Graded PU.1 Levels Regulate Granulocyte Vs. Macrophage Genes Via Multiple Enhancer Elements

PU.1 is a transcription factor absolutely required for normal hematopoiesis. Cumulating evidence indicates that precise levels of PU.1 expression are critical for differentiation to distinct blood lineages, and if perturbed, even modest decreases in PU.1 can lead to leukemogenesis. In contrast to extensive knowledge of regulation of PU.1 gene itself, the mechanism of how target genes senses different PU.1 levels remain largely unknown.

To address this, we used PU.1^-/- mouse myeloid progenitors encoding inducible PU.1 transgene (PU.1ER, PUER, Walsh 2002) that allows tight control of PU.1 activity. Interestingly, intermediate PU.1 activity induced differentiation of PUER progenitors into granulocyte like cells, while high PU.1 produced macrophages, supporting the model that different PU.1 expression is not a consequence but a driver of cell fate choice. Global expression analysis using 4 different levels of PU.1 at 8 time points (2-96 hrs) revealed that granulocyte specific genes were activated exclusively by intermediate PU.1 levels in 3 distinct modes:

1. not expressed in progenitors while strongly induced at intermediate PU.1 (e.g. Gelatinase B (Mmp9) and Neutrophil collagenase (NC)

2. moderately expressed in progenitors while strongly activated at intermediate PU.1 and repressed at high PU.1 (e.g. Myeloperaxidase (Mpo)

3. highly expressed in unstimulated progenitors with expression maintained at intermediate PU.1 but strongly repressed at high PU.1 (e.g. Neutrophil elastase (NE), Proteinase 3 (primary granule proteins), Cebpe and Gfi1 (Growth factor independent1)

Majority of macrophage genes (incl. CD14, Csf1R, Egr2) were regulated as early PU.1 target genes; being gradually activated by high PU.1 activity within 8hrs. However, most granulocyte genes (NE, Mmp9, Mpo, NC but not Cebpe and GFI1) were late activated PU.1 targets (48 and 96hrs) indicating that these genes are coregulated by additional factor(s), likely an early PU.1 target.

Next we analyzed the regulatory sequences (+-50kb) of two genes activated exclusively by intermediate PU.1, Mpo and Mmp9, using own and public ChIP(seq) data of transcription factors (TFs) (PU.1, GFI.1), DNAseI hypersensitive sites, histone modifications (H3K4Me, H3K27Ac, H3K9Ac) and expression of enhancer specific bidirectional ncRNAs (eRNA) (CAGE).

14 Mpo and 16 Mmp9 putative enhancers, selected by above mentioned criteria, were cloned into luciferase vector containing their proximal promoter (PP) and were tested for functional activity in response to PU.1 levels. Interestingly, the PU.1 binding motifs within these regions have a low to intermediate affinity (log of score, Jaspar) and are often present in multiples and/or enriched for binding sites of other lineage determining transcription factors. Although PU.1 bound to all of these DNA regions resembling superenhancer, just a small fraction of PU.1 binding was functionally responsive. Specifically, we identified novel enhancer elements at -3.4 kb and -15kb of MPO which were activated by intermediate (but not high) PU.1 levels. Interestingly, activity of -3.4 kb enhancer required presence of PP, while the -15kb element required presence of both PP and the -3.4kb element. Similar phenomenon was observed at -5kb and +4.6kb (intronic) MMP9 enhancers. Collectively, these observations suggest that a cooperative assembly of several cell type-specific enhancers is required for optimal Mpo and Mmp9 activation.

This model is supported by our Chromosome conformation capture (3C) data identifying 3D interaction of these enhancer elements at intermediate PU.1 levels suggesting that PU.1 binding mediates DNA looping that allows enhancer cooperation. In addition, activity of these enhancers at intermediate PU.1 levels was associated with expression of bidirectional noncoding enhancer RNAs, confirming functionality of these elements.

In conclusion, our data support the model that PU.1 at intermediate concentration binds to low and intermediate affinity binding sites in several enhancers of granulocyte genes, causing their successive looping and interaction with proximal promoter that leads to transcription activation. The role of cooperating TFs, mechanisms of how granulocyte genes are switched off at high PU.1 concentration and deregulation of these mechanisms in AML are being further studied.

Grants 16-05649S P305/12/1033 16-31586A 16-27790A 16-31586A UNCE 204021 PRVOUK P24