(4r) Control of Spatiotemporal Dynamics of Living Cells through Biomolecular Phase Separation

My research interests are in the general areas of process modeling and control as well as systems and synthetic biology. Specifically, the main theme of my research is to design and optimize microbial communities for various engineering applications such as biomanufacturing, bioremediation, and precision medicine. Like any other biological processes, microbial community dynamics are inherently complex. Thus, engineering microbial communities require development of sophisticated computational and experimental methodology. To this end, my research group will focus on 1) designing and optimizing novel genetic circuit components for control cellular behaviors; 2) developing a data-driven mathematical model to predict and optimize cellular dynamics. By synergistically employing experimental and computational tools, my research group will engineer microbial communities that will achieve their engineering goals robustly.
1. Control of dynamics of microbial cells through biomolecular phase separation
One of the key objectives of synthetic biology is to control cellular fates and processes through tuning gene expression in a living cell. This is achieved by introducing engineered genetic circuits into living cells. Hence, a large amount of research efforts has been devoted to optimizing the components of gene circuits such as transcriptional factors (TFs) to increase the capability of these circuits. In this regard, biomolecular phase separation (PS) has emerged as a novel mechanism that can be potentially engineered as a new component of genetic circuits. PS is a physical phenomenon where a homogeneous mixture spontaneously “demix” into two or more coexisting phases, and now it is well known that a wide range of biomolecules such as proteins and RNAs can undergo PS under certain conditions. Once PS occurs, membraneless compartments form, which selectively enriches a set of biomolecules. Because of elevated concentrations of biomolecules inside these compartments, it has been suggested that these PS compartments can act as reaction hubs to accelerate corresponding biochemical reactions. However, preliminary studies in the literature showed mixed success of PS in controlling the gene expression.
Goal: Quantitative characterization of PS-mediated gene expression in a living cell. Although the intuition of PS-mediated gene expression is simple, no study has been conducted to systematically characterize the capability of PS in controlling cellular biochemical reactions. For this, I constructed gene circuits to express a TF fused with a fluorescent protein and intrinsically disordered proteins (IDPs), whose expression induces another fluorescent protein. By fusing TFs with IDPs, we can make TFs to undergo PS. By tracking concentrations of TFs and downstream gene expression, we can quantitatively characterize effects of PS in gene expression. Our preliminary study has shown that, for some TFs, the PS-mediated gene expression is inhibitory. That is, the occurrence of PS, surprisingly, suppresses gene expression, which contradicts with the common beliefs in the research field. This raises the possibility that PS does not always guarantee. Based on this current finding, the ongoing work expands the repertoire of TFs and IDPs to characterize the exact correlation between PS and gene expression. By generalizing this capability of PS, we will have a new engineering method to control spatiotemporal dynamics of cells.

2. Data-driven mechanistic hybrid model to model and predict microbial dynamics

Previous work and motivation. Due to the sheer complexity of a biological process, it is difficult to predict the dynamics of a cellular process. In this regard, I developed a hybrid modeling approach, where a physics-based mechanistic model is complemented by a data-driven model such as a neural network model, to improve the model prediction accuracy (Lee et al. IET Syst. Biol. 2019; Lee et al., PLoS Comp. Biol. 2020). The hybrid modeling approach was implemented to model the NF𝜅B signaling pathway under external stimulus, and the prediction accuracy was improved significantly while preserving the underlying mechanisms. Besides predicting a cell’s natural behavior, the hybrid modeling approach will be a very valuable method for designing synthetic gene circuits. One of the persisting problems in synthetic biology is unpredictability of cellular behaviors with designed synthetic circuits, mainly due to the host-circuit interactions that are difficult to predict a priori. A single bacterium contains thousands of genes. A circuit may interact with the host in multiple ways beyond its intended design. For instance, synthetic gene circuits often impose burden or toxicity on the host cell, which can modulate the circuit function via growth feedback, leading to circuit failure. Moreover, our knowledge of cellular processes is still not perfect: for the model bacterium Escherichia coli, the functions of ~35% of its genes have not been experimentally validated. Under these circumstances, a hybrid modeling approach can be implemented.
Goal: Development of a hybrid model to increase model accuracy by complementing mechanistic model with experimentation. For ensuring the robust performance of a designed gene circuit, a mathematical model is used to predict and optimize the behaviors of the circuit a priori based on the underlying logic of the circuit. However, its interactions with the host are difficult to know beforehand. In this end, hybrid modeling approach will be implemented to improve the prediction accuracy to increase the performance robustness of a genetic circuit. For this, a series of experiments will be performed with a host with and without a preliminary genetic circuit and monitor their growth and performance of genetic circuits through measuring gene expressions under a variety of environments. Based on this set of data and a pre-designed mechanistic model, a hybrid model will be constructed with a significant increase in its prediction accuracy. By generalizing this hybrid model with different host strains and different genetic circuits, predictive engineering of a gene circuit, where a set of genetic circuit components and a host cell will be chosen based on the preset goal, will be plausible for a wider use of synthetic biology beyond a laboratory to industry.

Teaching Statement

My goal as an educator is to help students cultivate their skills to think critically, systematically, and be independent so that they will thrive in future their careers. Given my undergraduate and graduate school education in chemical engineering as well as my postdoctoral training in synthetic biology, I am confident that I can teach a wide variety of undergraduate and graduate courses in the chemical engineering curriculum.

Based on my training in process systems engineering and systems/synthetic biology, my immediate teaching plans for undergraduate courses are to teach ones related to process modeling and simulation, numerical analysis, biochemical engineering, and chemical process control. If I have a chance to teach a process control course, I intend to introduce laboratory components into the curriculum to strengthen students’ understanding in the subject through hand-on experience. This will complement the more conventional instruction method of the process control course, where students gain experience mostly through simulations. For the laboratory component, I plan to maximize the utility of microcontrollers that are cheaply available through collaborations with Prof. Hedengren in University Bringham Young University or Prof. Svoronos in University of Florida. Besides these courses, I have been teaching assistants for other core chemical engineering courses such as mass and heat transfer, which I will be able to teach with sufficient preparation.

Regarding graduate curriculum, I plan to develop an introduction to systems and synthetic biology course, if permitted. In this course, I will present the overview of the research field such as design of genetic circuits, mathematical modeling of genetic circuits. Throughout the course, students will learn the application of mathematical modeling (primarily deterministic or stochastic kinetic modeling) to the analysis of the “design principles” of cellular networks, and strategies to engineer such networks in the real cell. At the end of the course, students will be asked to model biological systems of their interests and provide in-depth analysis of system dynamics.