Down-Regulation of CEBPA Expression Has a Stronger Impact on Myeloid Differentiation in Human Versus Mouse Cells: Mechanistic Implications for CEBPA Mutations in AML.

Acute myeloid leukemia (AML) is often associated with gene mutations that affect transcription factors regulating differentiation. CEBPA mutations occur in 5–10% of AML cases. About half of those result in the translation of an N-terminal truncated C/EBPα protein, termed p30. While C/EBPα is not required for adult hematopoiesis, its ablation leads to a block in granulocytic differentiation at the blast stage. Earlier studies showed that p30 dominantly inhibits the transcriptional function of wtC/EBPα. We have previously shown that expression of p30 in human but not murine hematopoietic progenitors leads to the accumulation of myeloblasts. Here, we investigated the role of p30 C/EBPα in AML induction by using different dimerizing mutants of C/EBPα. In addition, we sought to determine if down-regulation of CEBPA expression mimicked the effect of p30 in human hematopoietic progenitors. To evaluate the requirement for dimerization and transcriptional activity of the p30 C/EBPα, a mutant was generated in which the first two leucines of the leucine zipper were changed to valine (C/EBPα L12V), which has been previously shown to disrupt C/EBPα function. CD34⁺ cells isolated from human cord blood were transduced with retroviral vectors expressing either GFP alone, human or murine p30 C/EBPα, or murine p30-L12V and plated into methylcellulose. While murine mutC/EBPα and human mutC/EBPα showed the expected block in granulocytic differentiation and inhibition of erythrocyte lineage commitment (mean number of 0 ± 2 CFU-GM), mutC/EBPα L12V had no effect as shown by the same colony distribution as the control vector (mean number of 20 CFU-GM). This observation implies that mutC/EBPα requires dimerization and/or transcriptional activity for the inhibition of human hematopoietic differentiation. To predict if p30 homodimerization or heterodimerization between p30 C/EBPα and wtC/EBPα is required for this activity, we replaced the leucine zipper of the p30 C/EBPα with the GCN4 exchange mutant to form p30-GZ, which allows homodimerization but not heterodimerization with wtC/EBPα. We demonstrate here that expression of p30-GZ showes an almost identical phenotype as mutC/EBPα arguing that p30 homodimerization is important for its activity while heterodimerization with wtC/EBPα is not necessary. To test the effects of down-regulation of C/EBPα and to compare the induced phenotype with the one induced by p30 C/EBPα, we generated a short-hairpin (sh) RNA against human C/EBPα and cloned it into a retroviral expression vector. Human CD34⁺ cells from cord blood were transduced with these vectors and evaluated for the down-regulation of C/EBPα mRNA. Expression of C/EBPα shRNA reduced C/EBPα mRNA levels to 25%. Examination of liquid cultures (with GM-CSF, SCF, IL3) by using FACS analysis demonstrated a relative increase in erythroid cells: 32% CD71⁺/CD13− C/EBPα siRNA versus 7% CD71⁺/CD13⁻ control vector, and substantial defect in proliferation of myeloid cells (0% CD14 and 4% CD15 C/EBPα siRNA compared with 10% CD14 and 12% CD15 control vector). Most striking were results in methylcellulose cultures, where a dramatic reduction in CFU-GM colonies (mean number of 8 ± 2 CFU-GM C/EBPα siRNA versus 37 ± 4 CFU-GM control vector), but also BFU-E (mean number of 18 ± 3 BFU-E C/EBPα siRNA versus 42 ± 4 BFU-E control vector) was observed in cultures expressing shRNA against C/EBPα. In addition, an almost complete block in terminal differentiation of both monocytic and granulocytic differentiation was observed in cells dispersed from the colonies. Using shRNA technology, we show that C/EBPα is required at two stages of human myeloid differentiation:

1. during early stages of myeloid/erythroid commitment and

2. for terminal granulocytic and monocytic maturation.

In contrast to mouse studies, the block to terminal differentiation could not be relieved by GM-CSF stimulation. This shows that C/EBPα activity is not readily substituted by alternative pathways in human myeloid differentiation. We thus propose that p30 induces myeloid differentiation at early stages of differentiation but cannot promote terminal differentiation. This coupled with reduced absolute levels of C/EBPα leads to accumulation of myeloblasts, a hallmark of AML.