Chronic Myeloid Leukemia Stem Cells Possess Unique Properties That Predict Intrinsic and Acquired Resistance to Imatinib Mesylate.

Chronic myeloid leukemia (CML) is sustained by a clonally amplified, but still rare population of BCR-ABL⁺ hematopoietic stem cells (HSCs). Understanding their properties is assuming increasing importance in light of recent evidence that they are innately resistant to BCR-ABL−targeted therapies and that, over time, the CML clone acquires BCR-ABL tyrosine kinase domain mutations. The relative resistance of CML HSCs (lin⁻CD34⁺CD38⁻ cells) to imatinib mesylate (IM) may be explained at least in part by their elevated expression of BCR-ABL and higher tyrosine kinase activity than is seen in the more prevalent lin⁻CD34⁺CD38⁺ leukemic cells. We have now obtained formal support for this concept from experiments with a BCR-ABL−transduced BaF3 cell line in which p210^BCR-ABL expression can be regulated by doxycycline (dox)-mediated activation of an upstream tet-regulated repressor. Treatment of these cells for 48 hr with 0.1–5 μM IM in the absence of IL-3 and in the presence of 0.1–1 μg dox/ml confirmed the expected dox-control of intracellular phospho-CrkL levels and demonstrated a corresponding p210^BCR-ABL kinase activity-dependent effect on IM sensitivity. To investigate the possibility that CML HSCs possess other unique properties that contribute to their relative resistance to many therapies, we compared the expression of 3 transporter genes (OCT1, ABCB1 and ABCG2) in CML cells at different stages of differentiation. Interestingly, we found that transcript levels for OCT1 (which regulates IM uptake) were very low in the most primitive (lin⁻CD34⁺CD38⁻) normal bone marrow cells and progressively increased (>100-fold) as these cells differentiate into mature (lin⁺CD34⁻) cells (n=4). Importantly, this difference in OCT1 expression was exaggerated in CML samples where OCT1 transcripts were even lower in the most primitive populations (below the level of detection in 2 of the 4 CML HSC isolates analyzed). Conversely, transcript levels for ABCB1 and ABCG2 (which regulate the efflux of many drugs) were highest in the normal lin⁻CD34⁺CD38⁻ (HSC) population and lowest (>6-fold) in the most mature lin⁺CD34⁻ normal cells, and again this difference was enhanced in the CML samples. The combination of a very low expression of OCT1 (low IM uptake) and highly elevated expression of ABCB1 and ABCB2 (high drug efflux) and BCR-ABL (elevated kinase activity) in CML HSCs identifies multiple mechanisms that would predict their broad insensitivity to IM and other therapeutics. In addition, we have found both by direct sequencing of cloned transcripts and allele-specific RT-PCR analyses, that a significant fraction of BCR-ABL transcripts in CML HSCs from chronic phase patients who have never received BCR-ABL-targeted therapy contain readily detectable kinase domain mutations (20–33%, n=3). A high incidence of mutations was also found at 2 additional sites in the BCR-ABL gene 5′ to the kinase domain and extending into the region encoded by BCR. After primitive CML cells had been cultured for 3 weeks (± 5 μM IM), we frequently found new BCR-ABL mutants in their clonal progeny. These in vivo and in vitro findings suggest that primary CML HSCs are also characterized by a high degree of genetic instability. We have thus identified multiple unique features of chronic phase CML HSCs that underscore the importance of evaluating these critical cells when considering new therapeutic approaches.