The adult hematopoietic system of mammals is a dynamic hierarchy of cells with the hematopoietic stem cell at its foundation. During embryonic development, the source and expansion potential of this cell remain unclear. Two sites of hematopoietic activity, the yolk sac and aorta-gonad-mesonephros (AGM) region, function in mouse ontogeny at the pre-liver stage of hematopoiesis. However, cellular interchange between these tissues obscures the embryonic site of hematopoietic stem cell generation. Here we present the results of a novel in vitro organ culture system demonstrating that, at day 10 in gestation, hematopoietic stem cells initiate autonomously and exclusively within the AGM region. Furthermore, we provide evidence for the in vitro expansion of hematopoietic stem cells within the AGM region. These results strongly suggest that the AGM region is the source of the definitive adult hematopoietic system, which subsequently colonizes the liver.

Emergence of hemopoietic stem cells in the mammalian embryo has yet to be definitively allocated. Previously, we detected multipotent hemopoietic precursors in the region surrounding the dorsal aorta (paraaortic splanchnopleura) beginning at 8.5 days postcoitum (dpc). However, as circulation is already established, it remained unclear whether hemopoietic precursors arise in situ or are blood-delivered. By adding an organotypic step to our former culture system, we now detect lymphocyte and multipotent myeloid precursors from the intraembryonic splanchnopleura as early as 7.5 dpc. Under identical conditions, yolk sacs from the same embryos are unable to generate lymphoid progeny and have a reduced potential for myeloid differentiation and maintenance. Thus, if isolated before circulation, the yolk sac does not produce multipotent precursors and therefore does not contribute to definitive hemopoiesis in the mouse.

The extent to which primitive embryonic blood progenitors contribute to definitive lymphoid-myeloid hematopoiesis in the adult remains uncertain. In an effort to characterize factors that distinguish the definitive adult hematopoietic stem cell (HSC) and primitive progenitors derived from yolk sac or embryonic stem (ES) cells, we examined the effect of ectopic expression of HoxB4, a homeotic selector gene implicated in self-renewal of definitive HSCs. Expression of HoxB4 in primitive progenitors combined with culture on hematopoietic stroma induces a switch to the definitive HSC phenotype. These progenitors engraft lethally irradiated adults and contribute to long-term, multilineage hematopoiesis in primary and secondary recipients. Our results suggest that primitive HSCs are poised to become definitive HSCs and that this transition can be promoted by HoxB4 expression. This strategy for blood engraftment enables modeling of hematopoietic transplantation from ES cells.

It is widely accepted that during murine embryogenesis, totipotent haematopoietic stem cells first originate in the yolk sac, then migrate to the fetal liver and finally colonize the bone marrow shortly before birth. This view is based on in vitro studies showing that yolk sac cells can differentiate into various haematopoietic lineages and in vivo studies showing that yolk sac contains spleen colony-forming units (CFU-S) beginning at day 8 of gestation. However, some investigators have failed to find statistically significant numbers of CFU-S arising from day 9 yolk sac and, although one group reported that yolk sac could repopulate the haematopoietic system of W mutant mice, others have failed to confirm yolk sac-derived repopulation of adults. In the avian and amphibian systems, the yolk sac gives rise only to early, transitory haematopoiesis whereas the definite adult haematopoietic stem cells in these vertebrates are derived from the mesodermal region containing the dorsal aorta. Because this analogous area of the mouse embryo has not been previously examined for haematopoietic activity, we directly compared the CFU-S activity of the aorta, gonad, mesonephros (AGM) region with the yolk sac and fetal liver during embryogenesis. Here we report that this intra-embryonic AGM region contains CFU-S activity at a higher frequency than that in embryonic yolk sac and that such activity appears in the AGM region before the fetal liver.

We previously demonstrated the essential role of the flt-1 gene in regulating the development of the cardiovascular system. While the inactivation of the flt-1 gene leads to a very severe disorganization of the vascular system, the primary defect at the cellular level was unknown. Here we report a surprising finding that it is an increase in the number of endothelial progenitors that leads to the vascular disorganization in flt-1(-/-) mice. At the early primitive streak stage (prior to the formation of blood islands), hemangioblasts are formed much more abundantly in flt-1(-/-) embryos. This increase is primarily due to an alteration in cell fate determination among mesenchymal cells, rather than to increased proliferation, migration or reduced apoptosis of flt-1(-/-) hemangioblasts. We further show that the increased population density of hemangioblasts is responsible for the observed vascular disorganization, based on the following observations:(1) both flt-1(-/-) and flt-1(+/+) endothelial cells formed normal vascular channels in chimaeric embryos;(2) wild-type endothelial cells formed abnormal vascular channels when their population density was significantly increased; and (3) in the absence of wild-type endothelial cells, flt-1(-/-) endothelial cells alone could form normal vascular channels when sufficiently diluted in a developing embryo. These results define the primary defect in flt-1(-/-) embryos at the cellular level and demonstrate the importance of population density of progenitor cells in pattern formation.

Mouse kif5B gene was disrupted by homologous recombination. kif5B-/- mice were embryonic lethal with a severe growth retardation at 9.5-11.5 days postcoitum. To analyze the significance of this conventional kinesin heavy chain in organelle transport, we studied the distribution of major organelles in the extraembryonic cells. The null mutant cells impaired lysosomal dispersion, while brefeldin A could normally induce the breakdown of their Golgi apparatus. More prominently, their mitochondria abnormally clustered in the perinuclear region. This mitochondrial phenotype was reversed by an exogenous expression of KIF5B, and a subcellular fractionation revealed that KIF5B is associated with mitochondria. These data collectively indicate that kinesin is essential for mitochondrial and lysosomal dispersion rather than for the Golgi-to-ER traffic in these cells.

The site of origin of lymphohematopoietic stem cells (HSC) that initiate definitive blood cell production in the murine fetal liver is controversial. Contrary to reports that the preliver yolk sac does not contain definitive HSC, we observed that CD34+ day 9 yolk sac cells repopulated multiple blood cell lineages in newborn hosts for at least 1 year. Furthermore, 100 CD34+c-Kit+ day 9 yolk sac or para-aortic splanchnopleura (P-Sp) cells, known to give rise to embryonic HSC, similarly repopulated hematopoiesis in recipient hosts. Surprisingly, 37-fold more CD34+c-Kit+ cells reside in the day 9 yolk sac than in the P-Sp. In sum, definitive HSC are coexistent, but not equal in number, in the murine yolk sac and P-Sp prior to fetal liver colonization.

We show by an in vitro approach that multipotent hemopoietic cells can be detected in the body of the mouse embryo between the stages of 10-25 somites (8.5-9.5 days of gestation)--i.e., prior to liver colonization (28-32 pairs of somites). Interestingly, hemopoietic cells appear in parallel in this location, the paraaortic splanchnopleura, and in the yolk sac, where they represent a new generation by reference to the primitive hemopoietic stem cells. Lymphoid cell clones, which could differentiate into mature B cells, were obtained from yolk sac and paraaortic splanchnopleura cell preparations but not from other tissues of the embryonic body. These B-cell precursors were first detected around the stage of 10 somites; thereafter, their initial minute numbers increased in parallel in the yolk sac and the paraaortic splanchnopleura, suggesting that their emergence in the two sites was simultaneous. By single cell manipulation, we show that these precursors can generate B and T lymphocytes and myeloid cells; these precursors can thus be defined as multipotent hemopoietic cells.

Definitive erythropoiesis in birds originates from stem cells that emerge in the splanchnopleural mesoderm near the embryonic aorta. The yolk sac is still generally held to be the unique provider of haematopoietic stem cells during mammalian ontogeny, although there may be an alternative intraembryonic source of stem cells in the mouse fetus. Here we search for a possible non-yolk-sac source of stem cells by grafting intraembryonic splanchnopleura from 10- to 18-somite mouse embryos into adult immunodeficient SCID mice. We find significant amounts of donor-derived serum IgM, normal numbers of IgM-secreting plasma cells, and the B1a (IgM(a)brightB220dullCD5+) cell subset to be fully reconstituted by donor progenitors 3 to 6 months after engraftment. The haematogenic capacity revealed in our experiments is present in a previously unrecognized site, the earliest described in the embryo, 12 hours before fetal liver colonization.