ABSTRACT: A major problem with linking transcription factor binding to function is that many factors bind to a large number of, at least in any given cell type, seemingly irrelevant regions. This makes it hard to filter out which binding sites are responsible for the regulation of a given gene. PU.1 (Spi1, Sfpi1) is an excellent example of a transcription factor that works both to mediate developmental choices and to serve the alternative developmental fates that emerge from these choices. Its role in T-cell development is confined to the early stages where high PU.1 expression persists through multiple cell divisions and is sharply downregulated over the DN2a-to-DN2b T-cell commitment stage. Even though it is known that PU.1 is necessary for the survival of the earliest T-cell progenitors, where it is needed for optimal proliferation, control of alternative lineage genes, and correct timing of the access to T-lineage genes, it has not been well-studied how PU.1 finds it targets and exerts its functions within this context. Here, we show in detail how PU.1 selects its binding sites and subsequently regulate gene expression. We show that PU.1 has two effective affinity thresholds for occupancy depending on current chromatin openness as measured by ATAC-sequencing, but that it is easily capable of binding sites in closed chromatin. Unexpectedly, although its binding to promoters is least constrained, promoters are the only major class of sites where it exerts a predominantly negative effect; otherwise it works locally as an activator, mainly mediated through binding to sites in closed or dynamically closing distal enhancer elements, where it can rapidly open chromatin and induce histone acetylation. However, its ability to open the chromatin depends not only on its own affinity but also on the presence of collaborating factors, because we show that PU.1 introduction into PU.1-negative cells triggers a massive reorganization of occupancy patterns of at least three other factors: Runx1, Satb1, and Gata3. Most strikingly, PU.1’s “theft” of Runx1 and Satb1 from many sites where they were binding in the absence of PU.1 enables PU.1 to exert a novel form of repression even on genes where it has no binding sites itself. We show here that PU.1 requires domains outside of its DNA binding domain to properly open chromatin, and this structural requirement is directly connected with its ability to bind to Runx1 and to Satb1, and moreover, Runx1, in particular, is an important collaborator to activate many of its target genes in the context of early T-cell development. Thus, PU.1 regulates gene expression via two distinct mechanisms. First, PU.1 steals Satb1 and Runx1 from many genomic sites, thereby repressing T cell gene expression indirectly. Second, PU.1, opens chromatin, recruits Satb1 and Runx1 to new sites, and directly activates its target gene expression. In summary, we here present a model where a transcription factor can work through redeployment of other factors and not only through sites that it binds itself.
