Molecular Signature of Dormancy in CD34+CD38- Acute Myeloid Leukaemia Cells

Chemotherapy drugs tend to spare cells in a state of dormancy (G0 phase of the cell cycle). Relapse of acute myeloid leukaemia (AML) is likely in part due to dormant cells evading remission-induction chemotherapy. Dormant AML cells have been identified in the bone marrow endosteal region which is characterised by an excess of TGFβ1 and a shortage of nutrients. We developed and characterized an in-vitro model of AML cell dormancy by exploiting these features.

Following preliminary investigation of several cell lines, the CD34+CD38- line TF1a was selected for in depth investigation. TF1a cells showed 72% inhibition of proliferation (p<0.001), with features of dormancy, in response to 72 hours TGFβ1+mTOR inhibitor treatment (mTOR pathway inhibition mimics major effects of nutrient scarcity).This treatment caused loss of Ki-67, as well as upregulating ALDH and CD34 and causing nuclear translocation of FOXO3a. In contrast to conventional serum-withdrawal assays for dormancy, the treatment had no impact on cell viability, assessed by Annexin V assay. Nor did the treatment lead to cell differentiation, assessed by CD11b staining and morphology.

Using whole human genome expression microarray and by intersecting differentially regulated genes in dormancy-induced TF1a cells in comparison to their proliferating (untreated) counterparts, we identified 240 genes which are significantly up-regulated >2 fold including genes involved in stemness, chemoresistance and tumour suppressor genes in addition to genes involved in canonical cell cycle regulation. There was striking upregulation of genes involved in adhesion and migration: raised expression levels of SPP1 (the gene coding for osteopontin), ITGB3, ITGB4, ITGA3 and CD44 in dormant cells were confirmed by real-time PCR. The most upregulated gene was SPP1/osteopontin (16 fold). Immunocytochemistry of biopsy material from AML patients confirmed high levels of osteopontin in the cytoplasm of blasts near the paratrabecular bone marrow. Osteopontin and other genes identified in this model, including well-characterised genes (e.g. CD44, CD47, CD123, ABBC3 and CDKN2B) as well as little-known ones (e.g. PTPRU, ITGB3 and BTG2), are potential therapeutic targets in dormant AML cells.