Commonly used in vivo models of prostate cancer metastasis include syngeneic rodent cancers and xenografts of human cancer in immunodeficient mice. However, the occurrence of osseous metastases in these models is rare, and in xenograft models, species-specific factors may limit the ability of human cells to metastasize to rodent bones. We have modified the severe combined immunodeficient (SCID)-human model to test the ability of circulating human prostate cancer cells to home to macroscopic fragments of human bone and other organs previously implanted into SCID mice. We have also compared the growth of human prostate cancer cells in various human and mouse tissue microenvironments in vivo. Macroscopic fragments of human fetal bone, lung, or intestine (16-22 weeks gestation) or mouse bone were implanted s.c. into male CB.17 SCID mice. Four weeks later, human prostate cancer cells were injected either i.v. via the tail vein (circulating cell colonization assay) or directly into the implanted tissue fragments transdermally (end organ growth assay). Tumor growth was followed for 6 weeks by palpation and magnetic resonance imaging. After 6 weeks, tumors were enumerated in implanted human and mouse organ fragments and native mouse tissue. Tumors were characterized by histology, immunohistochemistry, and chromosomal analysis. After i.v. injection, circulating PC3 cells successfully colonized implanted human bone fragments in 5 of 19 mice. Tumors were easily followed by palpation and imaging and had an average volume of 258 mm3 at autopsy. Histological examination revealed osteolysis and a strong desmoplastic stromal response, which indicated intense stromal-epithelial interaction. Bone tumors were subcultured, and chromosomal analysis demonstrated that the tumors were derived from the parental prostate cancer cell line. Microscopic tumor colonies were also found in a few mouse lungs after i.v. injection of PC3, DU145, and LNCaP cells, however the volume of the lung nodules was less than 1 mm3 in all of the cases. No colonization of human lung or intestine implants, the mouse skeleton, or other mouse organs was detected, demonstrating a species- and tissue-specific colonization of human bone by PC3 cells. Direct injection of 10(4) prostate cancer cells into human bone implants resulted in large tumors in 75-100% of mice. PC3 and DU145 bone tumors were primarily osteolytic, whereas LNCaP bone tumors were both osteoblastic and osteolytic. PC3 and LNCaP bone tumors showed a desmoplastic stromal response, which indicated intense stromal-epithelial interaction. All three of the cell lines formed tumors in implanted human lung tissue; however, the tumors were all < or = 10 mm3 in volume and showed minimal stromal involvement. No tumors formed after either s.c. injection or injection of cells into implanted mouse bone demonstrating both species- and tissue-specific enhancement of growth of human prostate cancer cells by human bone. The severe combined immunodeficient-human model provides a useful system to study species-specific mechanisms involved in the homing of human prostate cancer cells to human bone and the growth of human prostate cancer cells in human bone.