Fructose Treatment Targets Myeloid Leukemogenesis in IDH2 Mutants By Exhausting Alpha-Ketoglutarate Pools

Abstract

Acute Myeloid Leukemia (AML) is a heterogeneous group of hematological malignancies and represents the most frequent cause of leukemia-related deaths. The uncontrolled growth of myeloblasts in the Bone Marrow (BM) leads to accelerated metabolic processes and high consumption of nutrients, especially glucose. Isocitrate dehydrogenase (IDH) 1 and 2 are frequently mutated in human cancer, affecting up to 15% of AML patients. The neo-morphic activity conferred by IDH 1 or 2 mutations (mIDH1/2) converts α-ketoglutarate (α-KG) to the R-enantiomer of 2-hydroxyglutarate (2-HG), which competitively inhibits α-KG-dependent dioxygenases (Figueroa et al., 2010, Su et al.,2018, Li et al.,2017) and contributes to a block in myeloid differentiation and leukemic transformation. mIDH1/2 are clinically validated therapeutic targets as specific inhibitors have been approved by the FDA for the treatment of AML(DiNardo et al., 2018; Rohle et al., 2013; Stein et al., 2019; Wang et al., 2013) However, only 40% or less of mIDH AML patients respond to IDH inhibitor therapy (DiNardo et al., 2018; Stein et al., 2017) and little is known of how IDH mutations can be exploited for new metabolic vulnerabilities. We recently reported a novel metabolic dependency in AML cells that express low levels of the fructose transporter GLUT5. In fructose conditions these AMLs rewired their metabolism away from glycolysis and toward the serine synthesis pathway (SSP), (Jeong and Savino, Cell Metabolism 2021). Inhibiting the SSP by blocking PHGDH under these metabolic conditions resulted in reduced tumor burden. Additionally, patients with FLT-ITD alterations were sensitive to SSP blockade (Bjelosevic et. al. 2021). Although we previously found that myeloid leukemia cells grow normally in fructose, AML cell lines engineered to express IDH2 R140Q (TF1 knock in, and transduced MOLM13, U937 and OCIAML2) had a 50% reduction in cellular growth and 1.6-fold increase in differentiation (CD15 marker). Remarkably, a similar effect was observed in mIDH1 or 2 chondrosarcoma cell lines, suggesting that this metabolic program is more generally applicable. Excitingly, this new fructose induced therapeutic vulnerability could be recapitulated in vivo. We found that the fructose treated mice could delay myeloid leukemogenesis by 9-10 days (MOLM13 IDH2 R140Q) or in a primary patient sample successfully engrafted with IDH2 R140Q (n=6 mice/group p=0.0012; 3 weeks, I.P. 4 g/kg). To understand the basic mechanism for this new metabolic vulnerability, we first focused on the flux though the SSP. We found an increase in the end product of the SSP, glycine, as measured by LC-MS in the fructose condition compared to glucose. This SSP addiction under fructose was further demonstrated by reduced growth upon PHGDH inhibition (mIDH2: 75% reduction vs 50% WT IDH2 p=.0004; NCT503 5uM). Utilization of the SSP results in the generation of cytosolic α-KG, the substrate of IDH mutant enzymes. Thus, we hypothesized that mIDH AML cells might display enhanced dependency due to the requirement of α-KG for SSP, mIDH2 activity and to sustain the otherwise defective TCA cycle. Consistent with α-KG's central role in fructose driven metabolic addiction, we rescued the cell growth reduction by treating with cell permeable a-KG. Furthermore, in fructose conditions, we observed significant accumulation of the TCA cycle intermediate malate as well as aspartate, suggesting an engagement of the malate-aspartate shuttle in maintaining the redox balance. Previously, we found that AML cell lines (i.e THP1 cells) with moderate to high abundance of surface GLUT5, were not dependent on SSP in fructose. Surprisingly, by introducing IDH2 R140Q or by treating the cells with exogenous D-2HG, we found a 2-fold reduction of surface GLUT5 protein. This reduction of GLUT5 rewired the metabolism toward the SSP-addiction and resulted in a 50% reduction in cell growth in fructose. Interestingly, GLUT5 was controlled through transcriptional regulation of the locus as early as 8hrs post D-2HG treatment. Importantly, this relationship was found to be associated in patients, where PHGDH expression was significantly increased in mIDH2 AML patients with low vs high GLUT5 (n=15, n=16 p<.038) compared to other mutational subtypes (n=155, n=155 ns). In summary, we find a surprising new metabolic dependency based on sugar conditions in mIDH2 AML that could pave the way for a new therapeutic strategy.