Transient Inflammatory Signals Drive Persistent Changes in the PU.1 Transcriptional State of HSC

Robust cell-to-cell signaling is required for effective emergency myelopoiesis. How signals impact the internal state of recipient hematopoietic stem and progenitor cells (HSPC) is an area of active research with obvious clinical implications. Complicating these studies has been a growing appreciation for HSPC heterogeneity in the expression of key transcription factors (TF). We previously demonstrated marked cell-to-cell heterogeneity in the expression of the master myeloid TF, PU.1. As PU.1 is both a regulator of inflammatory cytokine receptors as well as a downstream target of inflammatory pathways, we proposed that variability in the basal expression of PU.1 could lead to variable receptivity to inflammatory signals. To test this hypothesis, we stimulated murine HSC with brief pulses of IL1β, TNFα, G-CSF, M-CSF, and GM-CSF, and used single molecule resolution, mRNA FISH (smFISH) to measure the transcriptional response of PU.1. We found that IL1β and TNFa lead to immediate (<1 hour) activation of de novo PU.1 transcription in HSC, whereas G-CSF, M-CSF, and GM-CSF had minimal effect. We then tested the persistence of this inductive effect on PU.1 transcription. 48 hours after cytokine wash out, IL1β and TNFα stimulated HSC maintained a high PU.1 transcriptional output, with ~2- to 5-fold increases in mRNA counts/cell as well as a 5-fold increase in active PU.1 transcription sites (TS). Interestingly, M-CSF also led to an increase in mRNA counts/cell with a more modest effect on TS frequency. To determine whether these changes were occurring in all HSC versus subpopulations, we expanded colonies from HSC ex vivo after a 1-hour pulse of IL1β, and measured the inter- and intra-colony transcriptional state heterogeneity as defined by PU.1 and its chief antagonist in HSC, Gata2. IL1β exposure caused PU.1 state skewing in the majority of HSC; however, a high PU.1 state was not seen uniformly amongst progeny, suggesting that the transcriptional memory of prior stimulation either decays or is actively reversed. Next, we asked how an increased time in the PU.1 state after IL1β changes the transcriptional activity of PU.1 target, Csf1r. Surprisingly, the increase in PU.1 mRNA did not lead to a corresponding increase in PU.1 protein levels at the single cell level. Additionally, Csf1r levels did not correlate strongly with nuclear PU.1 protein concentration. On the contrary, there was a strong correlation between the time ancestral cells spent in a high PU.1 state and the probability of expressing Csf1r. This finding supports that a PU.1- and time-dependent mechanism must operate in the licensing of the Csf1r locus. Furthermore, the licensing of Csf1r had functional consequences for cells despite overall low mRNA copy numbers as culturing colonies with M-CSF containing media after IL1β pulse led to expansion of a population of macrophage cells that were effectively not seen in non-stimulated colonies. In summary, we demonstrate that IL1β and TNFα lead to an almost immediate induction of PU.1 transcription and that this inductive effect persists for days after stimulation removal in nearly all HSC. We further show that this leads to the time dependent upregulation of the PU.1 target, Csf1r, which thereby renders cells responsive to M-CSF driven macrophage differentiation. We believe these studies have important implications for how we understand the inflammatory response and consequent emergency myelopoiesis. We also posit that these studies are relevant to our understanding of gene regulation, particularly at the single cell level. Finally, we hypothesize that time-dependent transcriptional regulation may be an essential, evolutionarily selected mechanism supporting stem cell robustness in the presence of recurrent physiologic stress.