A small number of Rho family GTPases participate in a broad array of fundamental cellular behaviors. Specificity is possible due to spatial and temporal control of GTPase “activation”; Guanine exchange factors (GEFs) generate activated, GTP-bound GTPases with precise timing and localization, while specialized interactions with adhesion molecules, membrane domains and other localized structures specify GEF-GTPase interactions. GEF/GTPase circuits are complex, with localized feedbacks, multiple GEFs controlling one GTPase, and vice versa. To dissect this spatiotemporally regulated circuitry requires imaging, and new analytical techniques that can dissect causal relationships from imaging data. Following the intentions of PAR- 19-158 (Bioengineering Research Grants), we propose a multidisciplinary collaboration leveraging organic chemistry, protein engineering, imaging, and computer science to fudnamentally advance signal transduction imaging and analysis. As a biological testbed we will explore the role of GEF-GTPase interactions in cell protrusion, single cell migration and collective migration. We will develop a generalizable approach to GEF biosensors, and adapt our proven GTPase biosensors to image GEF and GTPase activities in the same cell. Because GEF-GTPase interactions are heterogeneous and complex, multiplexed imaging is necessary to quantify their relative dynamics. However, perturbation of cell behavior is especially problematic when using two biosensors in the same cell. We will therefore develop new biosensor designs that greatly reduce cell perturbation. Even the most precise imaging of overlapping molecular activations has not revealed causal relationships. We will therefore adopt the framework of Granger Causality inference, which was originally devised for financial market analysis, to extract causal connections and feedback interactions from imaging data. Numerous steps will be necessary to translate the existing concepts of Granger causality to the analysis of spatially and temporally distributed molecular processes. Most importantly, we will implement a schema for Granger causality inference in multivariate time series models that will capture spatial relations, and we will combine principles of high-dimensional statistical regression with approaches from control theory to estimate information flows between variables that are coupled by strong feedbacks. We will also develop a novel clustering approach that preserves the neighborhood topology of data in a high-dimensional feature space and in the Euclidian space of the cell outline to identify signaling microdomains. Finally, to test and confirm our hypotheses, we will use new photo-activatable and photo-inhibitable analogs of GEFs together with GTPase biosensors to control one protein while observing another. This research plan will produce biosensors with reduced perturbation, biosensor/optogenetic multiplexing capabilities, and image analysis/modeling approaches necessary to shed light on the network topology of nonlinear, spatiotemporally controlled signaling pathways. All tools will efficiently deployed to the community.

In this project we are developing new ways to understand the mechanisms by which cells interpret, and in cancer misinterpret, the host of stimuli from outside and inside the cell that determine behavior. We are building molecules that can report, through light emission, the behavior of multiple key elements of the cellular “decision making circuits”, and in parallel developing new mathematical tools to interpret the output of these molecules, and to integrate this data into overall models that reveal information flow in the circuitry. We are also developing the ability to control one circuit component while watching the activity of another, enabling us to dissect out causal relationships within cellular circuits.