Abstract
Acute lymphoblastic leukemia (ALL) is the most prevalent pediatric malignancy, characterized by the uncontrollable proliferation of immature lymphocytes in the bone marrow (BM). Advances in treatment regimens have increased the overall survival of ALL patients significantly; however, approximately 20% of the cases exhibit resistance to treatment and relapse. The underlying molecular mechanisms of refractory or relapsed leukemia are yet elusive, but the protective role of the BM niche has widely been established. The microenvironment transforms into a favorable leukemic niche, where alteration of adhesion patterns, disruptions in the endosteal and vascular niche, protection of excessive reactive oxygen species and disrupted interactions between healthy hematopoietic and niche cells have been described. Thus, the studying of leukemia in the microenvironment context is necessary in order to recreate the protective cues that contribute to therapeutic resistance. To date, there is a lack of representative ex vivo systems to mimic the leukemic microenvironment, with traditional 2D co-culturing systems mostly used. 3D scaffold-based models have emerged as a promising tool for the study of ALL. These contain an extracellular matrix (ECM)-mimicking material that provides mechanical support and interaction sites for the leukemic cells, as well as multiple cellular BM compartments, thus imitating the niche more accurately. Here, we have developed a 3D hydrogel BM mimic that includes mesenchymal stromal cells (MSC) and endothelial cells (EC). This system recapitulates blood-vessel-like structures for leukemic patient-derived xenografts (PDX) and is characterized by robustness and compatibility with high-throughput readouts. We identified multi-lineage MSC differentiation and a vascular EC phenotype in a three-dimensional space, as well as leukemia-induced alterations such as ALL-MSC adhesive interactions through distinct cellular axes and EC-driven ECM degradation. We also demonstrated that leukemic PDXs can reproduce in vivo transcriptomic signatures through expression of essential developmental genes, that are absent when ALL cells are cultured alone. We further identified an epithelial mesenchymal transition (EMT)-like leukemic phenotype, described by enhanced invasiveness and drug resistance in other studies. This observation was validated by single-cell analyses that detected increased aggregation, migration and cell cycle diversity when cells are cultured in this artificial microenvironment compared to mono-cultured cells. Through these studies we have shown that this model can preserve cell heterogeneity and recapitulate relevant in vivo features. Another drawback of many 2D models is their limited ability to predict drug response accurately, as frequently resistance observed in vivo or in the clinical setting does not correlate to the in vitro response. Hence, 3D models have been proposed as alternative drug screening systems, as they recreate better the in vivo complexity. In this thesis, we utilized our system as a compound testing platform and detected recurrent response patterns, similar to what is observed in a well-established 2D drug response profiling platform, ergo validating the model’s relevance to predict response. Finally, 3D model response is thought to correlate better with in vivo due to the higher complexity, and indeed increased resistance has been reported in multiple ALL 3D studies. We also observed decreased efficacy in 3D for all compounds tested, through decreased cell death compared to 2D, and identified a remaining subpopulation upon treatment. Taken together, our 3D system recapitulates important in vivo elements both for leukemia and supporting cells and can be utilized for drug screening applications to potentially identify resistance patterns more readily compared to a 2D setting. $\textit{Full-text embargoed until: 2025-06-24}$
Kanari, Magdalini