Direct Pharmacological Inhibition of the Transcription Factor PU.1 in Acute Myeloid Leukemia

Functionally critical decreases in levels or activity of the ETS family transcription factor PU.1 are present in approximately 2/3 of patients with acute myeloid leukemia (AML), across different AML subtypes (Sive, Leukemia 2016) including at the stem cell level (Steidl, Nat Genet 2006; Will, Nat Med 2015). Thus, targeting PU.1 could be an appealing option for treatment. As complete loss of PU.1 leads to stem cell failure (Iwasaki, Blood 2005), we hypothesized that PU.1 inhibition could eradicate leukemic cells harboring already low levels of PU.1, with modest effects on normal cells.

We initially tested this hypothesis using 3 different shRNAs, and found that PU.1 inhibition led to a significant decrease in proliferation and clonogenicity, and increased apoptosis of mouse and human leukemic cell lines with low PU.1 levels, as well as the majority of primary human AML cells tested. We demonstrated that these effects were indeed due to decreased PU.1 levels by retroviral add-back experiments.

The direct pharmacologic targeting of transcription factors has proven challenging in the past. Besides the core ETS binding motif (GGAA) in the DNA major groove, PU.1 binding to chromatin depends on additional minor groove contacts enriched for AT nucleotides upstream of the ETS motif, which determine selectivity for PU.1. Using an integrated screening strategy utilizing biosensor surface plasmon resonance, DNA footprinting, and cell-based dual-color PU.1 reporter assays, we developed novel small molecules of the heterocyclic diamidine family acting as first-in-class PU.1 inhibitors. Targeted occupancy by our compounds in the minor groove induces perturbations in DNA conformation that are transmitted to the PU.1 site in the major groove and thus inhibits PU.1 binding via an allosteric mechanism. Consistent with this, the inhibitory effects were selective for PU.1 versus other ETS transcription factors. Treatment with 3 different compounds led to cell growth inhibitory effect with respect to PU.1 level and preferentially affects PU.1^low AML cells. Similarly to what we observed with shRNAs, treatment with our novel inhibitors led to decreased proliferation and colony forming capacity, increased apoptosis, and disrupted serial replating capacity of PU.1^low AML cells and a majority of primary AML cell samples. Targeted ChIP and expression analysis showed that the compounds disrupt PU.1-promoter interaction and lead to downregulation of canonical PU.1 transcriptional targets in AML cells, confirming on-target activity in AML cells. Genome-wide analysis showed highly significant enrichment of known transcriptional targets of PU.1, and selectivity over genes regulated by other ETS family members. Comparison with published transcriptomic and PU.1 ChIP-seq data sets, as well as ARACNe analysis of the PU.1 regulon in primary AML cells, demonstrated that the inhibitors antagonize PU.1-regulated pathways at a genome-wide level. ChIP-seq performed in PU.1^low AML cells confirmed a genome-wide decrease of PU.1 peaks after treatment and provides novel insight into the molecular mechanisms mediating the anti-leukemic effects of pharmacological PU.1 inhibition.

To test the effects of PU.1 inhibition on normal hematopoiesis, we treated normal hematopoietic stem/progenitors cells (HSPC) in colony forming assays and saw decreased production of mature granulo-monocytic cells, consistent with PU.1's known role in this lineage. However, this effect was reversible upon drug removal, and serial replating capacity was not affected suggesting no significant effects on more immature HSPC. Congenic transplantation assays of treated normal bone marrow cells led to no change in myeloid and T-cells and only a modest decrease in B-cell numbers.

Lastly, in vivo treatment with PU.1 inhibitors in mouse and human AML (xeno)transplantation models significantly decreased tumor burden and increased survival.

To conclude, our study provides proof-of-principle for PU.1 inhibition as a novel therapeutic strategy in AML. Furthermore, we present the development of first-in-class PU.1 inhibitors acting via an allosteric minor groove-mediated mechanism. Our work shows that the specific pharmacological targeting of the DNA interaction of transcription factors such as PU.1 is feasible in principle, and may open the way for targeting of other transcription factors through minor groove-directed approaches.