Model-guided design of mammalian genetic programs

Abstract

Genetically engineering cells to perform customizable functions is an emerging frontier with numerous technological and translational applications. However, it remains challenging to systematically engineer mammalian cells to execute complex functions. To address this need, we developed a method enabling accurate genetic program design using high-performing genetic parts and predictive computational models. We built multi-functional proteins integrating both transcriptional and post-translational control, validated models for describing these mechanisms, implemented digital and analog processing, and effectively linked genetic circuits with sensors for multi-input evaluations. The functional modularity and compositional versatility of these parts enable one to satisfy a given design objective via multiple synonymous programs. Our approach empowers bioengineers to predictively design mammalian cellular functions that perform as expected even at high levels of biological complexity.

Robust and tunable signal processing in mammalian cells via engineered covalent modification cycles

“…Combined with recent advances in utilizing TCS proteins to engineer synthetic receptors in mammalian cells 12,13 and to rewire the specificity of response regulators (RRs) in bacteria 64 , our platform will enable construction of sophisticated synthetic signaling systems that can connect intracellular and extracellular inputs to diverse target output in mammalian cells. While much work has so far focused on synthetic receptor engineering 10 , incorporation of downstream signal processing moieties to improve signaling pathway function has only recently begun to be explored 65 . In particular, the ability to easily tune signaling pathway activity through phosphatase expression and the ability to robustly control downstream gene expression processes will facilitate the creation of synthetic signaling systems that can operate across diverse cellular contexts.…”

GAMES: A dynamic model development workflow for rigorous characterization of synthetic genetic systems

“…One should also visually inspect the best model fits to the PEM evaluation data to check whether parameter sets yielding high R 2 values produce simulations that match the training data. Although it is possible to alternatively use 𝜒 5 to define the consistency check, in practice one would need to adjust the value of 𝜒 5 9:;; on a case-by-case basis.…”

“…At a fundamental level, this type of model is a mathematical description of the components and interactions in a system. These models have proven useful for investigating the behaviors of genetic programs-sets of biological parts such as genes and proteins that are designed to implement specific functions [1][2][3][4][5] . Efforts to date have often used ordinary differential equations (ODEs), which describe the time evolution of the concentrations of system components and provide a framework for representing continuous dynamic systems.…”

“…Efforts to date have often used ordinary differential equations (ODEs), which describe the time evolution of the concentrations of system components and provide a framework for representing continuous dynamic systems. These models have been used to explain and predict behaviors of genetic programs that perform a growing array of functions including regulating gene expression [3][4][5][6][7] , implementing logic gates 5,6 , and implementing feedback control [8][9][10][11][12][13] . Models can generally be employed for explanation, in which the objective is to help the user understand a set of experimental observations, or for prediction, in which the objective is to simulate the response of a genetic program to a previously untested experimental condition or design choice (terms in italics are summarized in a glossary in Supplementary Information).…”

“…Efforts to date have often used ordinary differential equations (ODEs), which describe the time evolution of the concentrations of system components and provide a framework for representing continuous dynamic systems. These models have been used to explain and predict behaviors of genetic programs that perform a growing array of functions including regulating gene expression 3–7 , implementing logic gates 5, 6 , and implementing feedback control 8–13 .…”