"Learning dynamical models of the interactions between the immune receptor NLRP3 and the circadian clock – application to lung cancer"
Lung cancer is a major health problem, with high incidence and mortality rates, due to an absence of effective treatment strategies. In the United States, it is the leading cause of death by cancer, with a 5-year survival of 23% (ACS source). The molecular basis of this disease is complex and heterogeneous, and large inter-patient variability is observed in treatment response, so that it is necessary to consider a mathematical approach to study the processes involved in lung cancer and personalize therapies. In this work, we focus on two deregulated mechanisms in cancer: the immune system and the circadian clock. At the cell level, the circadian clock is a 24-hour biological oscillator that regulates most intracellular processes. It consists of a regulatory network with several intertwine feedback loops that generate sustained oscillations with a period between 20 and 30h. Besides, NLRP3 is a sensor of innate immunity whose role in the immune response has been well studied which was recently identified as an interesting gene altered in lung tumors and predictive of poor prognosis. Previous studies have shown that NLRP3 transcription is regulated indirectly by REV-ERB α (a nuclear receptor of the circadian clock) in macrophages . However, the links between NLRP3 and the circadian clock and in particular the impact of the clock on the function of NLRP3 have been very rarely investigated. Our goal is then to characterize the interactions between NLRP3 and the circadian clock that are emerging as major components in the pathophysiology of lung cancer. To this end, we have undertaken a combined experimental and mathematical approach. We have studied the interactions of NLRP3 and the circadian clock in human bronchial epithelial cells (HBEC) which were synchronized by serum shock. Transcriptomics (RNA-Sequencing) and proteomics (Mass Spectrometry) data as well as intracellular localization (nucleus/cytoplasm) were assessed. Clock gene components were defined using the Reactome database (v84). Circadian rhythms were studied using cosine wave fitting and using CMAES for the minimization task. Model learning method was developed to automatically learn the structure of quantitative systems biology models based on ordinary differential equations from multimodal data. Parameter estimation was performed using a modified least-square approach using CMAES for minimization. The analysis of mRNA levels of 67 clock genes revealed a functional clock in HBEC cells with a period of 30h+/-2h . NLRP3 can interact with clock proteins and the data suggest that they could regulate the intracellular localization of NLRP3 to orchestrate its functions. On the other hand, loss of NLRP3 expression may disrupt the circadian regulation necessary for normal lung function. An existing circadian clock model  using ordinary differential equations (ODE) was extended by adding equations describing the influence of the clock on NLRP3 transcription and interactions of clock and NLRP3 proteins. As a start, a collection of models were considered that included a single additional reaction as compared to the initial clock model. Datasets used for the fit were: mRNA levels of 7 clock genes, protein level of 7 clock genes and circadian rhythms of nucleus/cytoplasm localization of NLRP3, BMAL1, PER2 and CRY1. A systematic fit of each model was performed which allowed to eliminate unlikely reactions. Models involving more than one additional reaction are being investigated. Such model learning pipeline will help prioritize future experiments to fully determine NLRP3 interactions with the clock and identify potential drug targets to restore NLRP3 functions in NLRP3-altered cancer cells.
 Pourcet, B., Zecchin, M., Ferri, L., Beauchamp, J., Sitaula, S., Billon, C., ... & Duez, H. M. (2018). Nuclear receptor subfamily 1 group D member 1 regulates circadian activity of NLRP3 inflammasome to reduce the severity of fulminant hepatitis in mice. Gastroenterology, 154(5), 1449-1464.  Wang, S., Lin, Y., Yuan, X., Li, F., Guo, L., & Wu, B. (2018). REV-ERBα integrates colon clock with experimental colitis through regulation of NF-κB/NLRP3 axis. Nature communications, 9(1), 1-12.  J. Hesse, J. Martinelli, O. Aboumanify, A. Ballesta, and A. Relogio. A mathematical model of the circadian clock and drug pharmacology to optimize irinotecan administration timing in colorectal cancer. Computational and structural biotechnology journal, 19:5170–5183, 2021.
Additional authors: Léa Bardoulet, INSERM U1052 - CNRS 5286, Cancer Research Center of Lyon, France ; Julien Martinelli, Probabilistic Machine Learning Group - Aalto University, Finland ; Samuel Bernard, DRACULA team, INRIA, Lyon ; Anne-Laure Huber, INSERM U1052 - CNRS 5286, Cancer Research Center of Lyon, France ; Annabelle Ballesta, Institut Curie - INSERM Unit 900, Paris, France