A Systems Biology Approach to the Modeling and Control of Circadian Rhythms

Abstract

Circadian rhythms are endogenous, daily, biological oscillations which emerge from two cellular-level transcriptional-translational feedback loops. These rhythms align to environmental cues in order to appropriately time various biological functions, including the sleep-wake cycle, alertness, metabolism, and the immune response. While many of these patterns are observed at the organismal level including oscillations in core body temperature and hormonal levels, the underlying cellular level rhythms in the core circadian clock genes and proteins result in the oscillation of the expression of genes and proteins throughout the body, impacting the efficacy of pharmaceutical therapies that target these molecular species and the performance of biological processes linked to these species. The relationship between this molecular oscillator and the timing of biological functions makes it critical to human health that the phase of the oscillator is appropriately aligned with the environment; circadian misalignment, which may result from jetlag, shiftwork, or disease, has been identified as a risk factor for cardiovascular disease and cancer. For this reason, we aim to develop control strategies that allow us to shift the phase of the circadian oscillator to align it with the environment, while accounting for the internal heterogeneity and external noise that is inherent to biological systems. To do so, we use a model predictive control (MPC) framework, which is well suited for the nonlinear nature of oscillating systems, to determine the timing and dosing of inputs to shift circadian phase. Such a framework requires a model of the circadian system, and we develop a physiologically-based model of the molecular-level circadian oscillator that captures the action of known small molecule inputs to the circadian system and allows us to generate predictions about the most effective control targets. We find that inputs that affect the negative feedback loop are able to achieve phase shifts faster than those that affect the positive feedback loop and use this model to demonstrate the potential advantages of using multi-input control strategies. We further adapt our phase shifting MPC algorithm to account for anticipated environmental changes, errors in our ability to sense phase, and heterogeneous responses of populations to our control inputs. All of these considerations are necessary as we continue to develop in silico control approaches to generate hypothesized strategies that are practical to implement and test in vivo. The work presented in thesis represents important steps in the development of personalized, chronotherapeutic techniques for shifting phase at the molecular level that will allow us to reduce circadian misalignment and hence improve human health.