Human acute myeloblastic leukemia-ascites model using the human GM-CSF- and IL-3-releasing transgenic SCID mice.

To generate an appropriate model for human acute myeloblastic leukemia (AML), we have successfully established a human hematopoietic growth factor-dependent AML cell line (TF-1 and UT-7/GM)-ascites model using human granulocyte-macrophage colony-stimulating factor (hGM-CSF)- and human interleukin 3 (hIL-3)-releasing transgenic (Tg)-SCID mice. When 1×107 cells of TF-1, a human erythroleukemia cell line, were transplanted into the peritoneum of irradiated Tg-SCID mice (TF-1 ip/Tg-SCID mice), TF-1 cells grew in both the single cell suspension form (asTF-1) and solid form in ascites and invaded various tissues: lungs, liver, pancreas, and genitals, 3–6 weeks following transplantation. Subsequently, 0.5–1×107 cells of UT-7/GM, a subline of the UT-7 human megakaryoblastic leukemia cell line, grown in the back of hGM-CSF Tg-SCID mice after subcutaneous inoculation, were transplanted into the peritoneum of other irradiated hGM-CSF Tg-SCID mice. After 4 weeks, UT-7/GM cells (asUT-7/GM) also grew in the same manner as TF-1 cells in hGM-CSF Tg-SCID mice. Analysis of the cells from the peritoneum and tissues by PCR amplifying ALU and human GM-CSF receptor β sequences and by immunohistochemical staining using anti-human CD45 revealed that they possessed the original characteristics of the parental cells. To confirm the usefulness of this human AML-ascites model, experimental treatment of AML cells grown in these mice was carried out with a differentiation inducer, delta-aminolevulinic acid (δ-ALA), which induces hemoglobin synthesis for TF-1 in vitro and is thus regarded as an anti-leukemia drug candidate. Unexpectedly, growth promotion of TF-1 cells was observed in the treated TF-1 ip/hIL-3 Tg-SCID mice without differentiation to erythroid cells after treatment with δ-ALA (5 mM) for 7 days. These results indicate that Tg-SCID mice can support the growth of human hematopoietic growth factor-dependent AML cell lines which are usually rejected by SCID mice, without modification of the parental cell characteristics. In addition, this Tg-SCID leukemia-ascites model may become a useful preclinical tool for estimation of drug efficacy in vivo, since the drug candidate which was promising in vitro did not act in the same manner in vivo.