Chronic myeloid leukemia (CML) stem/progenitor cells are relatively insensitive to tyrosine kinase inhibitor (TKI) monotherapies, and may not exclusively rely on BCR-ABL activity for survival. This population rapidly generates therapy-resistant clones and is often responsible for relapse after TKI therapy discontinuation. It has been reported that Hedgehog (HH) signalling in hematopoietic stem/progenitor cells is essential for CML induction by BCR-ABL in a mouse model. Inhibition of Smoothened (SMO), a key regulator of the HH pathway, reduces the number of BCR-ABL+ stem cells in vitro and slows disease progression in vivo. However, it is not known if SMO and other key regulators are differentially expressed in stem/progenitor cells derived from imatinib (IM)-nonresponders and IM-responders. If so, they may serve as potential biomarkers and/or therapeutic targets. We have assessed gene expression changes in the HH pathway using RNA sequencing (RNA-seq) in 6 highly purified CD34+ stem/progenitor samples from newly-diagnosed chronic phase (CP)-CML patients. After IM therapy was initiated, 3 patients were classified retrospectively as IM-responders, and 3 as IM-nonresponders. This study identified 27 differentially expressed HH genes between healthy and CML patients (>1.5-fold). In particular, SMO and GLI2, key components of the HH pathway,were upregulated in CP-CML patients relative to healthy bone marrow (HBM) controls (n=3).In particular, GLI2 was highly overexpressed in CP-CML compared to HBM (48-fold upregulated in IM-responders and 166-fold upregulated in IM-nonresponders). To validate the RNA-seq results, expression of the principal HH genes (PTCH1, GLI1, GLI2, SMO) was assessed in 18 CD34+ CP-CML patient samples and 8 HBM samples using qRT-PCR. Interestingly, the expression levels correlated with patients' TKI resistance status, with IM-nonresponders expressing significantly higher levels of SMO (p<0.01)and GLI2 (p<0.05) compared with IM-responders. GLI2 and SMO expression was then assessed in CD34-subpopulations in 5 IM-responders and 4 IM-nonresponder samples. GLI2 was exclusively and highly expressed in the most primitive population (lin-CD34+CD38-), while its expression was very low in more mature populations (lin-CD34+38+ and CD34-). This effect was amplified in IM-nonresponder samples, where GLI2 is greater than 40-fold upregulated in the lin-CD34+CD38- subpopulation compared with IM-responders. In contrast, SMO was more uniformly expressed in both stem and progenitor subpopulations, but was more highly expressed in IM-nonresponders than IM-responders.These results suggest that key regulators of the HH pathway are highly expressed in IM-nonresponder stem cells and that these cells may be more responsive to SMO inhibition than IM-responder cells.

We then performed a viability and apoptosis analysis of 5 CD34+ CP-CML patient samples. IM-nonresponder samples were more sensitive to SMO inhibition using a highly selective SMO inhibitor (PF-04449913) compared with IM-responders, and this effect was enhanced in combination with the new TKI bosutinib (BOS). To investigate the effects of BOS and PF-04449913 on CML progenitor cells, colony forming cell assays were performed on CD34+ CP-CML cells from 4 IM-responders and 3 IM-nonresponders. Overall, combination treatment with BOS and PF-04449913 only modestly reduced colony forming ability compared to either agent alone. The inhibitory effects of combination therapy were, however, strikingly enhanced in a replating experiment, resulting in a greatly reduced replating efficiency compared to BOS treatment alone (15% vs. 30%), suggesting that the combination specifically targets primitive CP-CML cells. Taken together, we have demonstrated that both SMO and GLI2 are highly expressed in primitive CP-CML cells, and that the transcript levels of GLI2 are significantly increased in CP-CML stem cells from IM-nonresponders, which may serve as a new biomarker to predict patient TKI responsivity. We also found that IM-nonresponder stem/progenitor cells are more sensitive to SMO suppression in short- and long-term assays compared with IM-responders, suggesting that HH pathway activation may comprise a potential mechanism of resistance during TKI therapy, and that dual inhibition of the BCR-ABL and HH pathways may constitute a rational approach to abrogate drug resistance and disease progression in CML.


Woolfson:Pfizer Inc: Employment.

Author notes


Asterisk with author names denotes non-ASH members.