Uncovering spatio-temporal, mechano-transcriptomic signatures of development

Contact PI: Dumitrascu, Bianca PROJECT SUMMARY/ABSTRACT Embryonic development requires the complex coordination of molecular, cellular, and mechanical processes across spatio-temporal scales. Specialized gene expression programs are established, and cellular aggregates generate force, thus exhibiting synchronized movement resulting in major self-organizing events such as heart looping or neural tube folding. Dysregulation at any level can lead to developmental defects and lethality. Yet, it is unclear how gene expression, cell morphometrics, and tissue-level mechanics work together to accomplish cell type diversity and inform shape. The goal of my research program is to understand the interaction between genomics, morphology, and mechanics by disentangling their contribution to coordinated cellular movement in developmental contexts. Despite the crucial role of cell morphometrics and tissue-level mechanics in development, methods for rigorously quantifying and exploring these data types jointly are currently lacking. During this award, we will fill this gap by: i) identifying gene expression programs that are predictive of tensional dynamics in multi-species, spatio-temporal contexts and ii) inverse engineering the spatio-temporal distribution of tensional profiles inferred from microscopy data to evaluate how feedback from gene programs shapes tissues. To do so, we innovate through the building of an integrative, statistical machine learning framework for 4D mammalian development, leveraging tools from computer vision, differential geometry, factor analysis, and graph neural networks, to assess the interplay between genomics, morphology, and mechanics. We take a multi-species approach, using data from mouse, fly, frogs, and zebrafish, to identify principles of self-organization shared across biological systems, with eventual relevance to processes leading to understanding human disease. More broadly, these techniques will allow us to generate a first computational framework for understanding how the physical properties of individual cells and their environments determine self-organization, development, and regeneration in multi-cellular contexts. We expect that this framework will lead to novel experimental design approaches for controlling pattern formation and generating tissues that fold according to predefined shapes.