The development of asciminib (ABL001),1 a highly potent and specific ABL1/ABL2/BCR-ABL1 tyrosine kinase inhibitor (TKI) with an allosteric mechanism of action that culminated from research in academic and pharmaceutical industry laboratories spanning more than a decade.Currently in clin. trials for relapsed/refractory Philadelphia chromosome-pos. (Ph+) leukemia patients (ClinicalTrials.gov: NCT02081378), asciminib represents a major advance from the structurally related, prototype compounds GNF-2 and GNF-5.2,3,4 We investigated mechanisms of asciminib resistance and identified two major categories: (1) upregulation of the ABCG2 efflux pump, resulting in undetectable intracellular asciminib levels and (2) emergence of BCR-ABL1 mutations at the myristoyl-binding site and at a distant residue.We generated five asciminib-resistant, BCR-ABL1-pos. cell lines by adapting to increasing concentrations of asciminib (range: 20-10 000 nm): K562asciminib-R, LAMA84asciminib-R, KYO1asciminib-R, Ba/F3 BCR-ABL1asciminib-R and KCL-22asciminib-R cells (Supplementary Methods).In K562asciminib-R cells, methanethiosulfonate-based cell proliferation assays demonstrated a ∼60-fold increase in asciminib IC50 compared to parental K562 cells, despite remaining sensitive to the ATP-competitive TKIs imatinib, nilotinib, dasatinib and ponatinib (Figure 1a; Supplementary Table S1).Immunoblot anal. revealed similar results, showing marked restoration of BCR-ABL1 tyrosine kinase activity and downstream STAT5 tyrosine phosphorylation in the presence of asciminib but not ponatinib (Supplementary Figure S1a).However, next-generation sequencing (NGS) and Sanger sequencing of the BCR-ABL1 kinase domain identified no mutations (Supplementary Table S2).5 To check for the possibility of reduced intracellular drug concentrations, asciminib levels were measured following treatment using a customized liquid chromatog.-mass spectrometry/mass spectrometry (LC-MS/MS) method.6 Asciminib was undetectable in K562asciminib-R cells, but present at substantial levels in parental K562 cells (Supplementary Figure S1b), with an inhibitor-based screen for potential involvement of efflux pumps (Supplementary Figure S1c) and anal. by qPCR and immunoblot (Figure 1b) all implicating ABCG2.Cell proliferation experiments revealed that the ABCG2 inhibitor Ko143 (100 nm) restored asciminib effectiveness against K562asciminib-R cells but had no effect on the asciminib IC50 for K562 cells (Figure 1c).Similar results were obtained for LAMA84asciminib-R (Supplementary Figure S2; Supplementary Table S1) and KYO1asciminib-R cells (Supplementary Figure S3; Supplementary Table S1).Our findings support ABCG2-mediated efflux of asciminib as the major mechanism of resistance in these cell lines, warrant its profiling among patients with asciminib resistance in the clinic, and suggest that combining asciminib with an ABCG2 inhibitor could override resistance, though development of clin. ABCG2 inhibitors is at the investigational stage.7, 8 In contrast, mutation-based resistance mechanisms were observed in Ba/F3 BCR-ABL1asciminib-R and KCL-22asciminib-R cells.In Ba/F3 BCR-ABL1asciminib-R cells, a>1000-fold increase in asciminib IC50 over Ba/F3 BCR-ABL1 cells was observed, despite undiminished sensitivity to ATP-competitive TKIs (Figure 2a; Supplementary Table S1).While immunoblot anal. demonstrated restored BCR-ABL1 signaling in Ba/F3 BCR-ABL1asciminib-R cells treated with asciminib (Supplementary Figure S4a), LC-MS/MS anal. showed similar amounts of asciminib in Ba/F3 BCR-ABL1 and Ba/F3 BCR-ABL1asciminib-R cells and there was no evidence of efflux pump involvement (Supplementary Figure S4b-e).NGS and Sanger sequencing of the BCR-ABL1 kinase domain identified a novel BCR-ABL1C464W mutation (Supplementary Table S2), mol. modeling of which suggested that the bulky tryptophan residue impedes access of asciminib to the myristoyl-binding pocket, consistent with high-level resistance (Figure 2b).While other BCR-ABL1 mutations within or near the myristoyl-binding pocket (for example, A337V; P465S; V468F) have been shown to confer asciminib resistance,1 this is the first report of BCR-ABL1C464W as an asciminib-resistant mutant.KCL-22asciminib-R cells exhibited two mutations at similar allelic frequencies:9 BCR-ABL1M244V near the ATP-binding site and BCR-ABL1A337V in the myristoyl-binding pocket (Supplementary Table S2), and single-cell sorting followed by clonal sequencing revealed a BCR-ABL1M244V/A337V compound mutation.10, 11 KCL-22asciminib-R cells were completely insensitive to asciminib (IC50: 10 000 nm as compared to 2.2 nm for parental KCL-22 cells), but remained sensitive to ATP-binding site TKIs (Figure 2c; Supplementary Figure S5a; Supplementary Table S1).LC/MS-MS showed comparable intracellular asciminib levels in KCL-22asciminib-R and KCL-22 cells, and there was no evidence of either altered effectiveness of asciminib against KCL-22asciminib-R cells by inclusion of an efflux pump inhibitor or upregulation of an efflux pump by qPCR or ABCG2 immunoblot (Supplementary Figure S5b-e).Sequencing of increasingly asciminib-resistant cells collected in the process of generating the final KCL-22asciminib-R cell line revealed a progression from BCR-ABL1A337V to BCR-ABL1M244V/A337V (Figure 2d).The reported asciminib IC50 value of 702 nM for KCL-22A337V cells1 is consistent with the inability of this mutation alone to confer the high-level resistance observed in KCL-22asciminib-R cells.The BCR-ABL1M244V/A337V compound mutant confers high-level asciminib resistance but inclusion of clin. achievable concentrations of imatinib lead to outgrowth of BCR-ABL1S229P/T315I at the expense of BCR-ABL1M244V/A337V (Supplementary Figure S6), highlighting a potential complication of addressing asciminib resistance by switching to or including an ATP-binding site TKI.Taken together, our findings suggest that mechanisms of acquired resistance to the allosteric BCR-ABL1 inhibitor asciminib involve restoration of BCR-ABL1 signaling, due to either asciminib efflux or select BCR-ABL1 mutations.It may be possible to override ABCG2-mediated drug efflux by dose escalation of asciminib.The achievable clin. dose is estimated to be at least 1μm.12 Another possibility to circumvent this resistance mechanism is to design an allosteric inhibitor that is not a substrate for efflux pump(s).While development of a next-generation allosteric inhibitor13 is possible, the potential for cross-resistant myristoyl-binding pocket mutations may be high, as those observed for asciminib to date completely block inhibitor access to the binding pocket.Alternatively, myristoyl-binding site mutation-based resistance to asciminib could be countered with an ATP-binding site TKI, though such a mutation occurring in tandem with an addnl. kinase domain mutation could result in resistance to both types of inhibitor.The strategy of simultaneously treating with asciminib and an ATP-binding site TKI (imatinib, nilotinib or dasatinib) to minimize opportunity for resistance is under clin. investigation.Ponatinib, which has activity against the T315I mutant and reported potency as an ABCG2 inhibitor,14, 15 may also warrant consideration in this setting.
