Single-Cell Metabolic Profiling Integrated with Extracellular Vesicle Analysis Reveals Novel Metabolic Vulnerabilities and Prognostic Biomarkers in Acute Myeloid Leukemia

Introduction Acute myeloid leukemia (AML) is an aggressive disease for which less invasive tools are needed to explore disease dynamics. Changes in cell metabolism and metabolic adaptation are hallmark features for AML, thus emerging as promising therapeutic targets. However, it is still poorly elucidated whether and which extrinsic microenvironment signals may regulate the metabolic reprogramming occurring in AML. In this regard, Extracellular Vesicles (EVs) represent a promising tool for non-invasive liquid biopsy and may reveal a novel role in energy metabolism as well as clinically relevant biomarkers for AML. Along with EV profiling from peripheral blood (PB), we mapped the metabolic signature of leukemic stem cells (LSCs) and immune cell subsets to identify metabolic vulnerabilities and novel prognostic biomarkers. Methods PB (and paired bone marrow) were collected from AML patients at diagnosis (n=98) and healthy donors (HD, n=12). EVs were purified from platelet-free PB plasma of AML patients by size-exclusion chromatography followed by ultrafiltration. We have submitted all relevant data of our EV experiments to the EV-TRACK knowledgebase (EV-TRACK ID: EV230004). MACSPlex Exosome Kit was used for screening up to 37 surface EV markers. Western blot analysis was also performed. Energy cell metabolism was explored by Single Cell ENergetic metabolism by profilIng Translation inHibition (SCENITH TM) in LSC-like cells (CD34 +CD38 low/-)and immune cell subsets (CD3 +/CD4 +/CD8 +/T regulatory cells) using conventional/spectral flow-cytometry. Results Here, we developed a novel method exploiting both LSCs and EVs from AML patients using fresh whole blood for metabolic exploration. Firstly, we observed that LSC-like cells display low reactive oxygen species (ROS) levels with both high glutathione (GSH, as antioxidant molecule) levels and mitochondrial functionality suggesting a more resistant phenotype. Consistently, poor clinical outcomes were experienced by AML patients treated with conventional chemotherapy and with high GSH levels and mitochondrial potential in LSC-like cells (log-rank p=0.04). We also found high dependence on glucose and a preferential skewing toward glycolysis in LSC-like cells revealing a distinct metabolic status for LSCs in circulation. However, according to European LeukemiaNet (ELN) 2022 risk classification, a more prominent mitochondrial dependence on LSC-like cells was reported for intermediate- or adverse-risk AML patients revealing a prognostic relevance. By contrast, exploring the immunometabolic reprogramming in circulation, we revealed a significant decrease in glucose dependence for immune cells (CD4 + and CD8 + T cells) in comparison to LSCs proving the presence of a complex metabolic network in AML. Interestingly, we investigated the potential metabolic effects of EVs on LSCs. Indeed, CD34 + LSC-like cells co-cultured with circulating EVs enhanced their GSH levels and mitochondrial functionality towards a more aggressive phenotype. Accordingly, we discovered the presence of the antioxidant enzyme, glutathione peroxidases 3 (GPx-3), in AML EVs. Then, we profiled the surface proteins of isolated EVs to detect their cell origins and provide a reference map for clinical prognosis. We found that circulating AML EVs were significantly enriched in CD4 (helper T cell marker), CD14 (monocyte marker), CD40 (B-cell associated marker), CD44 and CD133-1 (stem cell markers), CD105 (stromal cell marker) and HLA-II (antigen-presenting cell marker) compared to HD EVs. In particular, we found that intermediate-risk AML patients showed a strong depletion in the expression of EV markers including T cell markers (CD2 and CD8) and mesenchymal markers (CD49e and CD146) in comparison to favorable and adverse-risk patients. Interestingly, AML patients who had been treated not intensively and showed low expression of CD40 (HR=0.23, 95% CI, 0.05−0,99, p=0.04) or CD62P (platelet activation marker, HR=0.24, 95% CI, 0.05−1.02, p=0.05) on EVs experienced poor outcomes suggesting that dedifferentiated phenotype on EVs might be associated with worse prognosis. Conclusion Overall, towards a new redefined ‘metabolic’ stratification for AML patients, we identified novel prognostic biomarkers for AML patients that might reveal novel metabolic vulnerabilities by exploiting in parallel single-cell metabolic profile and EV-based liquid biopsy.