Sleep in mammals is regulated on multiple temporal and spatial scales. Recent experimental findings have revealed the importance of neuronal sleep/wake control systems in the brainstem and hypothalamus operating on the millisecond timescale that affect the organism's behavior on the hour timescale. These systems include mutually inhibitory sleep-promoting and wake-promoting nuclei that together comprise the "sleep/wake switch". It has been proposed that inputs to this switch, including the output of the master circadian pacemaker located in the suprachiasmatic nucleus (SCN) of the mammalian hypothalamus, and the sleep homeostatic process produce the observed sleep/wake cycles. This sleep/wake switch theory has become the basis for understanding some sleep disorders (e.g., narcolepsy) and the effects of aging on sleep. In the last 6 years, several mathematical models of this system have been proposed at different spatial scales, including neural mass (neuronal population level) and neuronal network (single neuron level) models. These models have been used for relating behavioral phenotypes to underlying physiology and testing assumptions about our theoretical models of sleep. However, the existing models are limited in their spatial and temporal scales, making it impossible to study multi-scale interactions, e.g., interactions between chronic sleep restriction (long timescale) and sleep fragmentation (short timescale). Moreover, model assumptions about the spatial and temporal properties of the underlying physiology have not been tested. Mathematical models have proven important in defining normal and abnormal physiological relationships, and testing potential treatments in many health- related areas. Therefore, integrating models of sleep across spatial and temporal scales will improve understanding of sleep physiology and pathophysiology, and be instrumental in designing interventions. To achieve integration of sleep models across spatial and temporal scales, we will pursue four linked goals. (i) To determine the conditions under which neural mass models and neuronal network models of the sleep/wake switch predict the same dynamics. We will develop a neuronal network version of our existing neural mass model and test the effects of parameter heterogeneity, network size, and network connectivity. (ii) To test the hypothesis that sleep is inherently fragmented on short timescales due to noisy input to the sleep/wake switch. To achieve this, we will incorporate noisy inputs into both neural mass and neuronal network models and compare predictions to human data from an inpatient protocol. (iii) To test the hypothesis that chronic sleep restriction effects on sleep and performance are due to an interaction between the sleep/wake switch and long timescale dynamics of adenosine receptors. To achieve this, we will add long timescale dynamics of adenosine receptor density to both of our models;up-regulation of A1 receptors has been proposed to underlie the observed sleep and performance responses to chronic sleep restriction. Model predictions will be compared to human data from a chronic sleep restriction protocol. (iv) Sleep/wake switch models predict that the stability of sleep and wake states should decrease as a sleep/wake transition approaches. Consequently, variability in firing rates and voltages of neurons in the sleep/wake switch should increase prior to a transition. We will test this prediction by comparing the predictions of both models to in vivo multi-unit recordings of the sleep/wake switch nuclei in freely behaving rats. Our work will: (i) Develop a standardized approach to modeling sleep physiology. (ii) Lead to the development of predictive models of sleep that are applicable to a wider range of domains, including sleep fragmentation and chronic sleep restriction. (iii) Provide the first direct quantitative test of the sleep/wake switch theory. These important goals are related but can also be achieved independently. Dr. Phillips'strong background in mathematical modeling of sleep and circadian rhythms provides him with the ideal expertise to tackle this project. He will also be greatly assisted by the experimental and mathematical expertise of his mentor, Dr. Klerman, who is head of the Analytic &Modeling Unit within the Division of Sleep Medicine (DSM) at Brigham &Women's Hospital, Harvard Medical School (HMS). The rich research environment at HMS provides extremely fertile ground for interdisciplinary and collaborative projects. In parallel with these research plans, we have prepared a detailed personal development plan to enable Dr. Phillips to build new skills and transition smoothly to independence. Dr. Phillips'immediate goals are to (i) expand his skills in computational neuroscience;(ii) receive focused training in biostatistics;(iii) build his experiece in teaching, mentorship, leadership, and organization;and (iv) produce high impact publications in the fields of sleep and circadian research. His plan to receive focused training in areas (i), (i), and (iii) is described in the proposal. The proposed research plan should also lead to high impact publications, due to its innovative approach, timeliness, and importance. This research and career development plan will build upon Dr. Phillips'existing research and expertise, and will open many new research directions for his career as an independent researcher. The proposed project will grant Dr. Phillips the necessary protected time to conduct research, achieve the necessary experience, and learn the requisite skills to achieve these long-term goals.
The sleep/wake switch in the mammalian hypothalamus and brainstem is a focal point for understanding sleep in both healthy and disease states. Mathematical models of the sleep/wake switch have recently been developed, but they have each covered only limited spatial and temporal domains, and their underlying assumptions have not been strictly tested. Here, we aim to integrate neural mass (neuronal population level) and neuronal network (single neuron level) models of the sleep/wake switch, extend their dynamics to both shorter and longer timescales, and directly test their predictions of sleep/wake dynamics against experimental data.
|Bermudez, Eduardo B; Klerman, Elizabeth B; Czeisler, Charles A et al. (2016) Prediction of Vigilant Attention and Cognitive Performance Using Self-Reported Alertness, Circadian Phase, Hours since Awakening, and Accumulated Sleep Loss. PLoS One 11:e0151770|
|Sano, Akane; Yu, Amy Z; McHill, Andrew W et al. (2015) Prediction of Happy-Sad mood from daily behaviors and previous sleep history. Conf Proc IEEE Eng Med Biol Soc 2015:6796-9|