Modelling the Circadian Clock

Abstract

Organisms have evolved molecular clocks to track 24-hour day/night light cycles which regulate all important physiological functions including metabolism, immunity, and cell-cycle. Cell-autonomous molecular clocks operate as a transcription-translation feedback loop (TTFL) driven by CLOCK:BMAL1, driving rhythmic transcription of genes including the negative repressors Cry and Per. Delayed negative feedback is initiated by CRY1 directly inhibiting BMAL1, followed by PER1/2 mediating removal of CLOCK:BMAL1 from target E-box DNA sites. Thus, protein-protein interactions are the key rate limiting steps which ensure correct phasing and periodicity. Previous studies have shown that the circadian clock is driven by impressively low abundance mRNA and protein copy number, but how the clock robustly generates cycles with low molecular numbers is unclear. The current TTFL model is based largely on qualitative genetic-based studies and lacks quantitative validation of how specific components of the cellular clockwork interact in time and space within the cell. I investigated how a finite pool of BMAL1 proteins regulate thousands of target sites over 24 hours. From this, I developed a quantitative model of CLOCK:BMAL1 binding DNA using single-cell data for fluorescent fusion protein dynamics and interactions measured using live-cell microscopy and Fluorescence Correlation Spectroscopy (FCS). I found that the approximately 1000 CLOCK:BMAL1 complexes are highly mobile, likely rapidly moving between the far greater number of DNA target sites. Modelling showed that CRY1 complex formation with PER2 regulates the DNA residence time, and that these PER:CRY complexes play a dual role as both transcriptional repressors and enhancers of CLOCK:BMAL1 mobility. This mechanism therefore allows low copy-number clock proteins to regulate a wide repertoire of thousands of gene targets. Despite the clear importance of interactions in the circadian circuit, ex vivo measurements of the affinities of protein-protein interaction are lacking. I therefore undertook a study to measure these for all major components of the molecular clock. This demonstrated that many of the protein-protein interactions in the circadian clock are remarkably strong and well conserved over multiple cell-types and different protein concentrations. Combining these data with modelling, I found that most interactions were direct, without the need for additional facilitating partner proteins, showing the clock operates as a set of serial pairwise interactions. To facilitate the use of FCS, I developed new analytical tools which improve accuracy and robustness of fit. I derived an approximate likelihood model and applied maximum likelihood estimation to directly analyse raw unprocessed FCS data to increase the available data density by three orders of magnitude. This new methodology can infer concentrations and diffusion rates with as little as a few milliseconds of data rather than the current several seconds. Lastly, I present a new application, Network Designer, to enable quicker exploration of models by graphically constructing networks with automated generation of differential and stochastic equations. I used this software throughout this body of work to create mathematical models of the circadian clock. By combining quantitative experiments and modelling of circadian proteins, I have offered new insights into when and how protein-protein and protein-DNA interactions may define the operation and generation of circadian rhythms in mammalian cells.

Date of Award	21 Jul 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Korbinian Strimmer (Co Supervisor), Andrew Loudon (Main Supervisor) & Thomas House (Co Supervisor)