(W-017) Integrating mathematical modeling with multiplexed immunofluorescence data to predict drug-induced cell arrest

Maria Rodriguez-Fernandez, Ph.D – Assistant Professor, Institute for Biological and Medical Engineering, Schools of Engineering, Medicine and Biological Sciences, Pontificia Universidad Católica de Chile; Peter Sorger, Ph.D – Otto Krayer Professor of Systems Pharmacology, Department of Systems Biology, Harvard Medical School; Fabian Fröhlich, Ph.D – Group Leader, Dynamics of Living Systems Laboratory, The Francis Crick Institute, London

Objectives: Pharmacological therapies for excessive accumulation of cells aim to generate cell cycle arrest or cell death. Nevertheless, effectivity of these treatments represents a challenge because of the high level of interconnectivity and multiplicity of pathways involved in the cell cycle regulation [1]. Computational tools that allow integrating cell level data with mathematical modeling would provide a deeper understanding of the effect of drugs and help in the choice and timing of drug combinations. Therefore, the main goal of this work was developing a method for integrating multiplexed immunofluorescence data with mathematical modeling to predict drug-induced cell arrest. Specifically, we seek: 1) to reconstruct dynamics of main cell cycle biomarkers using a method based in the work of Gaglia et al. 2) to propose a new approach to integrate multiple drug-induced cell arrest data with cell proliferation dynamics, and 3) to develop a mathematical model capable of integrating both biomarker trajectories during cell proliferation and arrest states induced by palbociclib and nocodazole.

Methods: Using snapshots of non-synchronized cells, ordered cell cycle progression trajectories of the cell cycle markers were generated through the classical multidimensional scaling (CMD). Data from MCF10a cell line untreated and treated with palbociclib or nocodazole from the work by Gaglia et al. were used [2]. To integrate cell arrest data with marker trajectories, a CMD embedding procedure was performed using the untreated data plus small subsets of treated data. The cell cycle progression and drug-induced cell arrest dynamics were used to fit a mathematical model of 6 ordinary differential equations inspired in the model developed by Gerard & Goldbeter [3]. Parameter estimation was performed through the Data2Dynamics modeling environment in Matlab 2023b.

Results: The reconstructed trajectories were found to align with the knowledge about cyclin dynamics throughout the cell cycle, as well as the arrest phases found for palbociclib (G1 phase), and nocodazole (early G1/G1-M checkpoint). The model was able to capture the dynamics of cell cycle and drug-induced cell arrest by palbociclib and nocodazole and, subsequently, predict a G1 arrest for lower values of growth factors. Finally, the model allowed predicting cell arrest and proliferation states under combinations of the effects of nocodazole, palbociclib, and low growth factor levels, showing that a linear threshold between both arrest and proliferation states constrains optimal combinations to a maximum additive cell arrest effect.

Conclusions: This work proposes a new methodology to integrate mathematical modeling with multiple datasets that exhibit continuous cycling and arrest in response to different experimental conditions. The application of this method helps to understand the functioning of the cell cycle, optimize doses and administration schemes of drugs by analyzing individual image data.

Citations: [1] Knudsen et al. Cell reports. 2022
[2] Gaglia et al. Nat Cell Biol. 2022
[3] Gérard & Goldbeter. Int Focus. 2011