In acute myeloid leukemia (AML), the chances of achieving disease-free survival are low. Studies have demonstrated a supportive role of endothelial cells (ECs) in normal hematopoiesis. Here we show that similar intercellular relationships exist in leukemia. We demonstrate that leukemia cells themselves initiate these interactions by directly modulating the behavior of resting ECs through the induction of EC activation. In this inflammatory state, activated ECs induce the adhesion of a sub-set of leukemia cells through the cell adhesion molecule E-selectin. These adherent leukemia cells are sequestered in a quiescent state and are unaffected by chemotherapy. The ability of adherent cells to later detach and again become proliferative following exposure to chemotherapy suggests a role of this process in relapse. Interestingly, differing leukemia subtypes modulate this process to varying degrees, which may explain the varied response of AML patients to chemotherapy and relapse rates. Finally, because leukemia cells themselves induce EC activation, we postulate a positive-feedback loop in leukemia that exists to support the growth and relapse of the disease. Together, the data defines a new mechanism describing how ECs and leukemia cells interact during leukemogenesis, which could be used to develop novel treatments for those with AML.

Endothelial cell (EC) activation and their subsequent binding with different cells have various mechanical consequences that, if monitored real time, can serve as a functional biomarker of many pathophysiological response mechanisms. This work presents an innovative and facile strategy to conduct such monitoring using quartz crystal microbalance (QCM), thereby relating the shifts in its frequency and motional resistance to morphological changes upon cell-cell and cell-substrate interactions. By activating ECs with TNF-α and then characterizing their binding with HL-60 and KG-1 leukemia cells, we are able to induce the mechanical changes in ECs especially in the region of cell-substrate contact which resulted in dynamically coupled mass and viscoelastic changes representing the extent of both activation and binding. The activated ECs suffered a decrease of cellular contact area, leading to positive frequency shift and decreased motional resistance. The binding of leukemia cells onto pre-activated ECs exerted a mechanical force to regain the cell surface contact which resulted in the obvious QCM responses opposite to that of activation, and proportional to the number of cells added, in spite of the fact that these added cells are extremely outside the extinction depth of the shear wave generated by QCM. Different cell lines demonstrate different attachment behavior, which was detected by the QCM. Despite these variations are quite subtle, yet the sensitivity of the technique for dynamic changes at the interface makes them detectable. Moreover, the reproducibility of the generated data determined at each step by deviation measurements (<10%) in response plot was very high despite the high possible heterogeneity in cell populations. The results are explained on the basis of simple theoretical and physical models, although, the development of a more quantitative and precise model is underway in our laboratory.