Where do the leukaemia relapses come from?

Insights into the cellular origin and etiology of the infant pro-B acute lymphoblastic leukemia with MLL-AF4 rearrangement

Infant acute lymphoblastic leukemia (ALL) involving mixed-lineage leukemia (MLL) fusions has attracted a huge interest in basic and clinical research because of its prenatal origin, mixed-lineage phenotype, dismal prognosis and extremely short latency. Over 90% of infant ALLs are pro-B ALL harboring the leukemic fusion MLL-AF4. Despite the fact that major achievements have provided a better understanding about the etiology of infant MLL-AF4+ ALL over the last two decades, key questions remain unanswered. Epidemiological and genetic studies suggest that the in utero origin of MLL rearrangements in infant leukemia may be the result of prenatal exposure to genotoxic compounds. In fact, chronic exposure of human embryonic stem cells (hESCs) to etoposide induces MLL rearrangements and makes hESC more prone to acquire subsequent chromosomal abnormalities than postnatal CD34(+) cells, linking embryonic exposure to topoisomerase II inhibitors to genomic instability and MLL rearrangements. Unfortunately, very little is known about the nature of the target cell for transformation. Neuron-glial antigen 2 expression was initially claimed to be specifically associated with MLL rearrangements and was recently shown to be readily expressed in CD34+CD38+, but not CD34+CD38- cells suggesting that progenitors rather than stem cells may be the target cell for transformation. Importantly, the recent findings showing that MLL-AF4 rearrangement is present and expressed in mesenchymal stem cells from infant patients with MLLAF4+ ALL challenged our current view of the etiology and cellular origin of this leukemia. It becomes therefore crucial to determine where the leukemia relapses come from and how the tumor-stroma relationship is defined at the molecular level. Finally, MLL-AF4 leukemogenesis has been particularly difficult to model and bona fide MLL-AF4 disease models do not exist so far. It is likely that the current disease models are missing some essential ingredients of leukemogenesis in the human embryo/fetus. We thus propose modeling MLL-AF4+ infant pro-B ALL using prenatal hESCs.