Unmasking retroviral vector-mediated leukemogenesis following bone-marrow transplantation in the mouse. (a) Li et al.  used a replication-defective retroviral vector, similar to that used in the patient trials , but this time encoding a commonly used marker gene dLNGFR. Because the bone-marrow cells were infected ex vivo, the re-infused population almost certainly contained a clone in which the dLNGFR retroviral vector had established a pre-transforming environment (thus providing transformation signals 1 and 2 by mechanisms described below). Transplanting this clone into an irradiated mouse meant that the pre-malignant clone had plenty of immunological space to differentiate into a T cell and then to undergo multiple cell divisions in order to re-populate the depleted T-cell compartment. In some progeny cells further genetic mutations would occur to mature the transforming process (signals 3 and 4). But within the life time of the recipient mouse, there was still insufficient expansion of the pre-malignant clone to produce a daughter cell with all of the required signals for full leukemic transformation. That mouse was therefore phenotypically normal with no apparent disease. This is where many of the pre-clinical studies on bone-marrow cell transplantation with vector-modified cells were terminated with an apparently safe outcome. (b) Li et al.  took the experiment one critical step further. Transplanting the bone marrow of the first mouse into a second irradiated recipient allowed for further homeostatic expansion of the pre-malignant cells, which was sufficient for the final necessary transforming mutations (signal 5, or more) to accumulate in the genome and to produce complete transformation of a single cell. Uncontrolled proliferation of this cell leads to leukemia. (c) Molecular analysis of the leukemias that developed in six of ten mice in the study by Li et al.  revealed the molecular pathways by which the dLNGFR gene could contribute to at least the early, pre-leukemic mutations. In these gene-replacement trials, the dLNGFR gene was meant to encode an inert molecule to tag infected cells. But it in fact encodes a receptor for a growth factor (a neurotrophin) and in the construct used by Li et al.  it was modified to remove its cytoplasmic domain. This deleted domain is usually proapoptotic. In hematopoietic cells, the deleted version of the receptor can contribute to signaling of growth signals (signal 1) but now presumably without the pro-apoptotic domain to keep its signaling activities in check. These growth signals alone are insufficient to transform T cells, but in the presence of additional growth-deregulating mutations within a cell the accumulated signals (1-5, for the sake of example here) can lead to complete malignant transformation of a T cell. The authors also demonstrated that, in all cases, the vector had integrated into the same cellular locus, the Evi1 gene. Evi1 is a transcription factor whose deregulation by the viral sequences in a precursor T cell probably provides a second predisposing transforming signal (signal 2 in this case). Co-operation between the dLNGFR and Evi1 signals would then set the stage for the evolution of further transforming signals (3, 4, 5, and so on) in the progeny cells as they undergo the enforced homeostatic proliferation described in (b) and (c).