Although many leukaemias have been associated with chromosomal translocations, the means by which a resulting fusion protein can transform a cell is not well understood. For instance, the AML1–ETO fusion protein, which is associated with acute myeloid leukaemia (AML), immortalizes haematopoietic progenitor cells by an unknown mechanism. In the July issue of Nature Medicine, Bryan Linggi et al. report that AML1–ETO can repress transcription of the ARF tumour suppressor.

The t(8;21) translocation, which is associated with 12–15% of all AML cases, fuses the DNA-binding domain of AML1 and the eight-twenty-one (ETO) corepressor. This fusion protein binds to the same DNA sequence as normal AML1, but functions as a transcriptional repressor instead of an activator. AML1–ETO expression extends the lifespan of primary myeloid progenitor cells in culture, so Linggi et al. investigated whether it affects components of the p53 checkpoint pathway.

The authors started by looking for AML1 binding sites in various gene promoters. Although the TP53 gene (which encodes p53) itself does not have an AML1 binding site, CDKN2A (which encodes ARF; also known as p14 in humans and p19 in mice) promoter contains eight AML1 binding sites. ARF antagonizes MDM2 to stabilize p53, and loss of ARF has been shown to impair p53-mediated growth arrest and apoptosis.

The authors found that expressing AML1 in cultured cells activates the transcription of the ARF transcript, and also induces senescence in fibroblasts. Expression of AML1–ETO, on the other hand, blocked AML1-mediated induction of the ARF-transcript promoter and repressed endogenous ARF-transcript expression, leading to immortalization. INK4a (also known as p16) has the same genetic locus as the ARF transcript, but is transcribed from a distinct promoter that lacks AML1 binding sites. Linggi et al. did not observe a significant difference in INK4A levels in cells expressing AML1–ETO, indicating that repression by this fusion protein is specific for ARF.

But is expression of the ARF transcript reduced in human leukaemia cells? Bone-marrow and blood samples from two cohorts of patients with t(8;21)-associated AML were independently tested for the expression of the ARF transcript by real-time polymerase chain reaction. These cells were found to express significantly less ARF-transcript mRNA than other types of leukaemia cells. The repression of expression of the ARF transcript might therefore explain why p53 is not mutated in these tumours. The authors suggest that ARF might also be repressed by other cancer-associated translocations that affect AML function.