To switch or not to switch: PU.1 expression is the question

In this issue of Blood, articles by van der Kouwe et al¹ and Trinh et al² characterize the complex regulation of PU.1 (SPI1) expression involving the interplay of a single enhancer with the canonical sense and the alternate antisense promoter.

The transcription factor PU.1 is a master hematopoietic regulator involved in hematopoietic stem cell maintenance and myeloid and B lymphoid lineages.^3,4 PU.1 does not function as an on/off switch; instead, graded expression levels determine its precise role in specific cell types,⁵ and expression levels must be tightly controlled throughout development. Levels of PU.1 increase during myelopoiesis,⁶ but expression is reduced in a range of acute myeloid leukemia (AML) cells, particularly those with core binding factor (CBF) fusions [RUNX1-ETO caused by t(8;21) or CBFβ-MYH11 caused by inv(16)].^4,7 Reduced expression, but not complete ablation, in mice leads to AML.⁶ Conversely, reexpression of PU.1 can cause AML cells to differentiate.⁷ A key PU.1 upstream regulatory element (URE) is located upstream of the sense promoter (proximal promoter [PrPr]), and this can be positively regulated by RUNX1, which binds directly to the URE^4,8-10 (see figure, panel A). PU.1 expression can be negatively regulated by a long noncoding RNA (lncRNA) (PU.1 antisense RNA [asRNA]), which reduces both PU.1 messenger RNA (mRNA) and the protein and is initiated by an antisense promoter (AsPr) located in exon 3¹¹ (see figure, panel A). As with the PrPr, AsPr interacts with and is regulated by the URE. However, the mechanisms of spatiotemporal control of PU.1 expression are not completely resolved.

In their article, van der Kouwe et al find that the ratio of PrPr to AsPr chromatin accessibility and transcriptional activity changes during development, with increased PrPr accessibility and PU.1 mRNA expression in the myeloid and B lymphoid lineages. In contrast, high AsPr accessibility precedes complete shutdown of the PU.1 locus during T-cell maturation. Because AsPr and PrPr both loop with and are regulated by the URE, the authors then investigated chromatin interactions at the PU.1 locus. In the T-lymphoid context, the URE is predominantly looped with AsPr with a concomitant high asRNA:mRNA ratio (see figure, panel B), whereas in myeloid cells, the URE interacts predominantly with PrPr, which favors sense transcript expression (see figure panel C). Interestingly, RUNX1 binds to the URE and both PrPr and AsPr, which suggests that it controls PU.1 expression during hematopoiesis by modulating the ratio of sense to antisense transcription.

Precisely how does RUNX1 occupation affect looping between URE and the promoters? The study by Trinh et al provides complementary new insights. They screened for RUNX1-interacting RNAs and identified a polyadenylated lnc-RNA termed LOUP that is myeloid specific and originates from the URE. LOUP facilitates chromatin loop formation between the URE and PrPr by recruiting RUNX1 to both elements thereby promoting PU.1 expression and, ultimately, myeloid differentiation with concomitant inhibition of proliferation (see figure, panel C).

Given the known roles of PU.1 and RUNX1 in AML, both studies focused on PU.1 regulation in leukemia. Similar to wild-type RUNX1, the CBF fusion proteins RUNX1-ETO and CBFβ-MYH11 are able to transactivate AsPr, which suggests the intriguing possibility that PU.1 perturbation in CBF leukemias might be mediated by AsPr. In line with this finding, CBF AMLs have increased AsPr accessibility and asRNA expression compared with AMLs with a normal karyotype. Furthermore, RUNX1-ETO also binds to the URE and diminishes LOUP expression. Knockdown of RUNX1-ETO, which also induces a more mature myeloid phenotype, increases LOUP levels, reduces the PU.1 asRNA:mRNA ratio, and promotes interaction between URE and PrPr (see figure, panel D). Thus, in RUNX1-ETO AML, PU.1 expression is actively repressed by a chromosomal structure that is unfavorable to interaction of the URE with the sense promoter and enhances expression of an antisense transcript that is known to repress PU.1. Furthermore, RUNX1-ETO–mediated repression of the PU.1 locus seems to require active maintenance, similar to other reports of RUNX1-ETO–induced chromatin remodeling.¹² The fact that PU.1 repression by this mechanism is reversible may be an opportunity for therapeutic targeting.

Key findings of these studies are: (1) that PU.1 is regulated by competition between a coding and a noncoding promoter for interaction with the enhancer, (2) that an enhancer-driven lncRNA promotes looping with the sense promoter, and (3) that CBF AMLs can co-opt these 2 mechanisms to reduce PU.1 expression. Questions remain regarding precisely how the RUNX1-ETO fusion protein mediates altered chromatin looping and whether this is an immediate consequence of fusion gene expression or requires previous chromatin remodeling or destabilization. RUNX1-ETO alters global chromatin conformation and promoter-enhancer interactions.¹² One possibility is that the global effects of fusion gene expression favor a chromatin conformation in which RUNX1-ETO can bind AsPr and modify its chromatin interactions. Alternatively, RUNX1-ETO might directly promote URE-AsPr looping by homodimerization or interaction with other factors bound to the URE or AsPr. It also remains unclear whether PU.1 promoter competition is a mechanism for PU.1 downregulation in other AML subtypes, or indeed whether similar regulatory modules are used by other genes. In this context, it is interesting to note that multiple established and potential therapeutic targets such as CCND2, CDK6, and WT1 show arrangements of sense and antisense promoters that are controlled by leukemic fusion proteins.¹³ 

These studies extend our understanding of how PU.1 expression is tightly controlled during normal hematopoiesis and add to the increasing catalog of epigenetic mechanisms by which oncogenes perturb normal gene regulation. The hope is that these collective insights will reveal leukemia vulnerabilities and point to new therapeutic possibilities for AML.