AML Cells Express Exclusively the Long Isoform of LEF1 and Are Highly Vulnerable to Blockage of LEF1-β-Catenin Binding

Despite progress in the understanding of the biology of acute myeloid leukemia (AML) in recent years, mortality of the disease is still high. One reason for this is drug resistance of leukemic stem cells (LSCs), which are responsible for leukemic growth and relapse. Thus, there is great interest in understanding factors driving LSCs and in key differences between normal hematopoietic stem cells (HSC) and their leukemic counterparts. LEF1 acts via the canonical Wnt pathway by interacting with β-catenin (long isoform). Of note, a natural isoform of Lef1 is expressed in hematopoiesis, which lacks the β-catenin binding domain at the N-terminus and whose function is not well defined in normal and leukemic hematopoiesis (short isoform). Here we demonstrate in a cohort of 111 AML primary patient samples with normal karyotype, that exclusively the long isoform of LEF1 is expressed. These analyses were extended to functionally validated LSC populations, using the NSG xenograft model, and confirmed the highly predominant expression of the long isoform of LEF1 and the virtual lack of expression of the short isoform. This stood in clear contrast to normal HSCs: in highly purified CD34^-/CD38^-/CD93^high HSCs, categorized as most immature human quiescent HSC population, as well as CD34⁺/CD49f⁺/CD90⁺ and CD34⁺/CD49f⁺/CD90^- human HSCs, no expression of the long isoform could be detected with an exclusive expression of the short isoform of LEF1. In multipotent progenitors MPPs (CD34⁺/CD49f^-/CD90^- ) there was a shift towards expression of both isoforms. Thus, LSCs show an inverse expression pattern of LEF1 isoforms compared to their normal counterparts. To understand the function of the long versus short isoform, we cloned the murine long isoform and the Lef1 isoform lacking the ß-catenin binding domain at the N-terminus into retroviral expression vectors and transduced primary murine bone marrow cells. In line with our observation of the exclusive expression of the long isoform of LEF1 in human AML stem cells, we previously showed that expression of the long isoform induces AML in transplanted mice (Petropoulos et al., JEM 2008). In contrast, expression of the short isoform did not have any major effect at the level of repopulating stem cells as determined by quantifying CRU frequency in limit dilution transplantation assays. Furthermore, ChiP-Seq analyses showed predominant binding of the short isoform on genes associate re-entry of HSCs into quiescence such as like CD81 and Wnt5a, in line with our observation of exclusive expression of the short isoform in the human CD34^-/CD38^-/CD93^high HSC population. Based on the observation that AML cells exclusively express the long isoform, which acts via β-catenin binding, we hypothesized that AML cells would be particularly vulnerable to drugs blocking the binding of LEF1-β-catenin binding. The small molecule inhibitors Cercosporin and Calphostin C (100nM), previously shown to block LEF1-β-catenin binding in CLL cells, reduced colony growth of the AML cell line THP1 by 100 % (p<0,0001) and impaired engraftment in NSG mice (26d control vs 32d Cercosporin 100nM, p=0,0013, and 53d Calphostin C, p= 0,0022). In contrast, normal CD34⁺ cord blood cells were much less vulnerable to these compounds (0,2% reduction of CFC growth with Cercosporin, 8% reduction in Calphostin C, not significant). Taken together, these data point to distinct LEF1 isoform expression in AML, thereby creating high vulnerability towards blockage of LEF1-β-catenin binding and forming a potentially targetable Achilles' heel at the level of leukemic stem cells.