Intersecting Chemical Genomic and Genetic Screens Identifies Glycogen Synthase Kinase-3α (GSK-3α) as a Modulator of Differentiation In Acute Myeloid Leukemia

Abstract 1000 The treatment of acute myeloid leukemia (AML) poses a vexing challenge despite an improved understanding of its molecular pathogenesis. A two-hit theory has been proposed for the pathogenesis of AML where the first hit imparts a proliferation defect and the second a block in differentiation. Drug discovery efforts for AML, however, have largely focused on the proliferation defect. Using the intersection of chemical biology and high-throughput genetic screening we sought to identify new AML differentiation targets by measuring the induction of a complex gene expression signature of myeloid maturation. We performed two independent small molecule library screens and a high-throughput shRNA screen for perturbations that induce differentiation in AML cells. We measured this differentiation signature using the previously described gene expression-based high-throughput screening (GE-HTS) approach in which gene expression signatures serve as surrogates for different biological states. Glycogen Synthase Kinase-3 (GSK-3) emerged as a target at the intersection of these three screens. GSK-3 is a multifunctional serine threonine kinase involved in diverse cellular processes including differentiation, signal transduction, cell cycle regulation and proliferation, with an emerging role in human leukemia. We demonstrate that the GSK-3 inhibitors scoring in the primary screens indeed induce the differentiation signature with a dose-response in AML cell lines. In order to further validate GSK-3 as a target, we extended testing to lithium chloride and SB216763, two commonly used GSK-3 inhibitors not in the original screens. Both of these molecules induced AML differentiation as measured by gene expression and morphological changes in multiple AML cell lines and in primary patient blasts in vitro. GSK-3 is expressed as two highly homologous but non-redundant isoforms, GSK-3α and GSK-3β, with small molecule inhibitors non-selectively reported to target both. In order to further validate the findings of the small molecule library screens, we performed a high-throughput shRNA screen targeting the human kinome for shRNAs that induce differentiation. Multiple hairpins against GSK-3α scored. In secondary testing of four AML cell lines, we found that genetic loss of GSK-3α induced differentiation as measured by induction of the complex gene expression signature, alterations in genome-wide expression, and morphological changes associated with maturation. Moreover, colony formation in methylcellulose was impeded. These effects could be rescued with a GSK-3α cDNA immune to the effects of the shRNA, further supporting the on-target activity of the hairpin. In contrast, GSK-3β-directed hairpins induced minimal differentiation. We next extended testing to an in vivo U937 orthotopic model of AML. While pan-GSK-3 inhibition with lithium chloride treatment did not demonstrate efficacy in this model, inhibition of GSK-3α with shRNA attenuated development of disease compared to an shRNA control. We hypothesize that the rise in β-catenin with pan-inhibition of GSK-3 may attenuate the response to lithium in vivo. In contrast, isolated knockdown of GSK-3α does not induce β-catenin. While much of the prior cancer literature has focused on the role of GSK-3β in human malignancy, these studies suggest a role for GSK-3α in AML differentiation and support a role for GSK-3α-directed targeted therapy. Disclosures: No relevant conflicts of interest to declare.