A Dynamic Personalized Human 3D Organoid for the Study of the Tumor Microenvironment and Metabolism in Acute Myeloid Leukemia Using Patient-Derived Xenografts

A significant limitation to improving treatment for Acute Myeloid Leukemia (AML) is the lack of representative culture systems that (1) recapitulate the three-dimensional (3D) human AML bone marrow (BM) Tumor Microenvironment (TME) where AML cells are protected during treatment, and (2) account for metabolic shifts that lead to drug resistance and relapse. Current in vitro two-dimensional plate cultures and 3D systems lack TME heterogeneity and complexity resulting in oversimplification of disease biology and utilize either (1) AML cell lines, (2) artificially manufactured TME with allogeneic feeder layers, or (3) exogenously added cytokines and serum, with inherent clonal selection and bias. AML patient-derived xenograft (PDX) mouse models remain inadequate for the study of AML due to 50% engraftment failure and stromal damage incurred during conditioning. Herein, we have developed the first human personalized long-term dynamic cytokine- and serum-free in vitro BM AML ex vivo organoid using PDX-derived cells, distinguishing our system from existing technologies. BM was harvested from busulfan-conditioned NSG mice engrafted with an AML PDX when human CD33+ AML blood cells reached >30%; NSG controls were harvested concurrently. We evaluated 3D organoid cultures for: (1) optimal seeding density, (2) TME seeding conditions (AML or naïve) and (3) metabolic shifts. PDX cells were seeded onto 3D polyurethane scaffolds in serum- and cytokine-free medium at different cell densities: 2e6, 4e6 and 6e6 cells per scaffold (n=3) and recharged with autologous PDX cells on day (D)13-14 (Leukemic seeding + Leukemic recharge; L+L). L+L organoids were compared with non-recharged controls. Organoids seeded with 4e6 and 6e6 cells/scaffold proliferated from D34 and exhibited higher dynamic growth kinetics over time when compared with controls (p<0.05). In contrast, 2e6 cultures proliferated after D56 similar to the non-recharged control (p>0.05), suggesting inadequate organoid seeding density. When comparing cell sources, naïve NSG BM cells seeded at 4e6 cells/scaffold (n=3) and recharged with PDX cells (Naïve seeding + Leukemic recharge; N+L) displayed similar kinetics to L+L, suggesting that PDX cells could harness a naïve microenvironment to proliferate, as typified in vivo. Organoids successfully supported proliferation of CD33+CD44+ AML cells, which had similar phenotype to the input population by flow cytometry and morphology, even at D70. AML cells increased from 6% (D0) to >50% at D70 and a murine CD45+ population was observed throughout culture (11% in L+L; 4% in N+L at D70) indicating ongoing interactions between normal and AML cells as described in vivo. Using confocal microscopy, 3D niches spontaneously formed with cell-specific production of IL-6 and IL-8 (assessed by single molecule RNA fluorescence in situ hybridization; RNA FISH) creating a platform for the study of niche biology. Extracellular metabolites from AML-PDX organoid supernatants (n=3) were evaluated with a Bioprofiler and analyzed by principal component analysis with hierarchical clustering. Three culture stages were identified: 1) pre/early recharge phase characterized by low glucose consumption and low lactate production, (2) D20 to D41 phase exhibiting low glutamine uptake and, (3) from D44 onwards where the organoid exhibited leukemic metabolism characterized by high glucose and glutamine consumption with high lactate, ammonia and glutamate production. The last stage of growth was concurrent with the highest cell counts, suggesting that organoids proliferated more after switching to leukemic metabolism, as observed in vivo. Though N+L organoids exhibited similar metabolic stages, the second stage started later (D26), suggesting that a latent leukemic metabolic switch occurred due to early exposure to a naïve microenvironment. We have created a long-term 3D personalized human BM AML-PDX biomimicry in serum- and cytokine-free conditions that not only out-lives the murine PDX donor, but also is an ideal system to understand BM TME and AML niche biology through the evaluation of constituent cells, cytokines and metabolic needs and shifts in a dynamic ex vivo culture. The organoids support self-organizing AML niches and mirror leukemic metabolism while preserving parental cell phenotype, representing a physiologically-relevant heterogeneous and dynamic platform to test novel therapeutics.