Development of Leukemia after HOXB4 Gene Transfer in the Canine Model.

HOXB4 is considered to be the only HOX gene that promotes self-renewal of hematopoietic stem cells without causing leukemia in mouse models. We investigated whether HOXB4 overexpression has similar effects in a clinically relevant canine model. A competitive repopulation assay was performed in three dogs in which CD34⁺ cells were transduced with MSCV-based gammaretroviral vectors expressing HOXB4GFP or control YFP. We observed up to 4-fold higher marking levels in granulocytes for the HOXB4GFP arm relative to the control 1 month after transplantation. The marking levels eventually decreased in all three animals and two dogs (G374, G450) have now been followed for more than 18 months. In G374, the marking levels for both arms stabilized at ~2% after 2 months post-transplantation. Between 14 and 20 months post-transplantation, the HOXB4GFP marking steadily increased to >95%, while YFP marking decreased to 0.1%. G374 was euthanized 21 months after transplantation due to declining health. Flow cytometry analysis showed that ~50% of BM cells expressed the monocyte marker CD14 and ~8% expressed the granulocyte marker DM5, all of which also expressed HOXB4GFP. CD3 and CD21 were expressed in 2% and 1% of cells, respectively, but these cells did not express HOXB4GFP. Bone marrow necropsy demonstrated significantly increased numbers of blast cells, consistent with a myelomonocytic leukemia. Southern blot analyses of G374 BM and PB samples identified 2 bands with the same intensity, suggesting a single dominant clone with 2 integration sites. LAM-PCR analysis identified two vector proviruses integrated ~100 kb upstream of c-myb, and into intron 3 of PRDM16. Western blot analysis confirmed expression of HOXB4 in cultured G374 BM cells but the levels of c-myb in these cells were not different from control HOXB4-transduced BM cells as determined by RT-PCR. The expression of PRDM16 exons 1–3 was not detected in cells from dog G374 or in control cells, however, PRDM16 exon 4 was expressed in G374 cells but not in control cells. RT-PCR using primers located in the MSCV LTR and in PRDM16 exon 4 identified a unique band and sequencing of this product showed that the 5′ LTR was spliced with PRDM16 exon 4 creating a short PRDM16 isoform which has been observed in human leukemias. These data suggest that HOXB4 overexpression in collaboration with integration-induced activation of PRDM16 led to the leukemia. Southern blot and SYBR green Q-PCR showed that the leukemic clone contributed to ~20% hematopoiesis in BM 6 months after transplantation, and gradually decreased to ~2% before final expansion of the clone, suggesting accumulation of other mutation(s) were required for overt leukemia. Karyotype analysis of BM cells has not shown any major abnormalities but we are currently performing analyses to search for minor abnormalities such as gene duplications and deletions. Recently, HOXB4GFP marking in dog G450 PB and BM has increased to 20% and 80%, respectively, while YFP marking has decreased to ~1%. Southern blot analysis has identified a single dominant band and a BM biopsy showed substantially increased blast cells. Of note, we have not observed leukemia in >30 dogs followed long term that received transduced cells without HOXB4. In summary, HOXB4 overexpression together with insertional mutagenesis by virus integration has induced leukemia in the canine model, demonstrating the utility of this model to study the safety of gene therapy.