Abstract

Hematopoietic cells have been shown to generate nonhematopoietic cells, although the true plasticity of stem cells has been questioned. Here we used the NOD/SCID/γcnull mouse model, which permits efficient engraftment of human hematopoietic stem cells and their multi-lineage differentiation including T cells, to investigate whether human hematopoietic stem cells can differentiate into human hepatocytes. Freshly collected cord blood was depleted of phagocytes with Silica® followed by CD34 positive selection using auto MACS®. These cells were intravenously transplanted into irradiated mice, after which the liver was either undamaged or damaged by chemicals. The livers of these mice contained hepatocyte-specific (albumin, CYP family, TAT, alpha1AT, CPSI, prealbumin, transferrin and RBP4), cholangiocyte-specific (CK19) and vascular endothelial cell-specific (eNOS) human mRNAs. Immunohistochemistry detected the human hepatocyte specific antigens, albumin and alpha-1-antitrypsin-positive hepatocytes, cholangiocytes and CD68+ Kupffer cells. We also found human albumin in the murine bloodstream. Human albumin levels in the peripheral blood of transplanted mice correlate with the degree of PB chimerism and increase with time after transplantation. Furthermore, after obtaining liver cells by collagenase perfusion, flow cytometry revealed the presence of human albumin-positive cells that bear both human and murine MHC molecules, suggesting cell fusion occurs. All of the above phenomena were found in both liver-damaged and undamaged mice. In addition, we found human CD34+ cells are recruited from the murine bone marrow to the liver only in the case of acute liver injury but do not acquire hepatic stem/progenitor characteristics. Our observation suggests there are two pathways that yield hepatic cells from hematopoietic stem cells. The first requires liver damage that recruits CD34+ cells from the bone marrow via the circulation while the second pathway does not involve liver damage and appears to represent a constitutive default pathway of hematopoietic to nonhematopoietic transition. Our model is thus a versatile tool for investigating the development of functional human hepatic cells from hematopoietic cells and the feasibility of using hematopoietic cells in clinical situations.

Author notes

Corresponding author