Modeling synthetic serum marker kinetics for monitoring deep-tissue gene expression.

Serum markers could theoretically enable monitoring of gene expression dynamics with a simple blood draw. However, such markers are typically used to measure long-term changes, such as the progression of disease over multiple days or months. In this theoretical study, we determine the maximum theoretical temporal resolution and precision - the ability to distinguish rapid changes in gene expression and to obtain the true frequency of such changes, respectively. As a model of such processes, we used Released Markers of Activity (RMAs) - a class of synthetic serum markers that are expressed in neurons in the brain but transported into the blood. RMAs are orthogonal to physiological processes and have tunable levels, production rate, and onset time, providing well-defined data for serum marker production, tissue transport, and detection. We explore several scenarios where RMAs were used, including monitoring tissue transduction, changes in endogenous gene expression, and drug-induced marker expression. We demonstrate that the temporal resolution of monitoring primarily depends on the extrinsic noise level of the RMA signal and protein serum half-life. Additionally, we find that decreasing the serum half-life results in improved temporal precision at the cost of signal intensity. To enable broad use of this model, we developed a library and interface for running simulations to approximate marker trajectories to inform experimental design decisions and optimize marker detection. Author SummaryGene expression is a foundational driver of biological processes. We use synthetic serum markers that can report on gene expression with a blood draw. These synthetic markers have well-defined parameters and thus form a useful model for understanding the maximum theoretical temporal resolution and precision of monitoring gene expression. These two variables describe, respectively, the fastest changes in gene expression and how accurately the kinetics can be reflected through a serum marker measurement. Developing more robust reporter systems requires understanding how biological properties of these markers, such as serum half-life or measurement noise, affect their kinetics. Here, we developed computational models describing in vivo behavior of synthetic serum markers to identify the most critical parameters for optimization. We show that serum half-life and measurement noise are the main contributors to the temporal precision and resolution of monitoring. Additionally, we demonstrate the ability to predict marker levels in various contexts, including constitutive and drug-induced expression.