Image-based spatiotemporal causality inference for protein signaling networks.

Area of Science:

• Cellular and Molecular Biology
• Systems Biology
• Bioimage Analysis

Background:

• Existing models of cellular signaling networks often neglect spatial organization or use low-resolution compartmental approaches.
• Fluorescence microscopy enables monitoring molecular spatiotemporal distribution, but large-scale image analysis for learning complex protein regulatory networks is lacking.

Purpose of the Study:

• To develop and evaluate methods for analyzing spatial relationships in cellular signaling networks.
• To identify how changes in protein concentration in one cellular region influence concentrations in other regions.

Main Methods:

• Development of methods integrating spatiotemporal alignment, automated region identification, and causal inference.
• Application to 3D confocal microscopy movies of GFP-tagged T cells undergoing costimulation.

Main Results:

• Learned models identified putative causal relationships among 12 proteins in T cell signaling, including novel predictions.
• Models accurately predicted protein localization over time in partially stimulated T cells, demonstrating statistical significance.
• The developed methods are anticipated for broad applicability across various biological systems.

Conclusions:

• The presented image analysis methods enable learning complex protein regulatory networks by considering spatial organization.
• This approach advances the understanding of cell signaling dynamics and provides a powerful tool for systems biology research.