Acute myelogenous leukemia (AML) is often associated with mutations in transcription factors that are essential for normal hematopoietic development and differentiation. The basic leucine zipper (bZIP) protein C/EBPalpha is mutated in 7–10% of AML. Two different classes of C/EBPalpha mutations are found in AML: 1) mutations that introduce a stop codon between the primary translational initiation site and a downstream ATG, resulting in translation of an N-terminally truncated protein (p30) 2) mutations that disrupt the basic region of the bZIP domain in the C-terminus, abolishing DNA binding. Of note, biallellic mutations are often but not always observed. It has been proposed that p30 is dominant-negative over wt C/EBPalpha (p42), blocking its transcriptional activity. We have previously shown that p30 blocks erythrocytic commitment and inhibits terminal granulocytic differentiation when expressed in human but not murine primary hematopoietic cells. To identify the molecular mechanism by which the p30 inhibits differentiation and to evaluate the role of homo- and heterodimerization between p30 and p42 in leukemogenesis, we created a p30 that carries point mutations in bZIP (p30-L12V) that disrupt dimerization. Expression of p30-L12V had no effect on the differentiation of human CD34+ cells, arguing for the requirement of dimerization. To inhibit potential heterodimerization between p30 and p42 but permit p30 homodimerization, we generated a p30-GZ, in which the leucine zipper was replaced by an artificial dimerization domain. Expression of p30-GZ induced a similar differentiation block as the normal p30, arguing against a dominant-negative function of p30 over p42 mediated by dimerization. In a second approach to understand the mechanism by which p30 inhibits differentiation, we wanted to identify gene targets of p30. For this purpose, p30 was fused to the modified ligand-binding domain of the estrogen receptor (p30-ERtm). Expression of p30-ERtm in CD34+ cells in the presence of tamoxifen showed a phenotype very similar to the one observed after the expression of p30, while in the absence of tamoxifen no effect was observed. We used this system of inducible p30 expression in human CD34+ cells to evaluate gene expression patterns. If p30 is dominant-negative and interferes with the transcriptional activation of p42, reciprocal expression patterns of target genes should be observed for wt C/EBPalpha and p30. Our data show that homodimerization of p30 is necessary and sufficient to block erythrocytic commitment and terminal granulocytic differentiation of human CD34+ cells and to induce known C/EBPalpha target genes. These data argue that the formation of p30/p42 heterodimers is not required for the effect of C/EBPalpha p30, questioning the proposed dominant-negative function.

Author notes

Corresponding author

Sign in via your Institution