Skip to main content
European Commission logo
polski polski
CORDIS - Wyniki badań wspieranych przez UE
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary
Zawartość zarchiwizowana w dniu 2024-06-18

Critical transitions and self-organization in sleep micro-architecture

Final Report Summary - CRITICALSLEEP (Critical transitions and self-organization in sleep micro-architecture)

The specific aim of the project was to (1) elucidate the basic mechanisms leading to Self-Organized Criticality (SOC) in sleep (i.e. the coexistence of scale-invariant (power-law) organization of arousals and scale-specific (exponential) structure of the sleep-stage durations during healthy sleep); (2) uncover how pathological conditions affect this complex SOC-type organization of arousals and sleep-stage transitions at short, sub-ultradian time scales; (3) derive novel SOC-based diagnostic markers of sleep disorders, which could be utilized to quantitatively assess effects of pharmacological treatments.

This specific aim leads to four main hypotheses:

1. Brief arousals are an integral part of healthy sleep regulation, and are generated by an SOC-type mechanism that also governs sleep-stage transitions. We will test that sleep dynamics exhibit a previously unrecognized architecture at small and intermediate scales (lower then 1-2 hours), which has no known parallels in other physiological and biological systems, and relates to the non-equilibrium critical behavior exhibited by only a few physical systems described by SOC.

2. SOC patterns in sleep dynamics relate to intrinsic and basic elements of sleep regulation, and thus, may change and even break down under different sleep disorders. Specifically, we will test which aspects of SOC change under given sleep disorders.

3. Disrupting specific sleep/wake regulatory pathways will affect SOC in sleep dynamics. We will examine which biochemical pathways of the interaction between sleep- and wake-promoting neurons may be responsible for the emergent SOC behavior at the system level.

4. Specific neuronal interactions and network topologies are necessary for transcending local signaling rules at the neuronal level into global SOC dynamics at the system level. We will apply models from statistical physics and develop approaches derived from
complex networks theory to investigate how SOC dynamics emerge from local neuronal interactions.

Overall Research Accomplishments

Related to Hypothesis 1:

- Sleep-stage and arousal transition dynamics and distributions in healthy human subjects as a function of age

We examined the distributions of wake and sleep bouts during night-time sleep from 198 healthy human subjects. Data were obtained from the EU-project SIESTA. The subjects were categorized in 6 age groups from 20 years old until above 70 years old. We found that the power-law distribution for wake bouts as well as the exponential distribution for sleep bouts is preserved across all age groups. However, our preliminary results suggest, that the power-law exponent and the characteristic time scale for the exponentially-distributed sleep bout durations decrease with age.

- SOC architecture in bursting activities of alpha- and delta- waves

We investigated the EEG recordings of the same group of subjects in order to probe the neurophysiological origin of SOC dynamics in sleep regulation and to confirm our results at shorter time scales of 5 sec (sleep micro-architecture) and independently of sleep-stage scoring (sleep stages are scored in 30 sec epochs and therefore neglect brief arousals that are shorter than 15 sec). We analyzed EEG brain activity in the alpha (8Hz-12Hz) and delta (0.5Hz-4Hz) frequency bands that are traditionally associated with arousals/quiet wake and deep sleep, respectively. Despite a large body of electro-physiological investigations of sleep, the temporal organization and interrelation of bursts of brain activity in these two frequency bands with opposing physiological function (wake vs sleep) is not known. We found the largest differences in alpha and delta activity in the occipital region, and by focusing on the corresponding EEG electrodes, we defined a alpha burst as a period of time when the ratio of alpha versus delta spectral power is greater than 1. Accordingly, a burst in delta activity was defined as a period of time when the ratio of alpha versus delta spectral power is smaller than 1. We observed that the micro-architecture of bursting activity of alpha- and delta-waves persists across a wide range of time scales. While the fine micro-structure of alpha- and delta-bursts notably changes by increasing the window of observation from 5 seconds to 30 seconds, the power-law distribution for alpha-bursts and the exponential form of delta-bursts remains stable with power-law exponent A and exponential time constant tau practically unchanged. Such stability indicates a robust SOC-type micro-architecture in brain dynamics that previously has not been observed.

In conclusion to Hypothesis 1, our research indicates that brief arousals and in particular the SOC-type organization of wake/sleep bouts is likely to be an integral part of healthy sleep. Our research also indicates that traditional sleep scoring in 30 sec windows largely neglects the micro-architecture of sleep/wake dynamics.

Related to Hypothesis 2:

- Changes in SOC patterns of arousal and sleep distributions with sleep disorders

Our work on Hypothesis 1 indicated that the traditional sleep/arousal scoring in 30 sec windows largely neglects the SOC-type micro-architecture. We therefore developed new rules to score arousals in windows of 5 sec, and rescored the sleep data of 70 healthy subjects as well as 70 subjects with insomnia traits. In these subjects, we compared the SOC-derived markers of power-law exponent and characteristic time scale and found that in insomnia subjects SOC patterns are preserved and do not break down. However, the SOC markers change with insomnia and show a smaller scaling exponent for the arousal/wake distribution. A similar behavior we found previously in patients with sleep apnea.

In conclusion to Hypothesis 2, the SOC patterns in sleep dynamics (power-law vs. exponential) are preserved even for sleep disorders. The SOC-derived markers (power-law exponent of arousal durations, characteristic time scale of sleep durations) change with sleep disorders, however, our preliminary results suggest that this change is not specific to the kind of sleep disorder.

Related to Hypothesis 3:

- Effects of lesions of the Locus Coeruleus (LC) on SOC in sleep

In order to probe the role of wake/arousal promoting neuronal groups in generating SOC dynamics and sleep micro-architecture, we investigated a rat model with lesions in the locus coeruleus (LC) brain area, where arousal-promoting (norepinephrine) neurons are deleted through injection of neurotoxins. We investigated a group of rats with LC lesions recorded over multiple days, and we compared with control rats where rats were injected saline solution (instead of neurotoxins). We studied EEG/EMG recordings from: 7 control rats and 6 LC lesion rats. EEG was recorded from two contralateral (frontal left vs. occipital right and frontal right vs. occipital left) screw electrodes while the EMG signal was recorded from two multi-stranded wires inserted bilaterally into the nuchal muscles. For both groups, we analyze the data during Dark period (7pm-7am) when rats are more active and Light period (7am-7pm) when rats are predominately sleeping to account for possible circadian effects. We discovered a remarkable temporal organization of arousals that is characterized by a power-law in the range of time scale from 10sec to 10min, indicating a scale-invariant behavior with absence of characteristic/dominating time scale in this regime. Our working hypothesis was that lesioning LC, one of the major wake-promoting brain areas, will lead a dramatic change in this scale-invariant temporal organization. Surprisingly, our investigation shows that LC lesion do not affect this scale-invariant micro- architecture in arousals and short wake bouts during sleep. Remarkably, the dynamics of short arousals and wake in both control and LC lesioned rats are characterized by the same scaling exponent A = 2, indicating that LC related wake-promoting mechanism is not involved in this scale-invariant organization of arousals, and that different mechanisms apart from LC may be responsible for generation and maintenance of arousal/wake events on scales < 10min. Thus, this result triggers new questions about the origin of the scale-invariant organization and the mechanisms that control them which are not addressed in current sleep research.

- Effects of Ventrolateral Preoptic Area (VLPO) lesions on SOC in sleep

To probe the role of sleep-promoting neuronal groups and brain-functional areas in generating SOC dynamics and sleep micro-architecture we investigated a rat model with lesions in the ventrolateral preoptic area (VLPO), where sleep-promoting neurons are deleted. We investigate a group of 6 rats with VLPO lesions recorded over multiple days, and compared with 7 wild-type control rats. The experimental protocol and data recording/scoring was similar to the LC experiment described above. While in the LC experiment we found that LC lesions do not affect the scale-invariant micro- architecture in arousals but instead the sleep distributions, in the VLPO experiment we found the opposite: the exponential distribution of sleep stage durations is preserved with exactly the same characteristic time scale, whereas the power-law exponent for the arousal distribution decreases significantly.

In conclusion to Hypothesis 3, our work shows the first examples of SOC type sleep micro-architecture where SOC type behavior is observed in healthy sleep dynamics and even when key elements in sleep regulation, LC and VLPO, are destroyed. Furthermore, our findings show a first evidence that arousals are intrinsic and that arousals and sleep are part of the same regulatory mechanism.

Related to Hypothesis 4:

- A stochastic modeling approach to address non-equilibrium aspects of sleep micro-architecture

Our studies of the SOC-type micro-architecture in sleep dynamics represent a new framework in sleep research that focuses on short time scales. Furthermore, the presence of power-law scaling in wake durations is indicative of an underlying non-equilibrium process. Current modeling efforts of understanding sleep-wake dynamics focus either on physiologically-motivated models that operate according to the principle of sleep homeostasis at large time scales of hours, or on stochastic models within the Markov framework. Neither of these approaches account for the power-law type temporal organization observed in the sleep-wake micro-architecture.

In our approach, we propose a new stochastic framework to account for the micro-architecture in sleep dynamics. We employ a two-state stochastic model with in-state memory that differentiates between the effects of maintaining a state versus initiating a state. With this model we demonstrate the origin of the empirically observed power-law scaling in the the arousal duration distribution and its connection to sleep inertia. Moreover, this new framework provides quantitative definitions for key sleep parameters including sleep quality, sleep fragmentation and sleep inertia. Introducing the in-state memory also solves the problem encountered by the traditional Markov description of sleep dynamics, where only exponential type duration statistics is present. The model also elucidates the origin of the power-law distribution as a non-equilibrium process of wake consolidation on short time scales that is consistent with the concept of sleep inertia defined from the sleep physiology intuition. Furthermore, our approach provides an analytical bases to quantify key sleep parameters such as sleep inertia, sleep quality and sleep fragmentation. Even though the proposed two-state stochastic description is still at the phenomenological level, it still provides some analytical intuition and can be utilized for future modeling effort to explore how SOC-like behavior emerges from microscopic neuronal mechanisms.

In conclusion to Hypothesis 4, we showed by means of a two-state stochastic model that the micro-architecture in sleep dynamics is the result of a non-equilibrium process and cannot be understood by the traditional approach of sleep homeostasis. This simple model can be used as a first step towards more sophisticated modeling approaches that take into account neuronal interactions and the topologies of the network of sleep and wake-promoting neuronal centers in the brain.