Metformin induces ferroptosis associated with lipidomic remodeling in AML Open Access

• The FDA-approved drug metformin induces ferroptosis, with stronger impact on IDH1/2 and FLT3-mutated AML

• Impairing lipid detoxification via inhibition of DGAT1-mediated lipid droplet formation synergizes with metformin treatment

Metabolic reprogramming is a hallmark of cancer, essential for sustaining leukemogenesis. In acute myeloid leukemia (AML), high dependency on oxidative phosphorylation (OXPHOS) is often linked to poor outcomes and inhibiting this pathway has shown to be highly effective. However, most OXPHOS inhibitors are not clinically translatable due to significant side effects. Thus, repurposing safe FDA-approved drugs that can target OXPHOS is of great interest. Here, we evaluated metformin, an antidiabetic drug that inhibits OXPHOS, in a genetically diverse panel of primary AML samples to identify metabolic profiles predicting treatment susceptibility. Using label-free quantitative proteome analysis on sorted CD34+/CD117+ AML, we performed single-sample gene set enrichment analysis focused on metabolic terms and correlated enrichment scores with metformin sensitivity, followed by functional studies. Ex vivo treatment of AML samples with metformin showed a significant increase in ROS levels and ferroptosis induction, especially in samples with disturbed lipid metabolism, such as IDH2- and FLT3-mutant AMLs. In IDH2-mutant cells, co-treatment with palmitate, a saturated fatty acid (FA), increased metformin sensitivity, which could be rescued by CD36 knockdown, rendering these cells more resistant to treatment. Lipidomic analysis revealed profound alterations upon metformin treatment, including increased production of triglycerides and polyunsaturated FAs, further supporting a metabolic shift. We observed upregulation of genes related to lipid droplet formation, including DGAT1, a key enzyme in this process. DGAT1 inhibition was strongly synergistic with metformin, while iron chelators acted antagonistically. Our results underscore the potential of leveraging metabolic vulnerabilities in AML to identify more effective and personalized therapeutic strategies.