Within 1 wk of liquid culture in granulocyte/macrophage colony-stimulating factor (GM-CSF), normal B10 BR (H-2k I-E+) mouse liver nonparenchymal cells (NPC) formed loosely adherent myeloid cell clusters that have been shown to contain dendritic cell (DC) progenitors in similar studies of mouse blood or bone marrow. Mononuclear cell progeny released from these clusters at and beyond 4 d exhibited distinct dendritic morphology and were actively phagocytic. After 6-10 d of culture, these cells strongly expressed CD45, CD11b, heat stable antigen, and CD44. However, the intensity of expression of the DC-restricted markers NLDC 145, 33D1, and N418, and the macrophage marker F4/80, intercellular adhesion molecule 1, and Fc gamma RII was low to moderate, whereas the cells were negative for CD3, CD45RA, and NK1.1. Splenocytes prepared in the same way also had a similar range and intensity of expression of these immunophenotypic markers. Unlike the splenic DC, however, most of the GM-CSF-propagated putative liver DC harvested at 6-10 d expressed only a low level of major histocompatibility complex (MHC) class II (I-Ek), and they failed to induce primary allogeneic responses in naive T cells, even when propagated additionally in GM-CSF and tumor necrosis alpha and/or interferon gamma-supplemented medium. However, when 7-d cultured GM-CSF-stimulated liver cells were maintained additionally for three or more days on type-1 collagen-coated plates in the continued presence of GM-CSF, they exhibited characteristics of mature DC: MHC class II expression was markedly upregulated, mixed leukocyte reaction stimulatory activity was increased, and phagocytic function was decreased. Similar observations were made when Ia+ cells were depleted from the GM-CSF-propagated cells before exposure to collagen. Further evidence that the GM-CSF-stimulated class IIdim or class II-depleted hepatic NPC were immature DC was obtained by injecting them into allogeneic B10 (H-2b I-E-) recipients. They "homed" to T cell-dependent areas of lymph nodes and spleen where they strongly expressed donor MHC class II antigen 1-5 d later. These observations provide insight into the regulation of DC maturation, and are congruent with the possibility that the migration of immature DC from normal liver and perhaps other organ allografts may help explain their inherent tolerogenicity.