implementation of this in a two-level network, where the high network produces the contextual signal for the low network.

M-current modulation of cortical slow oscillations: Network dynamics and computational modeling

The slow oscillation is a synchronized network activity expressed by the cortical network in slow wave sleep and under anesthesia. Waking up requires a transition from this synchronized brain state to a desynchronized one. Cholinergic innervation is critical for the transition from slow-wave-sleep to wakefulness, and muscarinic action is largely exerted through the muscarinic-sensitive potassium current (M-current) block. We investigated the dynamical impact of blocking the M-current on slow oscillations, both in cortical slices and in a cortical network computational model. Blocking M-current resulted in an elongation of Up states (by four times) and in a significant firing rate increase, reflecting an increased network excitability, albeit no epileptiform discharges occurred. These effects were replicated in a biophysical cortical model, where a parametric reduction of the M-current resulted in a progressive elongation of Up states and firing rate. All neurons, and not only those modeled with M-current, increased their firing rates due to network recurrency. Further increases in excitability induced even longer Up states, approaching the microarousals described in the transition towards wakefulness. Our results bridge an ionic current with network modulation, providing a mechanistic insight into network dynamics of awakening.

Complexity of cortical wave patterns of the wake mouse cortex

Rich spatiotemporal dynamics of cortical activity, including complex and diverse wave patterns, have been identified during unconscious and conscious brain states. Yet, how these activity patterns emerge across different levels of wakefulness remain unclear. Here we study the evolution of wave patterns utilizing data from high spatiotemporal resolution optical voltage imaging of mice transitioning from barbiturate-induced anesthesia to wakefulness (N = 5) and awake mice (N = 4). We find that, as the brain transitions into wakefulness, there is a reduction in hemisphere-scale voltage waves, and an increase in local wave events and complexity. A neural mass model recapitulates the essential cellular-level features and shows how the dynamical competition between global and local spatiotemporal patterns and long-range connections can explain the experimental observations. These mechanisms possibly endow the awake cortex with enhanced integrative processing capabilities.

Simulations approaching data: cortical slow waves in inferred models of the whole hemisphere of mouse

The development of novel techniques to record wide-field brain activity enables estimation of data-driven models from thousands of recording channels and hence across large regions of cortex. These in turn improve our understanding of the modulation of brain states and the richness of traveling waves dynamics. Here, we infer data-driven models from high-resolution in-vivo recordings of mouse brain obtained from wide-field calcium imaging. We then assimilate experimental and simulated data through the characterization of the spatio-temporal features of cortical waves in experimental recordings. Inference is built in two steps: an inner loop that optimizes a mean-field model by likelihood maximization, and an outer loop that optimizes a periodic neuro-modulation via direct comparison of observables that characterize cortical slow waves. The model reproduces most of the features of the non-stationary and non-linear dynamics present in the high-resolution in-vivo recordings of the mouse brain. The proposed approach offers new methods of characterizing and understanding cortical waves for experimental and computational neuroscientists.

Assessing brain state and anesthesia level with two-photon calcium signals

Brain states, such as wake, sleep, or different depths of anesthesia are usually assessed using electrophysiological techniques, such as the local field potential (LFP) or the electroencephalogram (EEG), which are ideal signals for detecting activity patterns such as asynchronous or oscillatory activities. However, it is technically challenging to have these types of measures during calcium imaging recordings such as two-photon or wide-field techniques. Here, using simultaneous two-photon and LFP measurements, we demonstrate that despite the slower dynamics of the calcium signal, there is a high correlation between the LFP and two-photon signals taken from the neuropil outside neuronal somata. Moreover, we find the calcium signal to be systematically delayed from the LFP signal, and we use a model to show that the delay between the two signals is due to the physical distance between the recording sites. These results suggest that calcium signals alone can be used to detect activity patterns such as slow oscillations and ultimately assess the brain state and level of anesthesia.

Slow waves form expanding, memory-rich mesostates steered by local excitability in fading anesthesia

In the arousal process, the brain restores its integrative activity from the synchronized state of slow wave activity (SWA). The mechanisms underpinning this state transition remain, however, to be elucidated. Here we simultaneously probed neuronal assemblies throughout the whole cortex with micro-electrocorticographic recordings in mice. We investigated the progressive shaping of propagating SWA at different levels of isoflurane. We found a form of memory of the wavefront shapes at deep anesthesia, tightly alternating posterior-anterior-posterior patterns. At low isoflurane, metastable patterns propagated in more directions, reflecting an increased complexity. The wandering across these mesostates progressively increased its randomness, as predicted by simulations of a network of spiking neurons, and confirmed in our experimental data. The complexity increase is explained by the elevated excitability of local assemblies with no modifications of the network connectivity. These results shed new light on the functional reorganization of the cortical network as anesthesia fades out.

•

Complexity of isoflurane-induced slow waves reliably determines anesthesia level

•

In deep anesthesia, the propagation strictly alternates between front-back-front patterns

•

In light anesthesia, there is a continuum of directions and faster propagation

•

Local excitability underpins the cortical reorganization in fading anesthesia

Thalamo-cortical spiking model of incremental learning combining perception, context and NREM-sleep

The brain exhibits capabilities of fast incremental learning from few noisy examples, as well as the ability to associate similar memories in autonomously-created categories and to combine contextual hints with sensory perceptions. Together with sleep, these mechanisms are thought to be key components of many high-level cognitive functions. Yet, little is known about the underlying processes and the specific roles of different brain states. In this work, we exploited the combination of context and perception in a thalamo-cortical model based on a soft winner-take-all circuit of excitatory and inhibitory spiking neurons. After calibrating this model to express awake and deep-sleep states with features comparable with biological measures, we demonstrate the model capability of fast incremental learning from few examples, its resilience when proposed with noisy perceptions and contextual signals, and an improvement in visual classification after sleep due to induced synaptic homeostasis and association of similar memories.

We created a thalamo-cortical spiking model (ThaCo) with the purpose of demonstrating a link among two phenomena that we believe to be essential for the brain capability of efficient incremental learning from few examples in noisy environments. Grounded in two experimental observations—the first about the effects of deep-sleep on pre- and post-sleep firing rate distributions, the second about the combination of perceptual and contextual information in pyramidal neurons—our model joins these two ingredients. ThaCo alternates phases of incremental learning, classification and deep-sleep. Memories of handwritten digit examples are learned through thalamo-cortical and cortico-cortical plastic synapses. In absence of noise, the combination of contextual information with perception enables fast incremental learning. Deep-sleep becomes crucial when noisy inputs are considered. We observed in ThaCo both homeostatic and associative processes: deep-sleep fights noise in perceptual and internal knowledge and it supports the categorical association of examples belonging to the same digit class, through reinforcement of class-specific cortico-cortical synapses. The distributions of pre-sleep and post-sleep firing rates during classification change in a manner similar to those of experimental observation. These changes promote energetic efficiency during recall of memories, better representation of individual memories and categories and higher classification performances.