MLL-r impair self-renewal and differentiation properties of HSCs and HSPCs. 79, 80] often fail to reproduce the biphenotypic feature observed in MLL-r leukaemia patients with co-expression of some myeloid and lymphoid genes. Zeisig et al.  first reported a MLL-ENL transformation model based on a biphenotypic lymphoid/myeloid phenotype. Similarly, when infected murine BM cells were cultured in methylcellulose with Flt3-ligand, stem cell factor (SCF) and interleukin-7 (IL-7) to sustain lymphopoiesis , B220+CD19+ and B220+CD19? B cells appeared after 4?weeks. The latter generated leukaemia in vivo characterized by splenomegaly, lymph node enlargement and an overgrown thymus, where cells showed a myeloid morphology, yet expressed the B220 lymphoid marker. Non-retroviral models of MLL-r leukaemia All the studies described so far used retroviral models, which may not generate expression levels representative TRAM-34 of the endogenous gene loci involved in the translocation events. Since expression levels are crucial determinants of cellular programming, it is not surprising that constitutive and conditional knock-in mouse models have provided another powerful approach to analyse MLL-r leukaemias. In 1996, Corral et al.  designed expression of the MLL-AF9 oncogene via homologous recombination . Designed mice, bearing the MLL-AF9 fusion, developed leukaemia restricted to the myeloid lineage despite of the widespread expression of the fusion gene. Moreover, AML development was characterized by long latency suggesting the need of genetic alterations for complete leukaemic transformation. A Cre-Lox recombination approach generated MLL-AF9  and MLL-ENL  mouse models able to rapidly develop AML. Furthermore, de novo MLL-ENL translocations caused myeloproliferative-like myeloid leukaemia development in all mice in which Cre recombinase was expressed from Lmo2, Lck and Rag1 genes (expressed in non-differentiated cells, T-cell linage and early staged of lymphoid lineage, respectively); while no haematological malignancies were observed in MLL-ENL Cd19-Cre (gene expressed in B cell lineage) . Overall, these data demonstrate that this MLL-ENL fusion is usually leukaemic when expressed in stem cells and progenitors excluding the B-cell compartment [87, 88]. Importantly, an endogenous knock-in mouse model using the MLL-AF9 oncogene , exhibited that GMPs were refractory to leukaemic transformation in complete contrast to previous retroviral studies [76, 79]. While only 100 HSCs and 2500 CMPs from knock-in mice were able to produce AML in the majority of the recipients, all mice transplanted with higher doses of GMPs did not develop any disease. The ability of GMPs to be transformed in retroviral studies seems to be related to the different levels of oncogene expressed (170-fold higher then knock-in GMPs). By contrast, a doxycycline (Dox) inducible mouse model, targeting MLL-ENL to the 3 UTR of the Col1a1 gene , showed that both HSCs and MPPs failed to induce leukaemia in vivo . AML development was observed only when Dox was administrated to mice 4?weeks after HSC transplantation suggesting that MLL-ENL expression could interfere with in vivo homing. The authors therefore suggested that granulocyte-monocyte-lymphoid progenitors (GMLPs), as a subpopulation of the wider Lineage?Sca-1+c-kit+ (LSK) population , and GMP precursors (pGMs) represent the most permissive cellular environment for regulatory program perturbations that can cause leukaemia development. Nevertheless, a recent study published in 2016  using another MLL-AF9 Dox inducible mouse model (67), showed that both long-term HSCs (LT-HSCs) and GMP were transformed by MLL-AF9 induction, where transformation in LT-HSCs resulted in a more aggressive AML phenotype. Taken together, rather than providing conclusive answers to the molecular processes underlying AML TRAM-34 development in patients, these studies further spotlight the intricacies of perturbing regulatory programs and the complex interplay of parameters such as cellular context and oncogene expression level. Given that no mouse model seems perfect, it may be argued that research efforts need to be refocused onto molecular studies with human patient samples, especially since genome engineering has become so much easier TRAM-34 with the new clustered regulatory interspaced short palindromic repeats (CRISR) system . NR4A3 Concluding remarks Regulatory programs in HSPCs need to be finely balanced to maintain normal haematopoiesis and are vulnerable to genetic perturbations TRAM-34 that result in the development of malignant disease. Despite the remaining disagreements between the various retroviral and transgenic models, it is clear that the.