Development of a human in vitro model of bone marrow vascular niche on a chip to study CD34+ hematopoietic stem/progenitor cells function in diabetes

Abstract

Abstract Background/Introduction Protective bone marrow (BM) CD34+ hematopoietic stem/progenitor cells (HSPCs) are released in the bloodstream after cardiomacrovascular (CV) tissue damage such as myocardial infarction1. Diabetes, an high risk condition to develop CV diseases, results in BM damage2-4 and detrimental pro-apoptotic and pro-oxidant shifts in BM-derived CD34+HSPCs5,6. Being able to classify and study the mechanisms of CD34+HSPC-vascular cell interaction would be a great step forward in the CV research field. However, nowadays there is no available primary human cell-based in vitro model to study the BM vascular niche. Purpose We aim to develop a human-based in vitro model to reproduce the BM vascular niche that allows the analysis of molecular mechanisms involved in CD34+HSPC function. This model will be relevant to identify molecular targets beyond hypoglycemic treatment and potentiate personalized medicine applications for diabetic complications. Method Human endothelial cells and pericytes were isolated from femoral head of patient who underwent hip replacement or from ad hoc BM aspirate by magnetic bead separation and characterized by immunofluorescence staining and flow cytometry. A culture of endothelial and pericytic cells to replicate the BM vascular niche was developed on a chip. CD34+HSPC subpopulation alteration in diabetes was preliminarily evaluated through single cell sequencing analysis (scRNAseq) on CD34+HSPC cells isolated from BM of patients with or without diabetes (N=1). Results We successfully set up a new method to efficiently isolate autologous BM endothelial cells and pericytes derived from patients (N=3). The cells expressing typically endothelial (CD31, ZO-1, and VE-cadherin) and pericytic (CD146, CD140b) markers were cultured in a 2D chip model. ScRNAseq reveals 15 clusters of CD34+HSPCs and important differences between the non-diabetic and diabetic samples. In particular, cluster 2 is highly represented in the non-diabetic and completely disappears in the diabetic sample, while on the contrary cluster 9 is over-represented in the diabetic compared to the non-diabetic sample. Interestingly, though the trajectory analysis cluster 2 shows the highest probability of turning into cluster 9. Through Ingenuity Canonical Pathways analysis we identified the anti-oxidant pathway as the most affected in diabetes-derived CD34+HSPCs. Conclusions In vitro human vascular niche model could be the gold standard to study BM cell crosstalk. In our interest, cluster 2 completely disappears in diabetes-derived CD34+HSPCs, potentially representing the CV-protective CD34+HSPC fractions more prone to be mobilized. Developing of our in vitro model would be useful to validate this hypothesis.Figure 1:scRNAseq cluster UMAPFigure 2:Endothelial cells IF staining