The differentiation of myeloid progenitors is effected by cascading waves of coordinated gene expression that remodel...

The differentiation of hematopoietic progenitors into specialized types requires the transmittal of information from a few external or internal regulators to the thousands of genes that produce a cell type's characteristic phenotypes. While the main signaling pathways, transcription factors, and the genes eliciting the terminal phenotypes are known, how information flows from a few regulators to thousands of genes to change the state of the cell remains to be fleshed out. To profile this information transfer process, we sampled the differentiation of the PUER myeloid cell line into macrophages and neutrophils at 27 time points over seven days. There is extensive transient regulation; the number of transcripts modulated in time is twice the number differentially expressed between endpoints. Differentiation is marked by two sharp transitions, at ∼8h and ∼80h, when transcriptomic state changes suddenly. We utilized non-negative matrix factorization to identify behaviors, characteristic temporal patterns of gene expression, and to classify transcripts by behavior. Only 10 distinct behaviors are sufficient to recapitulate the expression of ∼36,000 transcripts with high fidelity. Gene expression in most of the behaviors occurs in pulses of varying initiation times and durations. This implies that information transfer during differentiation occurs in cascading waves of gene expression culminating in the permanent turning on of certain genes after ∼80h. Each behavior is enriched in specific biological processes, so that physiological remodeling proceeds in a characteristic order-signal transduction, translation and mRNA processing, metabolism, and, ultimately, myeloid phenotypic processes. The sharp transition at 8h corresponds to the completion of transcriptional and translational remodeling and the initiation of metabolic remodeling; the one at 80h corresponds to the elicitation of myeloid phenotypes. Our analysis shows that differentiation relies upon a series of transient, rapid, and complex gene regulatory events and highlights the importance of profiling it at a high temporal resolution.