A reference model for circulating hematopoietic stem cells with applications to diagnostics

With aging, deviation of human blood counts from their normal range accompanies the transition from health to hematological and other diseases. Hematopoietic stem and progenitor cells (HSPCs) deliver life-long multi-lineage output, but their variation across healthy humans and their diagnostic utility haven’t been characterized. To address this, we introduce an HSPC reference model using single-cell profiling of circulating CD34+ cells (cHSPCs) from 148 healthy individuals. We characterize physiological cHSPC composition, link it with standard blood counts and show that the ratio between lymphoid and myeloid progenitors is decreased in ageing males. Using this new reference model, we develop sensitive tools which could be used for diagnostics of myelodysplastic syndrome (MDS) from peripheral blood without bone marrow analysis. Profiling cHSPC compositions leads to the discovery of new MDS subclasses, ranging in their lymphocyte, basophil and granulocyte progenitor frequencies. The data and methodologies presented herein highlight the role of reference modeling in promoting clinical applications of single-cell genomics in MDS and in the future for other diseases too. This dataset can be explored as metacells at https://apps.tanaylab.com/MCV/blood_aging (for more about metacells, see https://github.com/tanaylab/metacells and https://doi.org/10.1186/s13059-022-02667-1). PBMCs were isolated from fresh blood using Ficoll (Lymphoprep filled Sepmate tubes by StemCell technologies), then enriched for cHSPCs using CD34 magnetic bead-based enrichment (EasySep human CD34 positive selection kit II by StemCell technologies). scRNA-seq was performed on the enriched fraction (10x Genomics Chromium Next Gem single cell 3’ reagent kit V3.1) *************************************************************** As our scRNA-seq data contain SNPs allowing donor identification, our IRB does not allow publishing raw data, and so we publish only processed data. ***************************************************************