More complexity in MLL-associated leukemias

The mixed lineage leukemia gene (MLL, also known as ALL-1, and HRX) has rightly attracted much interest as a major player in leukemia.¹ MLL's central role is clear by its involvement with over 30 different partner genes in recurrent translocations. As if this were not enough, MLL is also implicated in leukemia by overexpression in the absence of overt mutations or by acquisition of partial tandem MLL duplications. What, then, accounts for the leukemogenicity of MLL? Might there be some common functional thread tying together many of the fusion genes and MLL overexpression? At least one strong clue has emerged from the recognition that a major function of MLL, like its Drosophila homolog Trithorax, is to serve as a maintenance factor for the expression of many members of the Hox family of transcription factors. Hox genes are now recognized as major components of the regulatory machinery of primitive hematopoietic cells. Strikingly, multiple lines of evidence link Hox genes directly to leukemic transformation.^2,3  This evidence includes induction of leukemia in mice following engineered overexpression of certain Hox genes (eg, HOXA9 and HOXA10) and the observed overexpression of multiple Hox genes in human leukemia and, notably, in MLL-associated leukemias. Perhaps most convincingly, multiple members of the Hox family, HOXA9 for one, have been identified as translocation partners in leukemias with the common partner Nucleoporin 98.⁴  Thus, a satisfying model for some if not all MLL-induced leukemias would be through induced deregulation of key Hox target genes. Strong support for this has recently been reported by Cleary and colleagues, who found that MLL-ENL lost leukemogenicity in bone marrow cells taken from Hoxa7 or Hoxa9 knockout mice.⁵  The jump to a unifying model involving MLL and HOXA9, however, is not without a tumble or two as indicated in the article by Kumar and colleagues (page 1823) in this issue of Blood. Their studies reveal unabated leukemogenicity by the fusion gene MLL-AF9 in the absence of Hoxa9. While there were clear influences of Hoxa9 on the phenotype of the leukemia, the essential transformation was not altered by the presence or absence of Hoxa9. The striking differences between these 2 recent studies involving related partner genes may be a consequence of several experimental and biologic differences. Kumar and colleagues have used a gene knock-in model of MLL-AF9 fusion rather than retroviral overexpression; the fusion genes may indeed have differential effects on Hox targets, rendering other members of the cluster more or less important. Indeed, multiple members of the Hox A cluster were observed to be up-regulated by MLL-AF9, making it possible that additive levels of Hox gene expression may be critical for hematopoietic effects. These studies caution against oversimplification in thinking about the likely complex paths from MLL to leukemia but further highlight the potential linkage to Hox genes and the power of genetic mouse models for investigating human disease.