Single-Cell Drug Profiling Reveals Maturation Stage-Dependent Drug Responses in AML

Introduction

Acute myeloid leukemia (AML) is a disease of myeloid lineage precursors with poor overall survival and limited treatment options in relapsed patients. Ex vivo drug testing is an approach to identify potential responders for novel targeted agents. However, AML bone marrow is comprised of a mixture of malignant and non-malignant cell types, and may contain myeloid cells at distinct differentiation stages. Here, we evaluated the ex vivo sensitivity of primary AML samples at a detailed cell-population level using a multicolor high-throughput flow cytometer (FACS).

Methods

Bone marrow (BM) mononuclear cells of 23 AML patients were plated on 96-well plates in mononuclear cell media (PromoCell). Seven FDA approved drugs were used, including signal transduction inhibitors that had shown AML specific responses in our earlier studies ( Pemovska et al. Cancer Discovery 2013 ): trametinib, ruxolitinib, everolimus, sunitinib, venetoclax, plus standard of care drugs idarubicin and cytarabine. After 3-day incubation, cells were stained with six cell surface markers (CD45, CD34, CD33, CD14, CD123, CD38) to identify major cell populations present in many AML samples: blasts, promonocytes/monocytes, lymphocytes and granulopoietic cells. Flow cytometric analysis was performed using the iQue Screener PLUS instrument (Intellicyt). Apoptotic and dead cells were removed from the analysis (Annexin-V/7-AAD) and live cell counts of different cell populations after drug treatment were measured. For comparison, the viability of the samples was also assessed using the CellTiter-Glo (CTG) assay that uses the amount of ATP as an indicator of metabolically active cells.

Results

Immunophenotyping of the AML samples showed large inter-sample variation in cell composition. The BM blast cell population out of the CD45+ positive cells ranged between 17-94% and the non-malignant lymphocyte population ranged from 0.4 to 46%. High numbers of promonocytes/monocytes were observed in samples from FAB M4/5 patients as expected, while FAB M0/1/2 patient samples mainly consisted of blasts and lymphocytes. FACS analysis of cell population specific drug responses revealed that blast cells of different patients had highly different drug sensitivities whereas lymphocytes and granulocytes did not show differences in drug sensitivities between the AML samples (Figure 1A). Interestingly, promonocytes/monocytes that are mainly present in FAB M4/5 samples but also in small quantities in some FAB M0/1/2 samples were systematically more sensitive to MEK inhibition, but more resistant to BCL-2 inhibition when compared to blasts (Figure 1A-B). In line with this observation, the average overall BM cell sensitivity of myelomonocytic/monocytic (FAB M4/5) samples as measured by CTG showed decreased sensitivity to venetoclax (P<0.01) and increased sensitivity to trametinib (P<0.05) when compared to less mature (FAB M0/1/2) AMLs. Initial differential gene expression analysis revealed enrichment of genes in the RAS/MAPK pathway in samples diagnosed with FAB M4/5. Amongst the tested compounds, venetoclax showed the highest toxic effect against blast cells even though 40% of the samples were at least partly resistant. Intriguingly, we observed that by combining venetoclax (50nM) with trametinib (25nM) or ruxolitinib (300nM), chemo- and venetoclax-resistant blast cells were effectively killed ex vivo (>80%) in all tested samples.

Conclusion

FACS-based drug testing ex vivo enables accurate drug sensitivity profiling of AML blast cells even in samples with high numbers of more mature leukemic cells and healthy cells. Our data indicate heterogeneity in drug responses of AML cells at different stages of myeloid differentiation, with high sensitivity to BCL-2 inhibition and resistance to MEK inhibition of the primitive blasts compared to more mature monocytic cells. MEK inhibitor sensitivity in FAB M4/5 AMLs might be linked to the inherent characteristics of monocyte lineage cells. Furthermore, we show that venetoclax-resistant blast cells can be effectively targeted by combining venetoclax with either MEK inhibitor trametinib or JAK inhibitor ruxolitinib.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal