Transcriptional, Protein-Level and Functional Profiling of Human Fetal Liver-Derived Hematopoietic Stem Cells at Single Cell Resolution

Intro:

In the mouse, hematopoietic stem cells (HSCs) can be isolated and characterized at single cell resolution using a well-defined panel of markers. While it is possible to enrich for human HSCs using a panel of associated markers, similar resolution has not been attained. By profiling HSCs residing in the human fetal liver (FL) using a novel technique called CITE-Seq that combines single cell RNA sequencing (scRNAseq) and cell surface marker interrogation using oligo-tagged antibodies, we aimed to establish an accurate molecular signature of engraftable human HSCs shortly after they arise in development. As HSCs are defined functionally, we have coupled this transcriptomic and protein-level characterization with transplantation assays in immunocompromised NOD scid gamma (NSG) mice to connect expression profiles of cell subsets with functional engraftment.

Methods:

CITE-Seq was performed on human FL cells (week 19) that showed robust engraftment capability in NSG mice. CD34⁺ and CD34^- cells were magnetically separated and stained with a panel of 19 oligo-tagged antibodies that were deemed relevant to characterize HSCs, including classical HSC markers but also novel targets that were identified in a previous pilot scRNAseq experiment conducted on CD34⁺ FL cells. From the CD34⁺ fraction, we sorted live-gated cells (CD34⁺bulk) as well as a population of cells that was further enriched based on the expression of GPI-80, a marker tightly linked to engraftment potential (CD34⁺GPI-80⁺, ~3%). CD34^-GlycophorinA(GYPA)^- cells were also sorted to assay for the presence of CD34^- HSCs. These fractions were then loaded onto the 10x Genomics platform for capture of single cells and subsequent reverse transcription and amplification of both mRNAs and antibody-derived tags (ADTs).

Results:

Both mRNA and ADT libraries were successfully sequenced, yielding 29-43,000 reads/cell for the mRNA portion and >1,500 reads/cell for the ADT fraction. After quality control and filtering, this effort resulted in 8,775 CD34⁺bulk cells, 7,279 CD34⁺GPI-80⁺ cells, and 6,937 CD34^-GYPA^- cells available for further analysis. Simultaneous transplantation experiments of the fractions assayed by CITE-seq revealed superior engraftment potential of the CD34⁺GPI-80⁺ fraction, confirming enrichment for bona fide HSCs at the functional level. This was also reflected in the scRNAseq data where we found enrichment for known HSC markers such as VNN2 (GPI-80), PROM1 (CD133), PROCR (EPCR), THY1 (CD90), ITGA6 (CD49f), HMGA2, CLEC9A and HLF in the CD34⁺GPI-80⁺ fraction compared to CD34⁺bulk cells. As our pilot studies revealed considerable differences in transcriptional expression (via scRNAseq) as compared to protein-level expression (via cell surface marker expression), integration of the transcriptomic and cell surface marker expression data will further refine the signature of engraftable HSCs. Both layers of information at single cell resolution will allow for the identification of novel markers or unique combinations of markers that are directly correlated with engraftment potential.

Conclusion:

By isolating the GPI-80⁺ population within the CD34⁺ fraction in human FL, we have achieved unprecedented resolution of the signature of engraftable HSCs as confirmed by transplantation experiments. The in-depth characterization of this compartment as well as the surrounding CD34⁺ and CD34^- cells within the FL is expected to yield valuable insights with respect to several biological questions. This data can be directly harnessed in improving the purification and expansion of engraftable HSCs as well as in guiding the in vitro generation of HSCs from pluripotent stem cells.