Runx1 Haploinsufficiency Permits Lineage-Commitment but Impairs Activation of a Key Late Differentiation Gene

Abstract 2442

Inactivating or dominant-negative RUNX1 mutations are a frequent first-hit in the multi-hit process of leukemogenesis. First-hits that target RUNX1 occur in stem cells, or in familial acute myeloid leukemia (AML), in the germ-line. However, self-renewing AML cells (leukemia-initiating cells, LIC) frequently display surface markers of lineage-committment, and AML cells appear lineage-committed in critical aspects, expressing high levels of key lineage-specifying transcription factors (TF) such as CEBPA with promoter CpG methylation patterns resembling mature hematopoietic cells (HSC) (Negrotto et al, Leukemia 2011).

To investigate a possible basis for differentiation advancement of LIC from stem cell origins, hematopoietic stem cells from wild-type and Runx1 haploinsufficient (+/−) mice were cultured with granulocyte-colony stimulating factor (G-CSF). Repression of Hoxb4 (a stem cell associated factor) and activation of Cebpa by G-CSF was similar in wild-type and Runx1+/− lineage-negative hematopoietic precursor cells. However, the expression of Cebpe, a key late-differentiation driving TF upregulated with the transition from proliferating pro-myelocytes to non-proliferating myelocytes, and which terminates proliferation in myeloid and AML cells, was significantly decreased in the Runx1+/− cells (>2-fold, t-test p<0.01), even after 15 days of culture with G-CSF. This pattern of Cebpa and Cebpe expression was also documented at the protein level. High CEBPA, but relatively low HOXB4 and CEBPE expression, is also characteristic of human primary AML cells, including CD34+ subsets (Negrotto et al, Leukemia 2011) and leukemia initiating cells (analysis of Geo Database GSE24006 and GSE17054). Indicating epigenetic repression of Cebpe, methylation of CpG in the Cebpe promoter was significantly increased in Runx1+/− compared to wild-type cells (p<0.05). Consistent with partial maturation of Runx1+/− cells, expression of the granulocyte-lineage markers Ly6G and CD11b increased during culture with G-CSF, but to a much lesser extent than in wild-type control (30 v 40% and 11 v 30% respectively). After 15 days of culture with G-CSF, in morphological analyses, Runx1+/− cells included granulocytes (with less neutrophilic granulation than observed in wild-type), some cells with high nuclear-cytoplasmic ratio, and mitotic figures, in comparison to mostly mature granulocytes in wild-type control.

Repression of Cebpe might confer a growth advantage after lineage-commitment. In liquid culture with and without G-CSF, Runx1+/− cells demonstrated more rapid and persistent proliferation than wild-type control (cumulative increase in cell numbers >2-fold, p<0.001), with this advantage more prominent in the presence of G-CSF. In semi-solid media supplemented with G-CSF, Runx1+/− lineage-negative cells produced a greater number and larger-sized colonies than wild-type control.

CEBPE promoter CpG that become less methylated during G-CSF-induced differentiation of normal human CD34+ precursors into granulocytes were identified by mass-spectrometry. These CpG were in proximity to RUNX1 and CEBPA consensus binding sequences. Two of the three CpG sites were significantly hypermethylated in AML (n=27) compared to normal (n=11) and/or remission bone marrow cells (n=6) (1.5 to >2-fold, p<0.01, Wilcoxon test). In contrast to the CEBPE promoter CpG, methylation levels at LINE-1 repetitive DNA element CpGs were similar in normal, remission, and AML bone marrow cells.

In conclusion, Runx1+/− abnormality present in stem cells is permissive of lineage-commitment but represses a key late-differentiation gene, hence conferring a proliferative advantage to lineage-committed daughter cells. This mechanism could explain differentiation advancement of LIC from stem cell origins, and likely contributes to the terminal maturation observed upon treatment of AML cells (including LIC) with corepressor antagonists (histone deacetylase and DNMT1 inhibitors). In contrast, treatment of normal HSC with these drugs has the opposite effect, maintaining self-renewal, possibly by preventing repression of stem cell genes by differentiation stimuli (Hu et al, Mol Ca Ther 2010). Hence, the difference in maturation level of LIC and normal HSC can potentially be exploited for AML selective therapy.