Liver regeneration after partial hepatectomy results primarily from the simple division of mature hepatocytes. However, during embryonic and fetal development or in circumstances under which postnatal hepatocytes are injured, organ regeneration is believed to occur from a compartment of epithelial liver stem or progenitor cells with biliary and hepatocytic bipotentiality. The ability to identify, isolate, and transplant epithelial liver stem cells from fetal liver would greatly facilitate the treatment of hepatic diseases currently requiring orthotopic liver transplantation. Here we report the identification and immortalization by retrovirus-mediated transfer of the simian virus 40 large T antigen gene of primate fetal epithelial liver cells with a dual hepatocytic biliary phenotype. These cells grow indefinitely in vitro and express the liver epithelial cell markers cytokeratins 8͞18, the hepatocyte-specific markers albumin and ␣-fetoprotein, and the biliary-specific markers cytokeratins 7 and 19. Bipotentiality of gene expression was confirmed by clonal analysis initiated from single cells. Endogenous telomerase also is expressed constitutively. After orthotopic transplantation via the portal vein, Ϸ50% of the injected cells integrated into the liver parenchyma of athymic mice without tumorigenicity. Three weeks after transplantation, cells having seeded in the liver parenchyma expressed both albumin and ␣-fetoprotein but had lost expression of cytokeratin 19. These results provide strong evidence for the existence of a bipotent epithelial liver stem cell in nonhuman primates. This unlimited source of donor cells also should enable the establishment of a model of allogenic liver cell transplantation in a large animal closely related to humans and shed light on important questions related to liver organogenesis and differentiation.

Engineered retroviruses are widely used vectors for cancer gene therapy approaches. However, the ability to target cells of therapeutic interest while controlling the expression of the transferred genes would improve both the efficiency and the safety of viral vectors. In this study, we investigated the ability of a retroviral amphotropic envelope displaying single-chain variablefragment (scFv) directed against the c-Met receptor, to target the entry of recombinant retroviruses to human hepatocarcinoma cells. Four single-chain antibody fragments directed against the c-Met receptor were generated and inserted into the viral envelope protein as an N-terminal fusion. The modified envelopes were incorporated into virus particles and one of the chimeric viruses, 3D6-Env, transduced preferentially human hepatoma cells rather than proliferating human hepatocytes. In another construct, the urokinase cleavage site was inserted between the scFv moiety and the envelope. Chimeric scFv-urokinase-Env viruses transduced hepatoma cells with a similar efficiency to that of the control virus and their infectivity in human hepatocytes remained low. These results indicate that amphotropic retroviruses with engineered envelopes to display scFv directed against the c-Met receptor can efficiently and selectively deliver genes into hepatoma cells.

The main impediment to effective ex vivo liver gene therapy of metabolic diseases is the lack of experimental work on large animals to resolve such important issues as effective gene delivery, cell-processing techniques, and the development of appropriate vectors. We have used a nonhuman primate, as a preclinical model, to analyze the limiting steps of this approach using recombinant retroviruses. Seven monkeys (Macaca fascicularis) underwent the complete protocol: their left liver lobe was resected, a catheter was placed in the inferior mesenteric vein and connected to an infusion chamber, and the hepatocytes were isolated, cultured, and transduced with a retroviral vector containing the beta-galactosidase gene. The hepatocytes were harvested and returned to the host via the infusion chamber. Biopsies were taken 4-40 days later. No animal was killed in the course of the experiments. They all tolerated the procedure well. We have developed and defined conditions that permit the proliferation and transduction of up to 90% of the plated hepatocytes. A significant proportion of genetically modified cells, representing up to 3% of the liver mass, were safely delivered to the liver via the chamber. Polymerase chain reaction analysis detected integrated viral DNA sequences and quantitative analysis of the in situ beta-Gal-expressing hepatocytes indicated that a significant amount of transduced hepatocytes, up to 2%, had become integrated into the liver and were functional. These results represent substantial advances in the development of the ex vivo approach and suggest that this approach is of clinical relevance for liver-directed gene therapy.