Generation of engraftable hematopoietic stem cells (HSCs) from pluripotent stem cells (PSCs) has long been thought an ultimate goal in the field of hematology. Numerous in vitro differentiation protocols, including trans-differentiation and forward programming approaches, have been reported but have so far failed to generate fully functional HSCs. We have previously demonstrated proof-of-concept for the in vivo generation of fully functional HSCs from induced PSCs (iPSCs) through teratoma formation (Suzuki et al., 2013). However, this method is time-consuming (taking over two months), HSCs are generated at low frequencies, and additionally require co-injection on OP9 stromal cells and SCF/TPO cytokines. Here, we present optimization of in vivo HSC generation via teratoma formation for faster, higher-efficiency HSC generation and without co-injection of stromal cells or cytokines.


First, we screened reported in vitro trans-differentiation and forward programming strategies for their ability to generate HSCs in vivo within the teratoma assay. We tested iPSCs transduced with the following dox-inducible TF overexpression vectors: (1) Gfi1b, cFOS and Gata2 (GFG), which induce hemogenic endothelial-like cells from fibroblast (Pereira et al.,2013); (2) Erg, HoxA9 and Rora (EAR), which induce short-term hematopoietic stem/progenitor cell (HSPC) formation during embryoid body differentiation (Doulatov et,al., 2013); and (3) Foxc1, which is highly expressed the CAR cells, a critical cell type for HSC maintenance (Oomatsu et al.,2014). We injected iPSCs into recipient mice, without co-injection of stromal cells or cytokines, and induced TF expression after teratoma formation by dox administration. After four weeks, GFG-derived teratomas contained large numbers of endothelial-like and epithelial-like cells, and importantly GFG-derived hematopoietic cells could also be detected. EAR-teratomas also generated hematopoietic cells, although at lower frequencies. By contrast, hematopoietic cells were not detected in control teratomas or Foxc1-teratomas. Through use of iPSCs generated from Runx1-EGFP mice (Ng et al. 2010), and CUBIC 3D imaging technology (Susaki et al. 2014), we were further able to demonstrate that GFG-derived hematopoietic cells were generated through a haemogenic endothelium precursor. Next, we assessed whether HSPC-deficient recipient mice would allow greater expansion of teratoma-derived HSCs. This was achieved by inducing c-kit deletion within the hematopoietic compartment of recipient mice (Kimura et al., 2011) and resulted in a ten-fold increase in the peripheral blood frequency of iPSC-derived hematopoietic cells. We further confirmed similar increases in iPSC-derived bone marrow cells, and in vivo HSC expansion, through bone marrow transplantation assays. Finally, we have been able to shorten the HSC generation time in this assay by five weeks through use of transplantable teratomas, rather than iPSCs.


We have demonstrated that GFG-iPSCs induce HSC generation within teratomas, via a hemogenic endothelium precursor, and that use of HSPC-deficient recipient mice further promotes expansion of teratoma-derived HSCs. These modifications now allow us to generate engraftable HSCs without co-injection of stromal cells or cytokines. Additionally, use of transplantable teratomas reduced HSC generation times as compared with the conventional assay. These findings suggest that our in vivo system provides a promising strategy to generate engraftable HSCs from iPSCs.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.