Paused RNA polymerase II enables flexible lineage decisions but also underlies oncogenic vulnerabilities in human hematopoiesis Free

During inflammation, hematopoietic stem cells (HSCs) must integrate environmental stimuli to generate the appropriate myeloid and lymphoid cell types to mount an effective immune response. Repeated stimulation induces “inflammatory memory” in HSCs through stable activation of immune response genes (Zeng et al., bioRxiv, 2023). However, the molecular mechanisms that integrate immune stimulation to “choose” the correct differentiation programs remain unclear.

Mutations enriched in blood cancers suggest that the gene-expression programs that support the flexible inflammatory response of HSCs may also create lineage-specific vulnerabilities to oncogenic transformation. For example, MLL1/KMT2A translocations produce fusion proteins with transcriptional elongation factors that manipulate these flexible programs by disrupting Pol II pausing to drive constitutive activation of developmental genes (Hu and Shilatifard, Genes & Development, 2016). Myelodysplastic syndromes (MDS) frequently harbor mutations in Polycomb group proteins, including ASXL1, EZH2, and EED (Bose et al., Blood, 2018). These Polycomb proteins act as transcriptional repressors throughout development, but in HSCs they may also serve a second specialized role that restrains inflammatory transcription through maintenance of Pol II pausing.

To investigate the epigenetic mechanisms underlying human hematopoiesis, we developed single-cell “multifactorial” profiling methods that map two chromatin-bound proteins in the same cell. As a model of inflammatory stress, we profiled over 200,000 CD34+ HSCs and progenitors mobilized with G-CSF. In stimulated HSCs, immune response genes exhibit elevated Pol II occupancy and loss of Polycomb-mediated repression marked by H3K27me3. These results suggest Polycomb restrains the inflammatory response by contributing to Pol II pausing. By profiling MDS patient samples with ASXL1 mutations, we find H3K27me3 is reduced over the inflammatory memory genes in HSCs. Our results support a model in which Polycomb dampens inflammatory memory by limiting Pol II elongation at key cytokine response genes.

Next, we investigated lineage commitment by profiling Pol II together with the histone H3 lysine 4 methylome (H3K4me1-2-3). We identified H3K4 methylation patterns that delineate lineage-restricted hematopoietic progenitor cell types and uncovered two modes of Pol II transcriptional activation: (1) “pause-and-release” genes exhibit promoter-proximal pausing of Pol II in HSCs, and this holds Pol II in a non-productive but poised state for rapid activation during lineage commitment; (2) “initiate-and-release” genes are activated later in lineage restricted progenitors through de novo accumulation of Pol II. Although Pol II likely pauses transiently on initiate-and-release genes, this pausing occurs while the gene is expressed and does not serve as a developmental check point. FNDC3B is an example of a pause-and-release gene that is poised in HSCs for rapid activation in myeloid progenitors. Strikingly, FNDC3B is also directly up regulated by the MLL::AF4 oncoprotein in acute myeloid leukemias following lineage switching. This suggests MLL-fusion proteins promote therapeutic resistance by driving aberrant release of paused Pol II at developmental genes.

We find that Pol II pausing plays a critical role in the flexible inflammatory response of HSCs, but this same mechanism also underlies oncogenic vulnerabilities. By capturing transcriptional mechanisms in primary human cells, our single-cell multi-protein profiling platform reveals how developmental control is repurposed in disease and provides a framework for predicting resistance and guiding mechanism-based therapies.