Abstract (eng)
Acute myeloid leukemia (AML), the most common leukemia subtype accounting for 43% of leuke-mia deaths, arises from heterogeneous and dynamic patterns of driver mutations, most of which remain undruggable. Despite their diversity, individual mutations converge to dysregulate a lim-ited number of cellular processes, which is thought to result in leukemia-specific dependencies. Strategies aimed at exploiting “non-oncogene addiction” to chromatin and cell-cycle regulators have recently shown promise in a wide range of AML subtypes. To accelerate the search for rapidly translatable non-oncogene addiction targets, we have constructed a custom shRNA library target-ing 1133 candidate genes including known targets of established small-molecule inhibitors and tool compounds as well as potentially druggable genes that were found overexpressed in high-risk MLL-rearranged AML. Using multiplexed screens in an MLL-AF9;NrasG12D-driven AML mouse model and murine embryonic fibroblasts, we identify and validate a set of non-essential genes whose suppression triggers dramatic anti-leukemic effects. Strikingly, among the ten most potent leukemia-specific sensitivities, we identify five genes involved in diverse metabolic processes.
Both in the primary screen and secondary validation studies in murine and human AML contexts, shRNA-mediated suppression of GART, a tri-functional enzyme in de novo purine synthesis, turned out to be the most prominent leukemia-specific vulnerability. Remarkably, side-by-side compari-sons to previously proposed and established targets (including BRD4, PLK1 and various metabolic targets), reveal that the potency and leukemia-specificity of these effects are unmatched. Inter-estingly, available antifolate inhibitors of GARTfase (catalyzing the second GART-dependent enzy-matic step), do not phenocopy the effects of shRNA-mediated GART suppression. In addition, sev-eral cDNA rescue experiments with mutated versions of Gart suggests that the GARTfase activity does not represent the key bottleneck in AML and the AIRS activity should be targeted for a po-tential new therapy. Importantly, GART is located on chr21 in a region commonly amplified in pediatric B-ALL, suggesting that GART may have driver functions in iAMP21+ and Down syndrome-associated leukemia. Furthermore, reanalysis of expression profiling studies reveal that GART is one of the most commonly overexpressed genes in a wide variety of hematopoietic and other malignancies. In summary, our study identifies inhibition of GART as promising strategy to thera-peutically exploit the increased demand for nucleotide building blocks, which has been proposed as a hallmark of cancer cell metabolism.