Spatial transcriptomics identified dynamic and spatial-resolved niche cell and signal architecture for hematopoietic stem cell specification Free

Hematopoietic stem cell (HSC) transplantation remains limited by insufficient matched donors. Generating HSCs from pluripotent stem cells (PSCs) represents an attractive strategy, potentially providing unlimited autologous HSCs for therapeutic applications. However, the generated cells overall lack self-renewal capacity, indicating their inappropriate directed differentiation or immature cell state. This required deep insights into embryonic HSC specification.

During embryogenesis, HSCs first arise within the dorsal aorta of the aorta-gonad-mesonephros (AGM) region, progressing through sequential stages of hemogenic endothelial cell specification and endothelial-to-hematopoietic transition. These events occur mainly on the ventral side of the dorsal aorta within a defined developmental window (mouse embryonic days (E)9.5–E12.5). Despite its importance and recent advances, the precise cellular composition and signaling landscape within this niche remain poorly defined.

To address this, we combined single-cell RNA sequencing with spatial transcriptomics (Curio-seq) on trunk sections encompassing the AGM at E10.5, E11.5 and E12.5. Our analysis revealed distinct cellular distributions between the dorsal and ventral sides of the aorta. The dorsal side is enriched with Nfgr⁺ cells expressing BMP signaling components and a wt1⁺ mesonephric cell population. In contrast, the ventral aortic region exhibits complex cellular diversity and a variety of signaling pathways (including WNT, NOTCH, and retinoic acid). Notably, we identified a distinctive population of mesenchymal stromal cells co-expressing cdh2 (encode N-cadherin) and pdgfrα, forming a discrete layer between aortic endothelial cells and mesonephric cells, and directly contacting all identified pre-HSCs.

Using CellChat, we identified dynamic signaling crosstalk between niche cells and pre-HSCs via ligand-receptor interactions. For example, interactions between N-cad⁺ MSCs and pre-HSCs, such as Dlk1-Notch1, Tgfβ1-Tgfβr1 and Jag1-Notch1, peaked at E10.5, E11.5 and E12.5, respectively. Furthermore, we mapped these interactions directly on the AGM tissue, enabling spatial-resolved visualization. Curio-seq data thus provided a 10 µm depiction of the dynamic and spatially asymmetric cellular and signaling landscape within the AGM niche.

We next validated our Curio-seq findings using MERFISH, an imaging-based spatial transcriptomics approach. A customized RNA probe panel targeting 140 key genes (including cellular markers and signaling molecules) confirmed niche cell identities and signaling interactions at single-cell resolution on E11.5 AGM sections. Consistent with Curio-seq results, MERFISH verified the ventral enrichment of N-cad⁺ MSCs, their close physical proximity to pre-HSCs, and their role as principal sources of Dlk1, Cd200 and Sdf-1 signals around the aorta.

Given the unique localization of N-cad⁺ MSCs and their intricate interactions with pre-HSCs, we next dissociated and reaggregated E11.5 AGM cells, with or without sorting out Cd45^-Cd144^- N-cad⁺ MSCs, to culture ex vivo and mature pre-HSCs into transplantable HSCs. Depletion of N-cad+ MSCs substantially reduced phenotypic (by flow cytometry) and functional HSCs (by transplantation assays). To define the signals from N-cad+ MSCs involved in HSC specification, we supplemented reaggregated cultures lacking N-cad⁺ MSCs with recombinant mouse Jag1, Cd200, or Sdf-1. Supplementation with either Jag1 or Cd200, but not Sdf-1, significantly rescued HSC specification impaired by N-cad+ MSC depletion. To directly examine the effects of N-cad⁺ MSC-derived Jag1 and Cd200, we generated Ncad^CreER; Jag1^flox/flox and Ncad^CreER; CD200^flox/flox mice and conditionally deleted Jag1 or Cd200 from N-cadherin-expressing cells at E9.5. AGM cells at E11.5 were collected for transplantation assays, and the results are pending.

In summary, we utilized spatial transcriptomics to define dynamic, single-cell and spatially resolved profiles of niche cells and their emanated signals in the AGM region. We demonstrated a positive role for N-cad+ MSCs and their putatively derived Jag1 and Cd200 in HSC development. This approach validates our spatial transcriptomics results and provides a paradigm for future validation of additional signals of interest identified by spatial transcriptomics. Our findings enhance the understanding of the embryonic hematopoietic niche and provide new insights toward developing improved methods for PSC-to-HSC induction.