A Role for Lipid Metabolism in Tyrosine Kinase Inhibitor (TKI) Resistance of Chronic Myeloid Leukemia (CML)

Tyrosine kinase inhibitors (TKIs) targeting BCR-ABL1 are remarkably effective therapies in chronic myeloid leukemia (CML). Despite success, TKIs do not target the CML leukemic stem cell (LSC), meaning patients must be treated for life at a high economic burden and often with significant side effects. Our previous work demonstrated a tumor suppressor role for G0/G1 switch gene 2 (G0S2) in CML, which is profoundly downregulated in TKI resistance (>3-fold, p<0.02) (Gonzalez MA, et al. Cancer Res 2020 #648). Low G0S2 expression not only correlated with TKI resistance, but also with a shorter overall survival in chronic phase (p=0.036), and transformation from the chronic to the blast phase of disease (p<0.05). G0S2 is known to regulate apoptosis, quiescence, lipid metabolism, oxidative phosphorylation, and also acts as an inhibitor of adipose triglyceride lipase (ATGL), the rate-limiting enzyme in intracellular lipolysis. We hypothesized that G0S2-mediated ATGL inhibition underlies its tumor suppressor activity in CML and TKI resistance.

To address this hypothesis, we first assessed ATGL protein expression in CML cell lines (K562 and KU812) in the presence and absence of the BCR-ABL TKI, imatinib. Importantly, ATGL protein expression remained unchanged in graded doses of imatinib, suggesting that ATGL expression occurs in a BCR-ABL1-independent manner. To assess the functional role of ATGL in CML and TKI resistance, we performed shRNA-mediated ATGL knockdown (shATGL) in both cell lines. Although ectopic G0S2 impaired survival of CML cell lines and patient samples, we observed a different phenotype upon ATGL knockdown. Rather, shATGL alone increased colony formation by ~30% in both cell lines in the absence but not presence of imatinib (K562, p=0.04; KU812, p=0.0024). Knockdown of ATGL had no effect on colony formation of normal cord blood CD34 ⁺ progenitors (p=0.742). Interestingly, when we ectopically expressed G0S2 into K562 cells with simultaneous ATGL knockdown, the phenotype in colony formation assays mimicked ectopic G0S2 expression, reducing survival by ~50% (p=0.05). These data suggest that the tumor suppressor role of G0S2 in CML is independent of ATGL. However, while ectopic G0S2 expression alone had no effect on apoptosis of CML cell lines, when combined with ATGL knockdown, imatinib-mediated apoptosis was markedly increased. This was similar to data observed in blast phase CML patient samples upon ectopic G0S2 expression. Consistently, RNA sequencing (RNAseq) data for CML patients revealed that ATGL mRNA is highly downregulated in blast phase (n=14) compared with patients in the chronic (n=52) or accelerated (n=11) phases of the disease (p<0.0001). These data suggest that ATGL abrogates G0S2-mediated apoptosis in the presence of imatinib. We next performed colony formation assays of K562 cells with G0S2 knockdown (shG0S2) versus a non-targeting control (shNT) in the absence and presence of an ATGL inhibitor (Atglistatin, 50 µM) and a fatty acid oxidation (FAO) inhibitor (TMZ, 10 µM). Atglistatin had no effect on colony formation in the shNT-expressing cells, but reduced survival by ~40% upon G0S2 knockdown. Similarly, TMZ reduced survival of shNT cells by ~30%, while it reduced survival by ~90% upon G0S2 knockdown. These data suggest that the phenotypic effects of G0S2 knockdown in TKI resistance in part requires FAO.

The effect of G0S2 on lipid metabolism was further confirmed with RNAseq and metabolomics/lipidomics analyses. RNAseq revealed no overlapping pathways between K562 cells expressing ectopic G0S2 versus shATGL. Lipidomics analyses revealed that G0S2 knockdown reduced expression of tri- and di-glycerides, whereas ectopic G0S2 promoted triglyceride accumulation in K562 cells. G0S2 knockdown also resulted in substantial changes of phosphatidylethanolamine and phosphatidylcholine expression, implicating G0S2 in the production of lipid bilayer components. Finally, metabolomics data implicated a role for G0S2 as a negative regulator of the mitochondrial electron transport chain. Altogether, these findings suggest that G0S2 and lipid metabolism play a role in regulating leukemic stem and progenitor cell survival as well as TKI response in vitro. Therefore, restoring G0S2 expression and inhibiting FAO, combined with BCR-ABL1 inhibition, may be a novel clinical strategy to induce treatment-free remission in CML patients and eradicate the CML LSC.