Targeting cellular metabolism has emerged as a promising therapeutic strategy in acute myeloid leukemia (AML). Several inhibitors of key metabolic dependencies are currently being tested in clinical trials. The impact of inhibition of these metabolic enzymes on downstream signaling pathways, however, remains incompletely understood. We recently identified the mitochondrial creatine kinase, CKMT1, involved in arginine-creatine metabolism, as a new metabolic vulnerability in a subset of AML driven by the oncogene EVI1 (Fenouille et al., Nature Medicine, 2017). We determined that suppression of arginine-creatine metabolism by CKMT1-directed shRNAs, or by the small molecule cyclocreatine, selectively altered mitochondrial-dependent ATP production in EVI1 -positive cells, thereby reducing their growth.In an effort to identify downstream signaling pathways altered with CKMT1 inhibition, we interrogated RNA sequencing data from three EVI1 -positive cell lines, TF-1, UT-7, and UCSD-AML1, after 24 hours of cyclocreatine treatment. This analysis revealed that gene sets related to the GSK3 and WNT pathways were among those most enriched in genes whose expression was suppressed by cyclocreatine. We then established that creatine kinase (CK) pathway inhibition impaired the expression of GSK3 and WNT target genes, including genes upstream in the pathway, such as DVL1, DVL2, DVL3, FZD3, FZD6, FZD7, and FZD8, and more downstream targets, such as CTNNB1, MYC, PPARD, CCND1, and JUN. In line with these transcriptional effects, the level of GSK3 phosphorylation at the inhibitory serine sites of the two GSK3 isoforms (S21 on GSK3A and S9 on GSK3B), was strongly enhanced upon cyclocreatine treatment, confirming the inhibition of GSK3. Intracellular glycogen accumulation, a direct phenotypic consequence of GSK3 inhibition, was also observed in cyclocreatine-treated AML cells.To investigate whether inhibition of GSK3 signaling was important for the cyclocreatine-mediated differentiation and growth effects, we generated constitutively active GSK3 mutants by substituting serines 21 and 9 with non-phosphorylable alanines (S21A and S9A) on GSK3A and GSK3B, respectively. Co-expression of active mutants GSK3A/B S21A/S9A completely abrogated cyclocreatine-induced CD117 downregulation, a marker of stemness. We then compared the effects of CK pathway inhibition in two mouse models: NrasG12D + Evi1 -driven AML, containing a clonal Evi1 integration and expressing high transcript levels of Evi1, and NrasG12D -driven AML that does not express Evi1. We previously established that the NrasG12D + Evi1 -driven disease burden was selectively altered by CK pathway inhibition. In this context, we have now demonstrated that overexpression of Gsk3b S9A in NrasG12D + Evi1 cells attenuated the anti-leukemic effect of cyclocreatine leading to altered survival.Inhibition of both the alpha and beta isoforms of GSK3 is well reported to induce β-catenin stabilization, an undesirable effect in AML therapy. Surprisingly, however, in addition to its inhibitory effect on GSK3, cyclocreatine treatment also repressed mRNA and protein levels of CTNNB1 . This effect was rescued by the reactivation of arginine-creatine metabolism upon phospho-creatine supplementation. Despite GSK3 pathway inhibition, cyclocreatine treatment prevented nuclear CTNNB1 accumulation in EVI1-positive cells and thereby did not activate the CTNNB1 transcriptional program.Taken together, our results demonstrate that the blockade of the creatine kinase pathway inhibits GSK3 and silences the WNT pathway, impeding canonical control of GSK3 on WNT signaling. We thus identified a therapeutic strategy to promote the anti-leukemic and pro-maturating effects of GSK3 inhibition without inducing potentially pro-leukemogenic β-catenin signaling. DisclosuresHermine:Novartis: Research Funding; AB Science: Equity Ownership, Honoraria, Patents & Royalties, Research Funding; Celgene: Research Funding; Hybrigenics: Research Funding; INatherys: Equity Ownership, Research Funding. Stegmaier:Novartis: Consultancy, Research Funding.
Read full abstract