Collaborative Research: RECODE: Directing and Controlling Cardiac Differentiation Through Cellular and Microenvironmental Manipulation and Application of Machine-Learning

US mortality rates from heart disease are increasing, driven particularly by the increasing prevalence of patients with heart failure. Limited availability of native human cardiac tissue impedes research, drug discovery, and clinical cardiac regeneration efforts. Treatment with stem cell-derived cardiac tissues has exceptionally high potential to achieve clinically meaningful outcomes.

However, the generation of a heterogeneous mixture of cell types is a critical barrier to cell-based cardiac therapy. By employing developmental biology, tissue engineering, and machine learning, this Reproducible Cells and Organoids via Directed-Differentiation Encoding (RECODE)research builds the foundation for overcoming this obstacle and develops methodologies and design approaches to produce functional cell types needed in understanding and treating heart disease.

This project will support undergraduate students from Alabama State University – a historically black university – to participate in summer research experiences at Auburn University.

Despite significant advances in our understanding of human induced pluripotent stem cell and cardiac development biology, our ability to generate specific cardiac cell subtypes from pluripotent stem cells in sufficient quantities remains limited. Cardiac differentiation of human induced pluripotent stem cells has been broken down into a stepwise process from pluripotency to mesoderm to cardiac progenitors to first and second heart fields.

However, this progression occurs at differing rates, require differing concentrations, durations, and timing of exposure to key cell signaling molecules, and yield varying concentrations of cardiomyocytes. Understanding the population dynamics and probabilities that a given cell will move towards becoming one cell type versus another is necessary for making predictions and directing decisions to achieve a desired final cell type or a mixture of cell types.

The goal of this RECODE project is to establish a paired experimental process and guiding hybrid model utilizing real-time measurements from differentiating cardiomyocytes to predict both the outcome of ongoing cardiac differentiation and the process parameters that should be adjusted to achieve the desired result. The project work will (1) marry innovative machine learning tools and cardiac developmental stages to mine single cell RNA sequencing data to identify key developmental decisions and levers that control cell fate at these instances, (2) perform directed cardiac differentiation in 3D to address complex autocrine, paracrine, cell-cell and cell-matrix interactions that are absent in conventional 2D assays, (3) employ cardiac cell subtype-specific fluorescent reporter to quantify differentiation outcome in real-time, and (4) develop a process control analytical platform that integrates differentiation outcome data with experimentally-defined input parameters that can enable generation of specific composition of cardiac cell subtypes on-demand using robustly validated and reproducible differentiation design rules.

This RECODE award is co-funded by the Mechanics and Engineering Materials Cluster in the Division of Civil, Mechanical, and Manufacturing Innovation and the Engineering Biology and Health Cluster in the Division of Chemical, Bioengineering, Environmental, and Transport Systems.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.