TRANSCRIPTOME BASED PROJECTION OF SINGLE CELLS TO UNCOVER DEVELOPMENT AND HETEROGENEITY OF ABNORMAL HEMATOPOIETIC CELLS

Abstract

Recent advances in single-cell RNA-sequencing (ScRNA-seq) has accelerated the investigation of hematopoiesis. However, characterization of leukemic cells using scRNA-seq remains challenging as conventional approaches for analysis fail to consider that the transcriptomes of leukemic cells often systematically diverge from normal cells. This results in irrelevant separation of leukemic subpopulations from their healthy counterparts. We have developed a computational approach - Nabo - which compares otherwise unalignable cells. The accuracy of Nabo was validated by mapping multiple scRNA-Seq datasets from purified subpopulations to well-defined reference samples. Nabo predicted the identity of human hematopoietic stem and progenitor populations to the same accuracy as can be achieved by established cell-surface markers. We applied Nabo to uncover the heterogeneity of hematopoietic cells from different stages of acute myeloid leukemia (AML). This revealed that AML cells lacked the heterogeneity of normal CD34+ cells and were devoid of cells with a hematopoietic stem cell gene signature. Furthermore, leukemic cells mapped to distinct stages of myeloid progenitors in a patient-dependent manner. To ask whether the observed mapping of leukemic to healthy progenitor cells indicates the identity of leukemia-initiating cells (LICs), we induced the human fusion oncogene MLL-ENL in murine granulocyte-monocyte-lymphoid progenitors (GMLPs). Induced leukemic cells mapped to a distinct Flt3+ subpopulation within the healthy GMLPs. Functional assays demonstrated that Flt3+ GMLPs contained 3-4 fold more LICs than Flt3- GMLPs, indicating that LICs within GMLPs express Flt3. In summary, Nabo represents a novel tool for delineation of leukemia heterogeneity, with the potential to identify LIC populations in single-cell data.