Network Analysis Of Single Cells Identifies DDIT3 As a Nucleator Of Erythroid Signatures In Early Erythro-Myelopoiesis

Analysis of individual cells within phenotypically-uniform stem and progenitor populations reveals heterogeneity of cell states and affords exploration of their differential contribution to fate transitions. Herein, we investigate the transcriptional composition of lineage commitment transitions in the mouse hematopoietic system, obtained either spontaneously or driven by individual transcription factors (TF), to define alternative molecular states at the emergence of erythroid and myeloid lineage identity.

To this end, we transcriptionally profile hundreds of individual cells across a spectrum of lineage states by quantitative RT-PCR on a Fluidigm platform. We explore the dual binary (on-off) vs. continuous (expression level) nature of PCR data and respectively calculate odds ratio, and Spearman correlation on co-expressing cells, for any pair of genes, in order to infer the circuitry underlying alternative routes to lineage commitment. This approach identifies Ddit3 as a specific target of GATA-driven networks that antagonizes Cebpa transcriptionally at the emergence of myeloid fate.

Knockdown and knockout approaches indicate a requirement for Ddit3 in early erythroid decisions from stem cells. Enforced expression of Ddit3 in granulocytic-monocytic progenitors (GMP) results in down-regulation of individual Cebpfamily members and their downstream transcriptional programmes, including myelo-monocytic cytokine receptors. Notably, the resulting cells have robust self-renewal potential and express an early erythroid signature that is augmented in the presence of interleukin-3 and erythropoietin.

Taken together, the data suggest that Ddit3 can nucleate incipient erythroid-affiliated programmes in early erythro-myelopoiesis, and may contribute to fate choice through antagonism with Cebp networks. Interestingly, ex-vivo culture and expression profiling of individual wild-type and DDIT3-enforced GMP suggest that the erythroid-primed state captured through Ddit3 expression may be found in a small proportion of native GMP, suggesting some fluidity of lineage identity in early progenitors.