Crispr Engineering in CD34+ Progenitors Reveals Cis-Acting Regulatory Regions Mediating 3D Interactions and Stem Cell Fate Decisions

DNA methylation is a critical regulator of cis-regulatory elements that can impact the distribution of epigenetic regulators and transcription factors. We have mapped DNA methylation changes in human CD34+ cells and downstream progeny to detect changes in DNA methylation with differentiation. Some large regions of low DNA methylation emerge with differentiation, which we call dynamic (d-) canyons. In order to determine whether these correlate with changes in the 3D genome, we generated 6 high-resolution Hi-C contact maps from CD34+CD38- cells, Adult bone marrow CD34+ cells, 7 days cultured CD34+ cells, CD71+CD36+CD235+ Erythroid cells, CD4+ T cells and AML patient blast cells. We generated ~ 1 billion mapped reads in order to create each 3D map of the genome at kilobase resolution. We identified several sites of d-canyons which correlate with changes in Hi-C loops. To study the functional role of d-canyons, we performed CRISPR/CAS9 mediated genome engineering targeting three sites of interest in CD34+ progenitors. We designed sgRNAs to delete several d-canyon regions that exhibited cell-type specific methylation and dynamic H3K27ac marking.

First, we targeted a novel putative regulatory region of the RUNX1 gene, a critical master regulator of hematopoiesis that is frequently mutated in human leukemia. From genome-wide DNA methylation profiling, we identified a d-canyon located in the first intron of RUNX1, which overlaped with a H3K27ac peak in human CD34+CD38- hematopoietic progenitor cells. DNA methylation in this region is further depleted in T cells and increased in AML cells, suggesting a role for regulating RUNX1 in specific cell types. To study its function, we designed 2 sgRNAs to delete a 1.8 kb d-canyon. CRISPR/Cas9-mediated deletion of this region resulted in ablation of RUNX1 expression in cord blood hematopoietic stem cells and a significant decrease of engraftment activity in NSG mice along with an increase of erythroid colony forming ability in in vitro assays.

In addition, we identified d-canyons upstream of the master regulators GATA2 and in the HOXA cluster. We deleted 1.7kb d-canyon upstream of GATA2 gene, finding that it resulted in increased self-renewal of CD34+CD38- cells in NSG xenografts. In contrast, when we deleted one a d-canyon located ~2Mb upstream of the HOXA cluster, within a HiC loop signal present only in CD34+CD38- cells, we observed in decreased CD34+CD38- cell self-renewal but a significant increase of CD34+CD38+ differentiated progenitors in an in vitro culture system, as well as in NSG mice. Colony forming assay showed decreased colony size and numbers.

Finally, we identified a d-canyon associated with TCF3, also known as E2A, a transcription factor involved in B and T cell lineage differentiation. We designed 4 sgRNAs to delete 1.5kb d-canyon edge regions within the second intron of TCF3. Removal of this region resulted in a significant decrease of CD19+ B cells, but an increase of CD3+ T cells in NSG mice.

Taken together, these results suggest the functional importance of d-canyons for orchestrating genome architecture and fate decisions of hematopoietic stem cells. These findings advance our understanding of the relationship between DNA methylation changes and loop interactions, providing new insights into the potential impact of potential aberrant DNA methylation and chromatin structure.