Single-cell multiomic dissection of human hematopoietic stem cells reveals novel transcriptional fingerprints and regulatory programs in bone marrow and mobilized peripheral blood

Abstract

Abstract Hematopoietic stem and progenitor cells (HSPCs) are central to therapeutic strategies such as bone marrow transplantation and gene therapy/editing for a wide range of hematologic and genetic disorders. Their long-term clinical efficacy, however, hinges on successful engraftment and sustained hematopoietic reconstitution. However, due to the heterogeneity of HSPCs, accurate identification of long-term hematopoietic stem cells (LT-HSCs), possessing robust self-renewal and multilineage potential, remains essential for enhancing therapeutic outcomes. In this work, we aimed at defining the transcriptomic and epigenetic signatures that enable the robust identification of different human HSPC subpopulations, with a special emphasis on primitive HSCs. This molecular characterization would enable not only the improved gene targeting of specific cell populations to design more effective therapeutic strategies, but also the prediction of functional attributes and efficacy of engineered HSCs, thereby reducing the need for animal use in the evaluation of the safety and efficacy of HSC gene therapies. In these experiments we performed single-cell multiome profiling (scRNA-seq + scATAC-seq) on CD34⁺ bone marrow (BM) cells from five healthy donors, capturing both gene expression and chromatin accessibility at single-cell resolution. Given the importance of precise cell annotation, we implemented a two-step strategy: automated annotation using SingleR, followed by manual refinement based on well-established HSPC markers. Within BM CD34⁺ cells, we identified a transcriptionally defined HSC compartment. We applied clustering analysis alongside the StemPublic stemness score to resolve three distinct HSC subtypes: (i) quiescent LT-HSCs (qLT-HSCs), enriched for quiescence-related genes such as EMCN and TEK; (ii) activated LT-HSCs (aLT-HSCs), expressing MECOM and RUNX1; and (iii) short-term HSCs (ST-HSCs), marked by elevated CDK6 expression. Our results enabled the definition of a novel transcriptional fingerprint (StemFingerprint) that offers a novel and robust molecular tool for accurate identification of LT-HSCs and distinction from closely related progenitors. Given the widespread clinical use of mobilized peripheral blood (mPB) as an HSC source in gene therapy appraoches, we also performed single-cell multiome profiling of CD34⁺ cells from mPB of three healthy donors. Applying StemFingerprint, we successfully annotated LT-HSCs and ST-HSCs, but, consistent with prior reports, no qLT-HSCs were detected, likely due to their rapid exit from quiescence upon mobilization and BM egress. A comparative analysis between LT-HSCs and ST-HSCs across BM and mPB samples revealed a conserved expression of stemness-associated genes between the different HSC sources. Notably, we identified novel LT-HSC-associated genes, including FTH1 (iron homeostasis), RYR3 (calcium signaling), and SRGN (niche maintenance), potential regulators of human HSC function. Interestingly, our analysis also identified novel surface markers (undisclosed) which are currently being validated for the selective isolation of LT-HSCs and assessment of their engraftment potential, a critical feature for the development of improved gene therapy/editing strategies. To further dissect the regulatory landscape of HSC subtypes, we analyzed transcription factor (TF) motif accessibility and inferred gene regulatory networks. We observed that ETS1 showed the highest centrality in qLT-HSCs, consistent with its role in maintaining quiescence. In contrast, aLT-HSCs exhibited central roles for MYC and KLF6, consistent with their activation and proliferation signatures. Notably, MYC, a pivotal regulator of LT-HSC activation, was co-expressed with TFEB and TFR1, recapitulating the regulatory axis governing the quiescence-to-activation transition. Altogether, our data highlight differential dynamics of transcriptional and regulatory programs across HSC states. Overall, our study presents a comprehensive molecular map of human HSPCs in both BM and mPB, integrating transcriptomic and epigenetic features with functional potential. By defining key molecular and regulatory signatures, particularly those defining quiescent and active LT-HSC states, we provide tools for improved HSC identification, manipulation, and clinical application. These findings pave the way for more precise, effective, and ethically responsible strategies in stem cell and gene therapies.