A Specific Mechanism By Which NPM1 mutations Impede Myeloid Differentiation Also Explains The Link With DNMT3A Mutation

NPM1 and DNMT3A mutations are highly recurrent (∼30% of cases) in de novo acute myeloid leukemia (AML). Besides aberrant cytoplasmic localization of mutated NPM1, and decreased DNA methyltransferase activity of mutated DNMT3A, the molecular mechanisms of pathogenesis are mysterious. Here, novel specific molecular mechanisms are demonstrated. These results build on and are separate from data presented (poster) at ASH 2012.

LC-MSMS analysis of murine hematopoietic cells was used to analyze the protein interaction network of the key monocyte/macrophage differentiation-driving transcription factor Pu.1. Npm1 was noted to interact with Pu.1 in these analyses, confirmed by bidirectional co-immunoprecipitation-Western blot assays. This interaction suggested that PU.1 might be dragged into the cytoplasm by mutated NPM1 in AML. Accordingly, in the OCI-AML3 cell line that contains an NPM1 mutation, but not in OCI-AML2 cells with wild-type NPM1, PU.1 was co-dislocated into the cytoplasm together with NPM1, obvious by Western analysis of cellular fractions, and by immunofluorescence (IF) assays. IF analysis of primary AML cells from patients with NPM1 mutated AML confirmed this cytoplasmic dislocation of both NPM1 and PU.1 (n=3), with strong cytoplasmic instead of nuclear staining of both proteins. In contrast, primary AML cells with wild-type NPM1demonstrated the expected strong staining of these proteins in the nucleus but not in the cytoplasm (n=3).

NPM1 movement into the cytoplasm is mediated by CRM1 (exportin). Thus, we hypothesized that antagonizing mutated NPM1 interaction with CRM1 would retain mutated NPM1 and PU.1 in the nucleus. Decoy peptides based on NPM1 C-terminal sequences, to hopefully minimize disruption to CRM1 interactions with other proteins, were designed to antagonize NPM1 binding to CRM1. These decoy peptides were combined with different nuclear delivery sequences including the TAT peptide, and with fluorescent tags for tracking. All six peptides entered AML-OCI3 cells (shown by IF), with the strongest signals observed for peptides fused to TAT and TAT-NES (nuclear export signal). The decoy peptides significantly and substantially inhibited cell growth of AML-OCI3 cells (>3-fold reduction), accompanied by an increase in expression of the monocyte differentiation marker CD14 (quantified by flow-cytometry). In contrast, TAT peptide alone, as a control, did not inhibit cell growth or induce monocytic differentiation. Western and IF analyses was used to study PU.1 and NPM1 localization: decoy peptide treatment clearly increased nuclear presence of both PU.1 and NPM1, although these proteins remained detectable in cytoplasmic fractions also.

At ASH 2012, we showed that OCI-AML3 cells have high nuclear CEBPA and retain granulocytic differentiation potential, readily induced by ATRA. Thus, we hypothesized that the NPM1-induced differentiation block, that is specific for the monocytic lineage, creates selective pressure for cooperative mutations that derepress monocyte-commitment. DNMT3A interacts with polycomb proteins that repress lineage-programs, and in NPM1/DNMT3A double mutant versus NPM1 mutant AML, double mutation was significantly associated with M4/M5 (14/23 [61%]) versus M1 morphology (4/13 [30%], p<0.001), higher peripheral blood absolute monocyte counts (median 3.96 v 0.37, p=0.04) and bone marrow non-specific esterase positivity (14/21 [67%] v 7/19 [37%], p=0.06)(TCGA). Chromatin-immunoprecipitation (ChIP) analyses in OCI-AML3 (NPM1/DNMT3A double mutant) and THP1 cells (NPM1/DNMT3A wild-type) that have similar DNMT3A protein expression, demonstrated DNMT3A localization at all (n=7) PU.1 proximal promoter and distal enhancer elements quantified by QRT-PCR (2-40 fold enrichment over ChIP with IgG). Consistent with decreased methyl-transferase activity of the recruited mutated DNMT3A in OCI-AML3, methylation levels at 9 CpG in the PU.1proximal promoter (chr11: 47399783–47400031, pyrosequencing) were significantly lower in OCI-AML3 than in THP1 cells (30-60% of methylation level in THP1). PU.1 mRNA expression was strikingly higher (>40-fold, p<0.001) in OCI-AML3 cells compared to THP1, although THP1 cells also have M5 morphology.

Thus, specific molecular mechanisms explain the lineage-associations and coincidence of NPM1 and DNMT3A mutations in AML. These mechanisms provide guidance for novel treatment methods.