Abstract

Hematopoiesis is defined as the set of processes leading to the production of mature blood cells from a restricted pool of HSC. In both humans and mice, the fate of HSC is finely regulated by intrinsic and extrinsic mechanisms, the latter relying mainly on the cells of the bone marrow microenvironment (BMME) in which HSC reside. In addition to regulating physiological hematopoiesis, BMME also plays a role in the onset and progression of myeloid hemopathies, such as myelodysplastic syndromes (MDS). Although they have elucidated a number of mechanisms concerning the pathophysiology of myeloid malignancies, xenotransplantation models do not take into account the human component of the BMME and have proved very disappointing in terms of engraftment, particularly when applied to MDS. To get as close as possible to what actually happens in humans, several teams have been working over the last ten years to develop organoid models that mimic a human BMME in mice, capable of harboring human hematopoiesis. These models are based on the use of mesenchymal stromal cells (MSC), derived from human bone marrow (BM), which are capable of differentiating to give rise the humanized BMME. Using the model initially developed by Andreas Reinisch's team in 2016, I set out to study the impact of these organoids, also named “ossicles” (hOSS), on the development of normal and pathological human hematopoiesis. The aim of this project was to further characterize this model and apply it to MDS, in order to better understand the role of the BMME in the pathophysiology of this disease.

First, I studied the impact of non-pathological hOSS on normal human hematopoiesis. Using a ScRNA-seq approach on the human CD34+ fraction, we showed that hOSS were capable of harboring normal human hematopoiesis, in the same way as murine BM. We have also shown that hOSS appeared to have an impact on the development of human hematopoiesis by promoting the development of myeloid subpopulations.

The next logical step in my work was to adapt this model to MDS. Using cellular and bulk RNAseq approaches, we first validated the pathogenicity of MSCs derived from MDS patients. We have identified a new molecular player, EPB41L3, usually described as having a role in intercellular communication, as potentially involved in the pathophysiology of the disease. We showed that MSCs from all MDS subtypes were capable of generating hOSS (MDS-hOSS), and that MSCs re-amplified from hOSS retained their molecular characteristics, underscoring the relevance of this model for reproducing a humanized pathological microenvironment. We then showed that MDS-hOSS were capable of robustly and reproducibly harboring normal human hematopoiesis. From a quantitative point of view, we observed no major difference between hematopoiesis grafted into MDS-hOSS and hematopoiesis grafted into hOSS from healthy elderly subjects.