Toward a Whole-Cell Model of Human Embryonic Stem Cells: A Composable Model of Metabolism Built on a Scalable Framework for Integrating Multiple Pathway Submodels

Stem cell behaviors such as self-renewal and differentiation result from complex interactions among signaling, gene regulation, metabolism, and other pathways. To understand how these pathways collectively determine cell behavior, we aim to develop the first whole-cell computational model of a human cell, focusing on the well-characterized H1 human embryonic stem cell (hESC) line.

Our approach is to build and integrate submodels of each pathway. To begin, we have developed a scalable framework for integrating multiple pathway submodels and built a genome-scale submodel of single-cell metabolism.

The integration framework consists of state variables that represent the instantaneous configuration of a cell such as its mass, volume, and composition including the abundance and compartmental location of each molecular component. We developed the initial conditions for the framework by integrating extensive data including the DNA sequence of H1-hESCs; the location of each transcript, exon, and protein lifted from the hg19 reference; protein-complex compositions; macromolecular contents; small-metabolite concentrations; RNA expression; protein abundances; and the stoichiometry, catalysis, rate law, and kinetic data of each reaction.

We built the metabolism submodel by: (1) expanding Recon 2.2 to produce and recycle the metabolites that are needed and generated by other pathways, respectively; (2) refining the submodel for integration into a stochastic dynamical model of individual cells; and (3) using multi-omics and biochemical data to contextualize the submodel to represent H1-hESCs. The metabolism submodel correctly predicts the essentiality of 97% of metabolism-associated genes in hESCs. 87.5% of the false negative have non-metabolic functions, highlighting the need for integrated models that describe multiple pathways.

Going forward, we will create submodels of other pathways, integrate the submodels, and use the integrated model to investigate stem cell self-renewal. Ultimately, we believe whole-cell models have the potential to transform basic science and medicine through unprecedented in silico experiments personalized to each patient.