Detecting lineage-specific regulatory activity and stemness retention in single-cell multiomics of human hematopoiesis

Abstract Introduction: The prevalence of epigenetic regulator mutations in clonal hematopoiesis and hematologic malignancies highlights the necessity of understanding how mutations in these regulators affect the fate potential of hematopoietic stem and progenitor cells (HSPCs) at a fundamental level. Utilizing single-cell multi-omics data, including paired transcriptome and chromatin accessibility (10X Multiome) and paired histone post-translational modification profiles and RNA polymerase II profiling (sciCUT&Tag + sciCUTAC) of human bone marrow mononuclear cells from healthy donors, we have identified regulatory dynamics of cell fate gain and loss unique to single lineages and quantify the loss of the stem cell program over a given fate transition.These tools will unravel the complex regulatory landscape of priming and lineage biases in healthy hematopoiesis and provide a pathway for investigating the specific mechanisms by which clinically observed epigenetic regulator mutations disrupt HSPC fate decisions in hematologic diseases. Methods: To construct a combined low-dimensional representation of chromatin accessibility scATAC-seq and sciCUTAC data were processed using shared genomic features using Seurat (Hao et al. 2021) and Signac (Stuart et al. 2021), then integrated using Seurat IntegrateData(). A shared pseudotime axis could then be computed using Palantir (Setty et al. 2019). To quantify epigenetic mark enrichment and chromatin accessibility at the gene level, gene scores were computed using ArchR (Granja et al. 2021). Using these scores, regulatory activity trends are fit for individual genes using Gaussian process regression (Otto et al. 2024). Results: Co-occurring H3K27me3 depletion, transcription, chromatin accessibility, and H3K27ac enrichment are restricted to the B cell lineage for PAX5 and EBF1, both critical master regulators of B lymphopoiesis. Comparable patterns of repression and de-repression at key transcription factor loci are detectable in the myeloid and erythroid lineages. The pseudotime ordering constructed from combined scATAC-seq and sciCUTAC data allows elucidation of the relative ordering of changes in histone mark enrichment, chromatin accessibility, and gene expression throughout hematopoiesis. Conclusions: Preliminary results from this work demonstrate the potential of co-analyzing more than two single-cell data modalities simultaneously, without the need for the development of novel techniques that can measure three or more modalities paired in single cells. The observed epigenetic activity at key master regulators, including PAX5 and EBF1 in B lymphopoiesis, defines a signature of lineage-specific gene activation during hematopoiesis.