Modeling gene expression networks to predict interchromosomal organization during human embryonic stem cell differentiation

Cellular differentiation occurs through the regulation of lineage-specific gene expression networks that are facilitated by the spatial organization of the genome. Although techniques based on the chromatin conformation capture (3C) approach have yielded intrachromosomal genome-wide interaction maps, strategies to identify non-random interchromosomal associations is lacking. Therefore, we modeled the genomic organization of chromosomes based on the regulatory networks involved in the differentiation of pluripotent human embryonic stem cells (hESCs) to committed neuronal precursor cells (cNPCs). Importantly, transcriptional regulation has been demonstrated to be a driving force in non-random genome organization. Thus, we constructed coarse-grained in silico networks using gene expression data to identify potential physical associations among chromosomes occurring in situ and then analyzed the three-dimensional (3D) distribution of these chromosomes, assessing how their associations contribute to nuclear organization. Our analysis suggests that coordinate regulation of differentially expressed genes is correlated with the 3D organization of chromosomes in hESC nuclei induced to differentiate to cNPCs.Author SummaryThe cellular commitment and differentiation of stem cells is a hallmark of metazoan development. The ultimate fate of a stem cell is defined by the synergistic modulation of key gene regulatory networks within the nucleus. In our work, we formulate an in silico model describing how the similarity in the expression profile of differentially regulated gene networks is correlated with the higher-order organization of chromosomes during differentiation from human embryonic stem cells (hESCs) to committed neuronal precursor cells (cNPCs). Using graph statistics, we observe that the genome networks generated using the in silico model exhibit properties similar to real-world networks. In addition to modeling how gene expression relates to dynamic changes in chromosome organization, we test the model by calculating the relative proximity of multiple chromosome pairs using 3D fluorescence in situ hybridization (FISH). While various chromosomal properties, including gene density and overall length, have been attributed to chromosome organization, our previous work has identified the emergence of cell-type specific chromosomal topologies related to coordinate gene regulation during cellular differentiation. Here we extend these findings by determining whether our in silico model can predict chromosome association based upon coordinate gene expression. Our work supports the idea that gene co-regulation, in addition to inherent organizational constraints of the nucleus, influences three-dimensional chromosome organization.