Progressive divergence of definitive haematopoietic stem cells from the endothelial compartment does not depend on contact with the foetal liver

The yolk sac and the para-aortic splanchnopleura/aorta-genital ridges-mesonephros (P-Sp/AGM) region are the main sites of haematopoietic activity in the mouse embryo at the pre-liver stage of development. By day 11.5 of gestation, the AGM region is capable of autonomous initiation and expansion of definitive haematopoietic stem cells (HSCs). By day 12.5, HSC activity in the AGM region is reduced whilst a second wave of HSCs begins to emerge in the yolk sac. We show here that HSCs emerging in both locations are marked by co-expression of the endothelial-specific marker VE-cadherin and the pan-leukocyte antigen CD45. Phenotypic characterisation using CD31, TIE2, FLK1, Ac-LDL receptors, and CD34 markers demonstrated significant similarities between this VE-cadherin(+)CD45(+) 'double-positive' population and endothelial cells suggesting a common origin for these cells. The double-positive fraction also expressed the stem cell markers Kit, Sca1 and AA4.1. Long-term transplantation experiments demonstrated that the double-positive population, which constituted less than 0.05% of the day 11.5 AGM region and the day 12.5 yolk sac, is highly enriched for HSCs. In vitro assays showed that this population is also enriched for myeloid progenitors. During foetal liver colonization, circulating HSCs remained within the VE-cadherin(+) cell fraction, although their phenotypic similarity with endothelial cells became less prominent. Upon liver colonisation the majority of HSCs downregulated VE-cadherin, expression of which was completely lost in the adult bone marrow. Partial loss of VE-cadherin expression in HSCs can be observed extra hepatically in the advanced AGM region by E12.5. Similarly, the CD34(+)KIT(+) population in the placenta, recently identified as a reservoir of HSCs, partly lose VE-cadherin expression by E12.5. By culturing isolated E11.5 AGM region and E12.5 yolk sac we show that the developmental switch from a 'primary' VE-cadherin(+)CD45(+) to a more 'advanced' VE-cadherin(-)CD45(+) phenotype does not require contact of HSCs with the liver and is probably a function of developmental time.