BCR-ABL1 Compound Mutations Combining Key Kinase Domain Positions Confer Clinical Resistance to Ponatinib in Ph Chromosome-Positive Leukemia

Transcription

1 Cancer Cell Article BCR-ABL1 Compound Mutations Combining Key Kinase Domain Positions Confer Clinical Resistance to Ponatinib in Ph Chromosome-Positive Leukemia Matthew S. Zabriskie, 1,27 Christopher A. Eide, 2,3,27 Srinivas K. Tantravahi, 1,4 Nadeem A. Vellore, 5 Johanna Estrada, 1 Franck E. Nicolini, 6 Hanna J. Khoury, 7 Richard A. Larson, 8 Marina Konopleva, 9 Jorge E. Cortes, 9 Hagop Kantarjian, 9 Elias J. Jabbour, 9 Steven M. Kornblau, 9 Jeffrey H. Lipton, 1 Delphine Rea, 11 Leif Stenke, 12 Gisela Barbany, 13 Thoralf Lange, 14 Juan-Carlos Hernández-Boluda, 15 Gert J. Ossenkoppele, 16 Richard D. Press, 17 Charles Chuah, 18 Stuart L. Goldberg, 19 Meir Wetzler, Francois-Xavier Mahon, 21 Gabriel Etienne, 22 Michele Baccarani, 23 Simona Soverini, 23 Gianantonio Rosti, 23 Philippe Rousselot, 24 Ran Friedman, 25 Marie Deininger, 1 Kimberly R. Reynolds, 1 William L. Heaton, 1 Anna M. Eiring, 1 Anthony D. Pomicter, 1 Jamshid S. Khorashad, 1 Todd W. Kelley, 26 Riccardo Baron, 5 Brian J. Druker, 2,3 Michael W. Deininger, 1,4,28, * and Thomas O Hare 1,4,28, * 1 Huntsman Cancer Institute, University of Utah, Salt Lake City, UT 84112, USA 2 Division of Hematology and Medical Oncology, Oregon Health & Science University Knight Cancer Institute, Portland, OR 97239, USA 3 Howard Hughes Medical Institute, Portland, OR 97239, USA 4 Division of Hematology and Hematologic Malignancies, University of Utah, Salt Lake City, UT 84112, USA 5 Department of Medicinal Chemistry, College of Pharmacy and The Henry Eyring Center for Theoretical Chemistry, University of Utah, Salt Lake City, UT 84112, USA 6 Hematology Department 1F, Centre Hospitalier Lyon Sud, Pierre Bénite, INSERM U152, CRCL, Lyon 69495, France 7 Department of Hematology and Medical Oncology, Winship Cancer Institute of Emory University, Atlanta, GA 3322, USA 8 University of Chicago, Chicago, IL 637, USA 9 Departments of Leukemia and Stem Cell Transplantation and Cellular Therapy, University of Texas MD Anderson Cancer Center, Houston, TX 773, USA 1 Department of Medical Oncology and Hematology, Allogeneic Blood and Marrow Transplantation Program, Princess Margaret Hospital, University of Toronto, Toronto ON M5G 2M9, Canada 11 Service des Maladies du Sang, Hospital Saint-Louis, 751 Paris, France 12 Department of Hematology, Karolinska Institutet and University Hospital, Stockholm, Sweden 13 Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden 14 Hematology and Oncology, University of Leipzig, 413 Leipzig, Germany 15 Hematology and Medical Oncology Department, Hospital Clínico Universitario, 41 Valencia, Spain 16 Department of Hematology, VU University Medical Center, Amsterdam 181HV, the Netherlands 17 Department of Pathology and Knight Cancer Institute, Oregon Health & Science University, Portland, OR 97239, USA 18 Department of Hematology, Singapore General Hospital, Program in Cancer and Stem Cell Biology, Duke-NUS Graduate Medical School, Singapore, Singapore 19 John Theurer Cancer Center at Hackensack University Medical Center, Hackensack, NJ 71, USA Roswell Park Cancer Institute, Buffalo, NY 14263, USA 21 Laboratoire d Hematologie, Centre Hospitalier Universitaire de Bordeaux and Laboratoire Hematopoïese Leucemique et Cible Therapeutique, Inserm U135, Universite Bordeaux, 3376 Bordeaux, France 22 Departement d Oncologie Medicale, Centre Regional de Lutte Contre le Cancer de Bordeaux et du Sud-Ouest, Institut Bergonie, 3376 Bordeaux, France 23 Department of Experimental, Diagnostic, and Specialty Medicine, Institute of Hematology L. e A. Seràgnoli, University of Bologna, 138 Bologna, Italy 24 Service d Hématologie et d Oncologie, Université de Versailles, 751 Paris, France 25 Department of Chemistry and Biomedical Sciences and Centre for Biomaterials Chemistry, Linnaeus University, Kalmar, Sweden 26 Department of Pathology, University of Utah, Salt Lake City, UT 84112, USA 27 Co-first author 28 Co-senior author *Correspondence: michael.deininger@hci.utah.edu (M.W.D.), thomas.ohare@hci.utah.edu (T.O.) SUMMARY Ponatinib is the only currently approved tyrosine kinase inhibitor (TKI) that suppresses all BCR-ABL1 single mutants in Philadelphia chromosome-positive (Ph + ) leukemia, including the recalcitrant BCR-ABL1 mutant. However, emergence of compound mutations in a BCR-ABL1 allele may confer ponatinib resistance. We found that clinically reported BCR-ABL1 compound mutants center on 12 key positions and confer varying resistance to imatinib, nilotinib, dasatinib, ponatinib, rebastinib, and bosutinib. -inclusive compound mutants confer high-level resistance to TKIs, including ponatinib. In vitro resistance profiling was 428 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

2 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia predictive of treatment outcomes in Ph + leukemia patients. Structural explanations for compound mutationbased resistance were obtained through molecular dynamics simulations. Our findings demonstrate that BCR-ABL1 compound mutants confer different levels of TKI resistance, necessitating rational treatment selection to optimize clinical outcome. INTRODUCTION Tyrosine kinase inhibitors (TKIs) targeting BCR-ABL1 (Druker et al., 6) have dramatically improved the prognosis of chronic myeloid leukemia (CML) and, to a lesser extent, Philadelphia chromosome-positive (Ph + ) acute lymphoblastic leukemia (ALL). However, TKI resistance occurs in % 3% of CML patients (O Hare et al., 12) and is commonly attributable to point mutations in the BCR-ABL1 kinase domain. The TKIs approved for first-line therapy, imatinib (Apperley, 7; Azam et al., 3; Bradeen et al., 6), nilotinib (Weisberg et al., 5), and dasatinib (Shah et al., 4), and the second-line therapy, bosutinib (Cortes et al., 11; Redaelli et al., 9), demonstrate overlapping resistance profiles, with the BCR-ABL1 mutant a shared vulnerability (O Hare et al., 12). Additionally, some patients fail therapy despite inhibition of BCR-ABL1, implicating activation of alternative, BCR-ABL1 kinase-independent resistance mechanisms (Dai et al., 4; Donato et al., 3; Hochhaus et al., 2). Ponatinib (O Hare et al., 9) is a high-affinity, pan-bcr- ABL1 TKI with the unique property of inhibiting BCR-ABL1. Anti-leukemic activity has been observed in clinical trials of ponatinib, including patients with BCR-ABL1, although responses in patients with blastic phase CML (CML-BP) or Ph + ALL are typically transient (Cortes et al., 12, 13). After a hold due to safety concerns pertaining to vascular occlusion events, regulatory approval in the United States was reinstated for patients with refractory Ph + leukemia harboring BCR- ABL1 or for whom no other TKI is indicated (Senior, 14). A risk of sequential TKI treatment is the selection of BCR-ABL1 compound mutants, defined as a BCR-ABL1 allele harboring two or more mutations, that have the potential to confer resistance to multiple TKIs (Shah et al., 7). Vulnerability of ponatinib to certain two-component compound mutations was demonstrated in preclinical studies (O Hare et al., 9), suggesting they may emerge as a clinical problem in patients treated with ponatinib. Importantly, ultra-deep sequencing of serial samples from Ph + leukemia patients who had received sequential TKI treatment showed that the majority (76%) of BCR-ABL1 compound mutations were two-component mutations, as compared to 21% triple and 3% quadruple mutations (Soverini et al., 13). Progress in the development of a next generation sequencing approach spanning the BCR-ABL1 kinase domain in a single read was recently reported (Kastner et al., 14). The ability of available TKIs to address resistance due to clinically reported BCR-ABL1 compound mutants has yet to be investigated. In this study, we inventoried clinically reported BCR-ABL1 compound mutations and established in vitro TKI sensitivity profiles of BCR-ABL1 compound mutants against a panel of clinically available TKIs. RESULTS Key BCR-ABL1 Kinase Domain Positions Are Frequently Represented in Clinically Reported Compound Mutants Over BCR-ABL1 kinase domain point mutations have been linked with clinical imatinib resistance (Apperley, 7), and resistance profiles for newer BCR-ABL1 TKIs are mainly comprised of subsets of these mutations. In the current study, all uses of the term compound mutation refer to two-component compound mutations unless otherwise stated. Thorough inventory of clinical BCR-ABL1 compound mutations associated with TKI resistance reported in the published literature identified a limited list of 12 kinase domain positions (Figure 1A) comprising the majority of compound mutations, which we refer to as key positions. All clinically reported compound mutations (%) in Figure 1 include a key position, and the majority (65%) involve two (Figures 1B and 1C). Each position has been implicated in resistance to one or more TKIs: imatinib (Bradeen et al., 6; Gorre et al., 1), nilotinib (Bradeen et al., 6; Ray et al., 7; Weisberg et al., 5), dasatinib (Bradeen et al., 6; Burgess et al., 5; Shah et al., 4), bosutinib (Redaelli et al., 9), ponatinib (O Hare et al., 9), and rebastinib (Chan et al., 11; Eide et al., 11). The key residues in native BCR-ABL1 are: M244, G25, Q252, Y253, E255, V299, F311, T315, F317, M351, F359, and H396 (Figure 1A). Clinical examples of paired with all key positions except 299 and 317 have been reported (Figure 1B and Figure S1A available online). Among 66 possible pairings of the 12 key positions, 3 (45%) have been reported to date (Figures 1B, 1C, and S1B). Further variations at the specific substitution level also occur, for example /F359C and / (Figure 1B) or E255K/F317L and E255V/F317I (Figure 1C). Significance In patients with Ph + leukemia, control of TKI resistance due to BCR-ABL1 single mutants is now achievable, but the ability of clinically available TKIs to target BCR-ABL1 compound mutants has yet to be thoroughly investigated. Results from this study reveal that BCR-ABL1 compound mutations impart various levels of TKI resistance, underscoring the need for definitive sequencing screens to distinguish these from polyclonal mutations and suggesting that optimal therapy selection for patients harboring compound mutations will improve disease control in Ph + leukemia. These findings may also apply to other malignancies in which compound mutations are a predicted route of therapy escape, such as acute myeloid leukemia and non-small cell lung cancer. Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 429

3 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A M244 F311 C Coiledcoil BCR SH3 SH2 ABL1 Kinase domain B M244V/ G25E/ Q252H/ Y253H/ E255K/ E255M/ E255V/ F311I/ /M351T /E355G /F359C / /L387M /H396R /D444G /E453K /E459K E255 T315 Imatinib M244 G25 Q252 Y253 E255 V299 F311 T315 F317 M351 F359 H396 M244V/F317L M244V F317L M244V/M351T M244V M351T F317 M244V/ M244V V299 M244V/E459K M244V F359 Y253 G25E/E275K G25E E275K G25 G25E/V299L G25E V299L Q252 M351 G25E/T315A G25E T315A G25E/F317L G25E F317L G25E/M351T G25E M351T G25E/D444G G25E G25E/E459K G25E Q252H/E255K Q252H E255K Y253H/E255K Y253H E255K H396 Y253H/F317L Y253H F317L Y253H/F359I Y253H F359I Y253H/ Y253H E255V/V299L E255V V299L E255V/T315A E255V T315A E255K/F317L E255K F317L E255V/F317I E255V F317I BCR ABL1 E255K/M351T E255K M351T Coiledcoil domain SH3 SH2 Kinase E255K/ E255K V268A/ V268A V299L/M351T V299L M351T V299L/F359I V299L F359I M244 G25 Q252 Y253 E255 V299 F311 T315 F317 M351 F359 H396 Reference(s) V299L/ V299L M244V 5, 6 V299L/L384M V299L L384M G25E 5, 6 V299L/M388L V299L M388L Q252H 9 V299L/E459K V299L Y253H 1, 2, 5, 8, 1, * F311I/F317L F311I F317L E255K 3, 4, 5, 6, 8, 9 F311L/H396R F311L H396R E255M 9 T315A/T319A T315A T319A E255V 4, 5, 7, * T315A/E355G T315A E355G F311I 2, 5, 6 T315F/E355G T315F E355G M351T 4, 5, 9 T315A/L364I T315A L364I E355G 7 T315A/L387M T315A L387M F359C 4, 5, 6, * F317L/M351T F317L M351T 4, 6, 7, 9 F317I/ F317I L387M 8 F317L/ F317L H396R 6, * F317I/L387M F317I L387M D444G 5 F317L/E45G F317L E453K * F317L/E459K F317L E459K 4, 5 M351T/H396R M351T H396R E459K D444G E459K E459K E45G E459K Reference(s) , 9, * 6, 8, 9 6, , 4, 6, 8 5, 1 6, , * , , , 8 6, 8, * 6, 9 1 * * 1 Figure 1. Clinically Reported BCR-ABL1 Compound Mutations Are Centered on 12 Key Positions (A) Crystal structure of the ABL1 kinase domain in complex with imatinib (PDB entry 2HYY). The 12 key positions accounting for most clinical BCR-ABL1 TKI resistance, including compound mutation-based resistance, are highlighted (orange; T315 is in red). The phosphate-binding (yellow) and activation loops (green) are indicated. (B and C) Key resistance positions (orange) in clinically reported -inclusive (B) and non- BCR-ABL1 compound mutants (C). Substitutions at non-key positions are in gray. 1, Shah et al., 7; 2,Khorashad et al., 8; 3,Stagno et al., 8; 4,Kim et al., 1; 5, Kim, D.W. et al., Blood 1 abstract 3443; 6, Khorashad et al., 13; 7, Smith, C.C., et al., Blood 11, abstract 3752; 8, Soverini et al., 13; 9,Cortes et al., 13; 1, Hochhaus et al., 13; *, the current study. See also Figure S1 and Table S1. Relevance for these key positions in TKI resistance is further supported by baseline conventional sequencing traces of 439 patients entering the phase 2 Ponatinib Ph + ALL and CML Evaluation (PACE) trial (Cortes et al., 13). Enrollment required: (1) resistance to or unacceptable toxicity from nilotinib or dasatinib, or (2) a documented baseline mutation. Mutations occurring in more than 1 patient were confined to 16 positions, including 11/12 key positions (all except Q252). In total, 95.4% (27/283) of the mutations observed in >1 patient among baseline PACE specimens occurred at key positions. For PACE end of treatment (EOT) specimens, 93.8% (15/16) of two-component compound mutations inferred from conventional sequencing involved two key positions (Table S1). Clinically Available Tyrosine Kinase Inhibitors Exhibit Differential Activity against BCR-ABL1 Compound Mutants Proliferation assays comparing six TKIs were performed with Ba/F3 cells expressing native BCR-ABL1, BCR-ABL1 single mutants at each of the 12 key positions, and clinically reported 43 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

4 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A B IC 5 [nm] C 1 1 Imatinib parental Nilotinib Dasatinib Ponatinib Rebastinib Bosutinib parental M244V G25E Q252H Y253H E255V V299L F311I I315M F317L M351T H396R Native Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Bosutinib IC 5 [nm] D 1 1 Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Bosutinib parental M244V/ G25E/ Q252H/ Y253H/ E255V/ F311I/ /M351T / /H396R /E453K Native IC 5 [nm] E 1 1 Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Bosutinib parental F317L/ V299L/ V299L/M351T V299L/F317L E255V/V299L Y253H/F317L Y253H/E255V G25E/V299L Native Native Single Mutants M244V G25E Q252H Y253H E255V V299L F311I T315M F317L M351T H396R Compound Mutants, -Inclusive M244V/ G25E/ Q252H/ Y253H/ E255V/ F311I/ /M351T / /H396R /E453K Native Y253H/ E255V/ F311I/ / pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 NT 5 nm 25 nm 5 nm 25 nm nm 5 nm nm 5 nm 5 nm 25 nm Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Native Y253H/E255V Y253H/F317L E255V/V299L V299L/F317L F317L/ pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 pbcr-abl1 BCR-ABL1 Compound Mutants, Non- G25E/V299L Y253H/E255V Y253H/F317L E255V/V299L V299L/F317L V299L/M351T V299L/ F317L/ NT 5 nm 25 nm 5 nm 25 nm nm 5 nm nm 5 nm 5 nm 25 nm Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Figure 2. Single and Compound BCR-ABL1 Mutants Exhibit Differential TKI Sensitivity in Ba/F3 Cells (A and B) Cell proliferation IC 5 values of TKIs against BCR-ABL1 single mutants (A), -inclusive (B, left) and non- (B, right) compound mutants. Mean IC 5 values are plotted. Blue lines denote average steady-state plasma concentration of each TKI in patients taking the standard daily dose: imatinib ( mg; 3,377 nm) (Peng et al., 4; Gozgit et al., Blood 13, abstract 3992), nilotinib ( mg twice daily; 2,754 nm) (Kantarjian et al., 6; Gozgit et al., Blood 13, abstract 3992), dasatinib ( mg; 27 nm) (Gozgit et al., Blood 13, abstract 3992; Talpaz et al., 6), ponatinib (15, 3, and 45 mg/day doses are represented; 35, 84, and 11 nm, respectively) (Cortes et al., 12; Gozgit et al., Blood 13, abstract 3992), rebastinib (15 mg twice daily; 35 nm) (Chan et al., 11; Eide et al., 11), and bosutinib (5 mg; 287 nm) (Gozgit et al., Blood 13, abstract 3992). (C) Heat map of TKI IC 5 s for single and compound mutants. I315M (see Figure 6) and /E453K (see Figure 5) are included for comparison. A color gradient from green (sensitive) to yellow (moderately resistant) to red (highly resistant) denotes the IC 5 sensitivity to each TKI: imatinib (green: <1, nm; yellow: 1, 4, nm; red: >4, nm); nilotinib (green: < nm; yellow: 1, nm; red: >1, nm); dasatinib (green: <25 nm; yellow: nm; red: >15 nm); ponatinib (green: <25 nm; yellow: nm; red: >15 nm); rebastinib (green: < nm; yellow: 1, nm; red: >1, nm); bosutinib (green: <15 nm; yellow: 15 1, nm; red: >1, nm). (D and E) BCR-ABL1 tyrosine phosphorylation (Y393) immunoblot analysis of -inclusive (D) and non- (E) compound mutants. NT, no treatment. See also Table S2. BCR-ABL1 compound mutants (Khorashad et al., 8; Kim et al., 1; Shah et al., 7; Stagno et al., 8). Except for I315M (see below), each single mutant was effectively inhibited by at least one TKI and exhibited a half-maximal inhibitory concentration (IC 5 ) value below the average steady-state plasma TKI concentration reported for patients receiving the standard drug dose (Figure 2). The six TKIs displayed partially overlapping resistance profiles for BCR-ABL1 single mutants, with inhibited only by ponatinib and rebastinib. -inclusive compound mutants were insensitive to all TKIs except ponatinib and rebastinib, which exhibited only marginal efficacy in most cases (Figures 2B and 2C). The most resistant mutant, E255V/ (IC 5 : nm), exhibited and 22.7-fold higher ponatinib resistance than E255V (IC 5 : 55.6 nm) or (IC 5 : 29.1 nm; Table S2), respectively. The IC 5 for E255V/ is >6.5-fold the average steady-state plasma concentration (11 nm) for patients receiving ponatinib at the PACE starting dose (45 mg/day; Cortes et al., 12). The Q252H/, /M351T, /, and /H396R mutants exhibited marginal ponatinib sensitivity (IC 5 : nm) and high-level rebastinib resistance (IC 5 : nm). M244V/ was the only -inclusive compound Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 431

5 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia mutant in the panel predicted to be sensitive to ponatinib and rebastinib at clinically achievable levels (Figure 2; Table S2). In vitro sensitivity of -inclusive mutants was correlated to the degree of BCR-ABL1 Y393 phosphorylation (a marker of kinase activity) by immunoblot analysis (Figure 2D). All non- compound mutants analyzed were inhibited by one or more TKIs (Figure 2B). For some compound mutants (e.g., Y253H/F317L), several TKI options exist. For others, a single TKI stands out as the leading choice, most notably dasatinib for Y253H/E255V (Figures 2B and 2C; Table S2). Superiority of dasatinib compared to ponatinib (which demonstrate similar low nanomolar IC 5 s against native BCR-ABL1) against this particular compound mutant was further confirmed by immunoblot analysis, which revealed substantial residual pbcr-abl1 signal for ponatinib compared to dasatinib at nm (Figure 2E). All other non- compound mutants were effectively suppressed by ponatinib at concentrations below the steady-state plasma concentrations of 45 mg/day (11 nm) and 3 mg/day (84 nm) doses. In addition to Y253H/E255V, only E255V/V299L (IC 5 : 42.8 nm) and F317L/ (IC 5 : 53.2 nm) were not inhibited at the steady-state plasma concentration for the 15 mg/day (35 nm) dose. It is conceivable that ponatinib doses lower than 15 mg/day may not be able to prevent the emergence of additional compound mutants. In summary, non- BCR- ABL1 compound mutants exhibited a spectrum of TKI sensitivities, suggesting in vitro resistance profiles may serve as a guide for clinical TKI selection. Computational Modeling of Y253H/E255V Rationalizes Differential Tyrosine Kinase Inhibitor Sensitivity Ponatinib binds to the ABL1 kinase domain in the DFG-out mode, recognizing an inactive conformation of the kinase (O Hare et al., 9; Zhou et al., 11). The binding site of ponatinib is centered on the adenine pocket of the enzyme and extends from the phosphate-binding loop (P loop) to the C-helix region. By contrast, dasatinib binding is accompanied by fewer conformational constraints and is less dependent on direct P loop and C-helix interactions (Tokarski et al., 6). Since ponatinib compared favorably with dasatinib against all non- compound mutants except Y253H/E255V, we investigated structural features that account for the striking difference in the case of this compound mutant (ponatinib IC 5 : 3.5 nm; dasatinib IC 5 : 18.1 nm; Table S2). Molecular dynamics simulations were carried out for a protracted interval ( ns), and docking simulations were performed using the GlideXP method (Suite 12: GlideXP, version 5.8, Schrödinger, New York, NY, 12) on a collection of 5 Y253H/E255V conformations extracted at regular intervals. Introduction of Y253H and E255V noticeably shifted the P loop, impinging on the ponatinib binding site (Figure 3A). Loss of Y253-F382 aromatic p-p stacking also pushed F382 into the ponatinib site (Figures 3B and S2), and disruption of the critical K271-E286 salt bridge in the inactive conformation repositioned residues L248, K271, E286, and R362 (Figures 3B and S2). In contrast, modeling predicted that dasatinib forms a new hydrogen bond with H253 in the Y253H/E255V mutant and that realignments relative to native BCR-ABL1 do not obstruct dasatinib binding (Figures 3C and 3D). Thus, in vitro experimental results and computational modeling (Figures 2B, 2C, and 3E) identify dasatinib as the only TKI that retains potency against Y253H/E255V at clinically relevant levels. Conventional and Clonal Sequencing Establish Correlations between Baseline Mutations and Response to Ponatinib To understand the role of compound mutations for ponatinib response and resistance, we received and analyzed specimens from 64 patients treated on the PACE trial (n = 5) or ponatinib expanded access program (n = 14), using both conventional Sanger sequencing and clonal sequencing of an average of 85 individual BCR-ABL1 kinase domain amplicons per specimen. Clinical specimens originated from patients enrolled at centers participating in the PACE trial that elected to participate in an investigator-initiated companion protocol. The cloning and sequencing approach is an order of magnitude more sensitive and differentiates compound from polyclonal mutations, allowing greater insight into the role of compound mutations in TKI resistance. Pre-ponatinib baseline samples were evaluated for all patients; for 3 patients, longitudinal and/or EOT specimens were also analyzed. Patients were grouped according to baseline mutation status assessed by conventional sequencing: (1), (2) mutation other than, or (3) no mutation. There were 31 patients that were in the chronic phase (CML-CP), 14 in the accelerated phase (CML-AP), and 19 in CML-BP or with Ph + ALL. The cohort was heavily pretreated: 31 patients (48%) had been exposed to two TKIs and 29 (45%) to three or more TKIs. Prior TKI exposure, baseline mutation status, response, and outcome are summarized in Table 1. Patients with a Baseline Mutation was detected at baseline in 22/64 patients (34.4%), including eight CML-CP, six CML-AP, five CML-BP, and three Ph + ALL patients (Table S3). There were three patients that carried a second baseline mutation: K285E (#2), F317L (#1), or H396R (#18). Patients with a Mutation Other Than at Baseline Non- baseline mutations were found in 17/64 patients (26.6%), representing all non- key positions except 244 and 311: nine CML-CP, three CML-AP, two CML-BP, and three Ph + ALL (Table S4). Most baseline samples (11/17; 64.7%) harbored a mutation at a single position; six had mutations at two positions: F317L; E45G (#23), F317L; E459K (#27), E255V; F317L (#32), F317I; (#34), Y253H; E255V (#35), and G25E; F317L (#39). Patients with No Baseline BCR-ABL1 Mutation Lack of a baseline mutation was observed in 25/64 patients (39.1%): 14 CML-CP, five CML-AP, four CML-BP, and two Ph + ALL (Table S5). No patient lacking a baseline mutation who discontinued ponatinib harbored a compound mutation at EOT, suggesting a degree of BCR-ABL1-independent resistance prior to initiating ponatinib therapy as well as at ponatinib failure. In the following, we evaluated outcomes on ponatinib therapy for patients carrying a baseline or non- mutation. -Inclusive Compound Mutations Are Associated with Ponatinib Failure Our in vitro profiling of T315-inclusive compound mutants predicts that most pairings with a second key position will confer moderate- to high-level ponatinib resistance (Figure 2). 432 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

6 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A Y253H/E255V Native B K271 Y253H/E255V Native V255 H253 E255 Y253 C-helix L248 E286 Ponatinib F382 Ponatinib H361 R362 P-loop C Y253H/E255V D E T315 M318 1 Ponatinib Dasatinib F317 M29 8 Dasatinib H-bond H253 V255 K271 Number of conformations Docking score (kcal/mol) IC 5 Fold/Native Ponatinib Dasatinib Figure 3. Molecular Dynamics Simulations of ABL1 Y253H/E255V Provide a Structural Explanation for the Ineffectiveness of Ponatinib Compared to Dasatinib (A) Structural alignment of native ABL1 kinase (PDB entry 3IK3 bound to ponatinib; orange) with the ABL1 Y253H/E255V mutant kinase (blue). Ponatinib is shown in wireframe. Positions within 4 Å of ponatinib were designated for structural alignment using only the C a atoms. Native positions Y253 and E255 (green) and mutated positions H253 and V255 (red) are shown as spheres. The region of the backbone (P loop, C-helix, and DFG loop) that showed major conformational change is highlighted in either orange (native) or blue (Y253H/E255V). (B) Positions of selected residues in the ponatinib site of native ABL1 (orange) and ABL1 Y253H/E255V (cyan). (C) Dasatinib was docked to a representative model of the ABL1 Y253H/E255V mutant structure using molecular dynamics simulation. Key residues involved in binding of dasatinib are shown (cyan). The Y253H substitution establishes a new hydrogen bond with dasatinib, in addition to the existing T315-dasatinib hydrogen bond (red dashed lines). The chemical structure of dasatinib is shown using a space-filling wireframe representation. (D) Computational docking of dasatinib (black) and ponatinib (gray) into the mutant ABL1 Y253H/E255V kinase domain (Suite 12: GlideXP, Schrödinger, New York, NY, 12). (E) Fold-change in cellular IC 5 for BCR-ABL1 Y253H/E255V relative to native BCR-ABL1 for dasatinib (black) and ponatinib (gray). Error bars represent ± SEM. See also Figure S2. Accordingly, we observed three patients who discontinued ponatinib due to marked expansion of a -inclusive compound mutation. Patient #38 (Tables S4 and S6) presented with Ph + ALL previously refractory to imatinib and dasatinib. Cloning and sequencing (n = 84 clones) confirmed a predominant E255V mutation (85% of clones), including as an E255V/ compound mutation (17% of clones). Transient response to ponatinib was followed by rapid hematologic relapse and a detection of a dominant E255V/ compound mutation (69% of clones; Figure 4A). Molecular dynamics simulations traced the reduced affinity of ponatinib toward E255V/ compared to alone to reorientation of the P loop and C-helix necessary to accommodate the hydrophobic V255 side chain (Figure 4B). These adjustments force the L248 and I315 side chains into the ponatinib site, repositioning residues M29, F359, and D381 and reducing the distance between F382 and I315, which narrows the channel into which ponatinib normally binds (Figures 4C and S2). A second patient (#36, CML-BP; Table S4) had a baseline F359C mutation and later experienced disease progression attributable to a /F359C mutation that was not detectable in the baseline clonal sequencing profile (Figures 5A and 5B; Table S7). This compound mutant was recovered in Ba/F3 BCR-ABL cell-based resistance screens for ponatinib (O Hare et al., 9) and rebastinib (Eide et al., 11), in line with our in vitro profiles implicating mutant pairing of these two positions in moderate and high-level resistance to these TKIs, respectively. Last, a CML-AP patient (#12; Tables S3 and S7) with a baseline mutation treated with ponatinib (45 mg/day) experienced disease progression with a /E453K mutation (9% of clones) not detected at baseline by conventional sequencing or cloning and sequencing (Figure 5C). The E453K mutation has been reported in imatinib resistance (Soverini et al., 13), but not compound mutation-based resistance. Ba/F3 BCR- ABL1 /E453K cells showed a substantially higher level of Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 433

7 434 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. Table 1. Patient Characteristics: Prior TKI Exposure, Baseline Mutation Status, Response, and Outcome CML-CP CML-AP CML-BP/Ph + ALL Mutation Other than Mutation Other than Mutation Other than No No Total Mutation Mutation Total Mutation Mutation Total Mutation n=31 n=8 n=9 n=14 n=14 n=6 n=3 n=5 n=19 n=8 n=5 n=6 No Mutation Prior TKI exposure, n (%) 1 TKI 3 (1) 2 (25) () 1 (7) () () () () 1 (5) 1 (13) () () 2 TKIs 12 (39) 5 (63) 5 (56) 2 (14) 7 (5) 3 (5) 1 (33) 3 () 12 (63) 6 (75) 4 () 2 (33) R3 TKIs 16 (52) 1 (13) 4 (44) 11 (79) 7 (5) 3 (5) 2 (67) 2 () 6 (32) 1 (13) 1 () 4 (67) Best hematologic response on ponatinib, n (%) a <MaHR 1 (3) () () 1 (7) () () () () 3 (16) () 1 () 2 (33) MaHR 1 (3) 1 (13) () () 1 (7) () 1 (33) () 5 (26) 2 (25) 1 () 2 (33) CHR 29 (94) 7 (89) 9 () 13 (93) 13 (93) 6 () 2 (67) 5 () 11 (58) 6 (75) 3 () 2 (33) Best cytogenetic response on ponatinib, n (%) b <pcyr 12 (39) 2 (25) 5 (56) 5 (36) 6 (43) 2 (33) 1 (33) 3 () 8 (42) 1 (13) 2 () 5 (83) pcyr 4 (13) 1 (13) () 3 (21) 3 (21) 1 (17) 1 (33) 1 () 5 (26) 4 (5) 1 () () CCyR 15 (48) 5 (63) 4 (44) 6 (43) 5 (36) 3 (5) 1 (33) 1 () 6 (32) 3 (38) 2 () 1 (17) Best molecular response on ponatinib, n (%) c <MMR (65) 4 (5) 6 (67) 1 (71) 1 (71) 4 (67) 2 (67) 4 () 18 (95) 8 () 4 () 6 () RMMR 11 (35) 4 (5) 3 (33) 4 (29) 4 (29) 2 (33) 1 (33) 1 (2) 1 (5) () 1 () () Treatment follow-up Median duration of ponatinib treatment, months (range) 13.6 ( ) 11.1 ( ) 15.1 ( ) 13.6 ( ) 17.1 ( ) 19.3 ( ) ) 13.8 ( ) Remain on ponatinib therapy, n (%) 14 (45) 5 (63) 4 (44) 5 (36) 6 (43) 2 (33) 2 (67) 2 () () () () () Discontinued, n (%) 17 (55) 3 (38) 5 (56) 9 (64) 8 (57) 4 (67) 1 (33) 3 () 19 () 8 () 5 () 6 () Outcome follow-up, n (%) Alive at last follow-up 28 (9) 6 (75) 8 (89) 14 () 8 (57) 3 (5) 2 (67) 3 () 7 (37) 3 (38) 1 () 3 (5) Deceased at last follow-up 3 (1) 2 (25) 1 (11) () 6 (43) 3 (5) 1 (33) 2 () 12 (63) 5 (63) 4 () 3 (5) BCR-ABL1 cloning and sequencing, n (%) Baseline samples analyzed 31 () 8 () 9 () 14 () 13 (93) 6 () 2 (67) 5 () 19 () 8 () 5 () 6 () Longitudinal samples analyzed 7 (23) 2 (25) 4 (44) 1 (7) 4 (29) 1 (17) 2 (67) 1 () 3 (16) 1 (13) 1 () 1 (17) End of treatment samples analyzed 4 (13) () 1 (11) 3 (21) 5 (36) 3 (5) () 2 () 9 (47) 4 (5) 2 () 3 (5) Compound mutations emergent/persistent () () () () 1 (7) 1 (17) () () 6 (32) 3 (38) 3 () () in failure See also Tables S3 S5. a MaHR, Major hematologic response; CHR, complete hematologic response. b pcyr, partial cytogenetic response; CCyR, complete cytogenetic response. c MMR, major molecular response. 3.7 ( ) 3.6 ( ) 5.6 ( ) 2.9 ( ) Compound Mutants Cause TKI Failure in Ph + Leukemia Cancer Cell

8 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A Patient 38 Prior TKIs: I, D Feb 11 Pre-ponatinib: Ph + ALL Baseline conventional seq: E255V (7%) Pre-ponatinib: % of sequenced clones (n=84) sample 11-8 B I315 E255 V255 E255V/ C-helix Mar 11 EOT conventional seq: E255V; (%; 75%) Key EOT: % of sequenced clones (n=98) None Q252H/ E255V E255V/ E255V V299L E255V/ E255V/ F317L sample P-loop C L248 I315 M29 Ponatinib E255V/ F359 EOT due to progression Key None Q252H/ E255V E255V/ E255V V299L Key Mutation E255V/ Key + Additional E255V/ F317L F382 Ponatinib D381 Figure 4. BCR-ABL1 E255V/ Exhibits Severely Compromised TKI Binding and Is Detected at Clinical Relapse on Ponatinib (A) Summary of BCR-ABL1 cloning and sequencing for patient #38 (Ph + ALL) at baseline (upper) and EOT (lower) by cloning and sequencing individual amplicons. Conventional sequencing results and prior TKIs are shown. Light red bars represent mutations detected at key positions; dark red fractions of bars indicate additional mutations at other positions. For bars designated None, light and dark red fractions represent native BCR-ABL1 and all mutations not occurring at key positions, respectively. In this patient, a dominant, pan-tki resistant E255V/ mutant was detected at the time of ponatinib failure. (B) Molecular dynamics comparison of E255V/ (blue; V255 in red) and mutants (orange; E255 in green). (C) Structural realignments in E255V/ (cyan) compared to the mutant (orange). See also Figure S3 and Table S6. ponatinib resistance (IC 5 : 93.4 nm) relative to those expressing the mutant (Figure 5D) and were insensitive to all other TKIs tested except rebastinib (IC 5 : nm) (Figures 2B and 2C). Both ponatinib and rebastinib were effective only at clinically unachievable concentrations (Figure 2B). Among these three examples, the -inclusive EOT mutation was also detectable at baseline in only one case (E255V/), suggesting the mutation was acquired on therapy or was below the detection limit of cloning and sequencing in the other two cases. There were two additional patients in our study (#17 and #18; Tables S3 and S6) that had ponatinib-resistant EOT mutations that already predominated at baseline (Y253H/ and /H396R, respectively; Figure S3). Altogether, these findings suggest that -inclusive compound mutations significantly impair ponatinib binding and typically lead to clinical resistance and relapse. The I315M Mutation Emanates from and Confers High-Level Ponatinib Resistance Nearly every instance of BCR-ABL1 kinase domain mutationbased ponatinib failure was attributable to a compound mutation pairing two key positions. However, in the case of a Ph + ALL patient (#22; Figure 6A; Tables S3 and S8) with a baseline mutation who achieved a complete cytogenetic response (CCyR) on ponatinib (45 mg/day), but progressed after 7 months, longitudinal and EOT cloning and sequencing revealed a change of I315 to methionine (I315M) through a single nucleotide change (ATT to ATG). We previously recovered the ponatinib-resistant I315M mutant in Ba/F3 BCR-ABL1 cell-based resistance screens (Eide et al., 11; O Hare et al., 9), and in vitro profiling of Ba/F3 BCR-ABL1 I315M cells confirmed pan-tki resistance. The level of ponatinib resistance conferred by the I315M mutation (IC 5 : nm; Figure 6B) exceeded all tested single and compound mutants except E255V/ (IC 5 : nm). Molecular dynamics simulations demonstrated direct encroachment of the methionine residue on the ponatinib site (Figures 6C and 6D) and that adjustments at positions 269, 29, 317, 359, and 381 also disfavor ponatinib binding (Figures 6D, S2) and disrupt the hydrophobic spine architecture (Azam et al., 8). These findings illustrate that I315M as a single point mutation can lead to ponatinib treatment failure. Non- Compound Mutations Impart Differential Levels of Tyrosine Kinase Inhibitor Resistance In vitro evaluation of non- compound mutants showed varying levels of TKI sensitivity across the panel, with 7/8 mutants demonstrating sensitivity to ponatinib (Figure 2C). Among patients for whom EOT samples were available, only Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 435

9 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A Patient 36 Prior TKIs: D, N C Patient 12 Prior TKIs: I, D, N Feb 11 Pre-ponatinib: CML-BP Baseline conventional seq: F359C (%) Pre-ponatinib: %of sequenced clones (n=) sample 12-3A May 11 Pre-ponatinib: CML-AP Baseline conventional seq: (9%) Pre-ponatinib: % of sequenced clones (n=99) sample A May 11 EOT conventional seq: ; F359C (%; %) EOT due to progression B Ponatinib IC 5 Fold/Native Key EOT: % of sequenced clones (n=87) Key 3 1 None None Key Mutation / G25R/ / F359C/ H396Y G25R/ / F359C/ H396Y V299M/ F359C V299M/ F359C F311S/ F359C F311S/ F359C / F359C / F359C Key+ Additional D / F359C/ H396Y sample 12-3B / F359C/ H396Y Ponatinib IC 5 Fold/Native 3 1 F359C F359C /E453K May 12 Longitudinal conventional seq: ; E453K (%; %) Relapse and progression Jun 12 EOT conventional seq: ; E453K (%; 85%) EOT due to progression Key None Longitudinal: % of sequenced clones (n=) Key EOT: % of sequenced clones (n=99) Key Key Mutation None None Prominent Non-Key Mutation / E453K sample B / E453K sample / E453K E453K E453K E453K Key + Additional Prominent Non-Key Mutation + Additional Figure 5. Clinical Emergence of -Inclusive Compound Mutants Can Lead to Relapse (A) Summary of BCR-ABL1 cloning and sequencing of individual clones for patient #36 (CML-BP) at baseline (upper) and EOT (lower). Conventional sequencing results and prior TKIs are shown. Bar graphs are as in Figure 4. (B) Fold-change in ponatinib Ba/F3 cellular IC 5 values for BCR-ABL1 and BCR-ABL1 / compared to native BCR-ABL1. (C) Summary of BCR-ABL1 cloning and sequencing for patient #12 (CML-AP). Samples were analyzed as in (A) at baseline (upper), while on ponatinib (middle), and at EOT (lower). (D) Fold-change in ponatinib Ba/F3 cellular IC 5 values for BCR-ABL1 and BCR-ABL1 /E453K compared to native BCR-ABL1. All error bars represent ± SEM. See also Table S7. one demonstrated clear evidence of a non- compound mutation at failure (#37, Ph + ALL; Figure 7A; Tables S4 and S9). Following treatment with imatinib and dasatinib, this patient exhibited a baseline F317I mutation. The patient experienced disease progression and discontinued ponatinib (45 mg/day), with EOT sequencing revealing an E255V/F317I mutation (% of clones; Figure 7A). In contrast, three patients with non- compound mutations at baseline achieved durable responses on ponatinib. Patient #34 (Figure 7B; Tables S4 and S9) presented with CML-AP and was treated with imatinib and dasatinib prior to starting ponatinib. There were two mutations (F317I and ) identified in the baseline sample that were confirmed as a predominant F317I/ compound mutation by cloning and sequencing (Figure 7B). The patient achieved rapid complete hematologic response (CHR) and a major molecular response (MMR) within 7 months on ponatinib, at which time the majority (87%) of sequenced clones remained F317I/. At last follow-up of 16 months, the patient continued to maintain MMR. There were two additional patients (#23 and #27; Tables S4, S9) with non- compound mutations predominant at 436 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

10 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A Patient 22 Prior TKIs: I Apr 11 Pre-ponatinib: Ph + ALL Baseline conventional seq: (%) Pre-ponatinib: % of sequenced clones (n=97) Key None E255K E255K/ E255K/ E255V/ E255V/ / I315M I315M F317S sample I315M Fold/Native Ponatinib IC 5 B 15 5 I315M C M315 I315M C-helix Jun 11 Longitudinal: Longitudinal conventional seq: (%) %ofsequenced clones (n=83) sample A P-loop Ponatinib Responding Key None E255K E255K/ E255K/ E255V/ E255V/ / I315M I315M F317S I315M Nov 11 Longitudinal conventional seq: ; I315M (3%; 7%) Relapse and progression Nov 11 EOT conventional seq: ; I315M (25%; 75%) EOT due to progression Longitudinal: % of sequenced clones (n=) Key None E255K E255K/ E255K/ E255V/ E255V/ / I315M I315M F317S EOT: %ofsequenced clones (n=98) Key None E255K E255K/ E255K/ E255V/ E255V/ / I315M I315M F317S sample C I315M sample I315M D F317 M315 Ponatinib A269 I315 D381 M29 I293 I315M F359 Key Mutation Key + Additional Figure 6. Acquired I315M Point Mutation-Based Ponatinib Failure in a Patient Harboring at Onset of Therapy (A) Summary of BCR-ABL1 cloning and sequencing for patient #22 (Ph + ALL) at baseline (upper), longitudinally on ponatinib therapy (middle), and at EOT (lower) by cloning and sequencing individual clones. Conventional sequencing results and prior TKIs are shown. Bar graphs are as in Figure 4. (B) Fold-change in ponatinib Ba/F3 cellular IC 5 values for BCR-ABL1 I315M and BCR-ABL1 compared to native BCR-ABL1. Error bars represent ± SEM. (C) Ponatinib (in wireframe) in complex with ABL1 (orange); ABL1 I315M (blue) is superimposed. Structural alignment of the and I315M mutants was performed as in Figure 3. (D) M315 penetrates deeply into the ponatinib site, and structural adjustments at positions 269, 29, 293, 317, 359, and 381 also disfavor ponatinib binding. See also Figure S4 and Table S8. baseline that showed marked reduction in the abundance of the compound mutant clones on ponatinib therapy. After failing four successive TKIs, patient #23 exhibited an F317L/E45G baseline compound mutation (83% of clones; Figure 7C). The patient achieved a CHR and stable disease on ponatinib for over 2 years, discontinuing for undisclosed reasons. EOT sequencing showed no F317L/E45G (Figure 7C), suggesting this mutant is resistant to previous TKIs, but sensitive to ponatinib. Similarly, CML-CP patient #27 (Tables S4 and S9) exhibited an F317L/E459K baseline mutation (98% of clones; Figure 7D) following failure of imatinib and dasatinib. After a year on ponatinib and achievement of CHR, but neither CCyR nor MMR, F317L/E459K was reduced to a minor component (6% of clones), suggesting sensitivity to ponatinib. Compound mutants pairing E459K with M244V, G25E, V299L (Kim et al., 9), and have been reported (Figures 1B and 1C); the current study reports F317L/E459K as a confirmed clinical compound mutation. In summary, -inclusive compound mutations almost uniformly confer high-level resistance to all clinically available TKIs including ponatinib, while a fraction of non- compound mutants remain sensitive to one or more TKIs. DISCUSSION Drug-resistant compound mutations within the BCR-ABL1 kinase domain are an emerging clinical problem for patients receiving sequential TKI therapy. As we predicted for ponatinib and rebastinib, some of these mutations confer resistance that is several-fold higher than that of either contributing mutation in isolation (Eide et al., 11; O Hare et al., 9). We investigated the role of BCR-ABL1 compound mutations in TKI Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 437

11 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia A Patient 37 Prior TKIs: I, D B Patient 34 Prior TKIs: I, D Jan 11 Pre-ponatinib: Ph + ALL Baseline conventional seq: F317I (%) Pre-ponatinib: %of sequenced clones (n=) sample 11-3 Sep 12 Pre-ponatinib: CML-AP Baseline conventional seq: F317I; (%; %) Pre-ponatinib: %of sequenced clones (n=98) sample Jul 11 EOT conventional seq: E255V; F317I (%; %) Key E255V/ E255V/ V299M/ F317I F317I EOT: %of sequenced clones (n=95) V299L V299L/ F317I F317I sample 11-2 Apr 13 Longitudinal conventional seq: F317I; (%; %) Key None Longitudinal: % of sequenced clones (n=23) V299A/ F317I/ F317I/ sample C EOT due to progression Patient 23 Prior TKIs: I, D, N, B Key E255V/ E255V/ V299M/ F317I F317I Key Mutation V299L V299L/ F317I Key + Additional F317I D MMR Patient 27 Prior TKIs: I, D Key None Key Mutation V299A/ F317I/ F317I/ Key + Additional Oct 1 Pre-ponatinib: CML-CP Baseline conventional seq: F317L; E45G (9%, 9%) Pre-ponatinib: % of sequenced clones (n=95) sample Jan 11 Pre-ponatinib: CML-CP Baseline conventional seq: F317L; E459K (%; %) Pre-ponatinib: % of sequenced clones (n=99) sample B Key None F311S/ F317L/ E45G F317L F317S F317L/ F317L/ E45G E45G Key None F317L F317L/ E459K Apr 13 EOT conventional seq: No mutations EOT: %of sequenced clones (n=93) sample 14-5 Dec 11 Longitudinal conventional seq: F317L (%) Longitudinal: % of sequenced clones (n=99) sample C Undisclosed reasons Key None F311S/ F317L/ E45G Key Mutation Prominent Non-Key Mutation F317L F317S Key + Additional F317L/ F317L/ E45G Prominent Non-Key Mutation + Additional E45G Stable disease Key Key Mutation Prominent Non-Key Mutation None F317L F317L/ E459K Key + Additional Prominent Non-Key Mutation + Additional Figure 7. Certain Non- Compound Mutants Lead to Clinical Relapse, while Others Are Susceptible to Ponatinib (A) Summary of BCR-ABL1 cloning and sequencing for patient #37 (Ph + ALL) at baseline (upper) and EOT (lower) by cloning and sequencing individual amplicons. Conventional sequencing results and prior TKIs are shown. Bar graphs are as in Figure 4. (B) Summary of BCR-ABL1 cloning and sequencing for patient #34 (CML-AP) at baseline (upper) and while responding to ponatinib (lower). (C) Summary of BCR-ABL1 cloning and sequencing for patient #23 (CML-CP) at baseline (upper) and EOT (lower). (D) Summary of BCR-ABL1 cloning and sequencing for patient #27 (CML-CP) at baseline (upper) and while responding to ponatinib (lower). See also Table S Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

12 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia resistance, focusing on ponatinib due to its unique effectiveness against the single mutant and clinical availability. In the United States, ponatinib is approved for patients with refractory CML or Ph + ALL harboring a mutation or for whom no other TKI is indicated, based on results of the PACE trial demonstrating significant activity at a median follow-up of 15 months (Cortes et al., 13). Despite the impressive efficacy of ponatinib, our findings indicate that compound mutations are an important route of therapy escape, and it is conceivable that dose reduction from 45 mg to 3 mg/day, as recommended for patients with a good response to ponatinib, may increase the emergence of drug-resistant compound mutants. Our in vitro studies predict that a 3 mg/day dose would maintain efficacy against 7/8 non- compound mutants tested in our panel. At a daily dose of 15 mg, ponatinib is predicted to preempt outgrowth of 5/8 non- compound mutants in our panel, with the Y253H/E255V, E255V/V299L, and F317L/ mutants remaining potentially problematic. In contrast, therapeutic utility is less promising with respect to -inclusive compound mutants, where 9/1 in our panel showed little or no sensitivity to ponatinib or any of the other TKIs tested. We provide examples of clinical ponatinib failure attributable to E255V/, /F359C, Y253H/, /H396R, and / E453K. Given the unique efficacy of ponatinib against the single mutant and its current revised U.S. clinical indication, a significant fraction of future patients treated with ponatinib will be expected to harbor a mutation at baseline. More sensitive, routine screening of baseline samples from these patients may be warranted to determine whether problematic -inclusive compound mutants are present. Detection of two mutations by conventional sequencing does not provide sufficient information to identify the best treatment option since this may represent two clones, each with a single mutation. In contrast, cloning and sequencing discerns compound from polyclonal mutations. For example, while patients #1, #26, and #35 each had two baseline mutations by conventional sequencing, cloning and sequencing demonstrated mutual exclusivity of these mutations at the clonal level. Verification that Y253H and E255V exist in different clones as opposed to as a highly ponatinib-resistant Y253H/E255V compound mutant (patient #35) is of importance for clinical decisionmaking. In our study, no patient beginning ponatinib with native BCR- ABL1 by conventional sequencing exhibited a causative compound mutation at EOT, similar to reports on BCR-ABL1 single mutants (Khorashad et al., 8; Soverini et al., 9). These results suggest that patients who failed multiple TKIs without a BCR-ABL1 mutation are unlikely to experience ponatinib failure due to emergence of a resistance-conferring compound mutation. Effective therapy for these patients may require a synthetic lethality approach, involving blockade of a second pathway in addition to BCR-ABL1. Additionally, consistent with inferred compound mutational status data reported in the PACE trial (Cortes et al., 13), we found detection of compound mutations at EOT to be more frequent among CML-BP and Ph + ALL patients than those with CML-CP, suggesting an increased risk of compound mutation-based ponatinib resistance in advanced disease. Although we identified a number of resistance-conferring compound mutants at EOT, longer follow-up and a larger number of specimens will be required to make definitive prognostic use of baseline sequencing profiles of patients beginning a new TKI. Among the 12 key positions identified, there appear to be pairing constraints for generation of TKI-resistant compound mutants. For example, the current snapshot of reported compound mutations includes position 315 in compound with 9/11 other key positions, whereas position 252 has only been reported in compound with positions 255 and 315. We also observed that among the 66 possible pairwise combinations of the 12 key residues, the spectrum and frequency of pairings reported to date appear to represent a nonrandom distribution (c2 = 42.39; df = 3; p <.1). While additional resistant pairings will undoubtedly be observed in the future, findings to date may suggest that only a limited number of compound mutation possibilities avoid deleterious, nontolerated effects on kinase function (Corbin et al., 2) or fitness of the mutated clone (Griswold et al., 6; Shah et al., 7; Skaggs et al., 6) and are consistent with our observation that the number of missense mutations tolerated by the kinase appears to be limited. The broad potency of ponatinib against BCR-ABL1 point mutants can be traced to the extensive network of contacts that stabilize its binding to the kinase domain. However, certain pairings of mutations, each of which is susceptible to a given TKI in isolation, confer increased resistance when present as a compound mutation. Molecular dynamics-guided modeling performed for Y253H/E255V, E255V/, and I315M reveals commonalities that could aid in designing TKIs to treat compound mutants. For instance, several compound mutations involving the P loop result in significant distortion of this region, suggesting it may prove advantageous to minimize direct TKI/P loop interactions. Also intriguing is our characterization of an I315M point mutation in clinical resistance to ponatinib due to direct encroachment of the mutant side chain on drug binding (patient #22). Notably, ponatinib is a poor inhibitor of kinases in which methionine is the native gatekeeper position analogous to BCR-ABL1 position 315. For example, the insulin receptor is 5-fold less sensitive to ponatinib than ABL1 (O Hare et al., 9). By contrast, the T315A mutant is uniquely resistant to dasatinib and inhibited by each of the other five TKIs including ponatinib (Burgess et al., 5; Shah et al., 4). These findings argue that efforts to develop future BCR-ABL1 TKIs should also consider the capacity to accommodate multiple different specific substitutions at the gatekeeper position. Computational methods have been applied to BCR-ABL1 single and compound mutants to predict TKI binding, including ponatinib (Gibbons et al., 14; Tanneeru and Guruprasad, 13). The use of different computational methods and the inherent limitations of modeling mandate experimental validation of predictions. For example, one computational approach identified / as 4-fold more resistant to ponatinib than Y253H/ (Gibbons et al., 14). In contrast, our comprehensive, direct experimental comparison of compound mutants and findings in cell-based resistance analyses identify Y253H/ as 3.5-fold more ponatinib resistant than / (O Hare et al., 9). It is impossible to predict exactly how many patients diagnosed with CML or Ph + ALL will develop resistance due to BCR-ABL1 compound mutations. Fortunately, most CML-CP Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 439

13 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia patients treated with TKIs upfront achieve and maintain profound responses and are at a low risk for acquiring compound mutations. In contrast, TKI failure remains common in CML-BP and Ph + ALL, and the incidence of compound mutations may increase with the number of successive TKI therapies (Shah et al., 7). Although patients with compound mutations represent only a minority of Ph + leukemias, they currently lack a targeted therapy option and their prognosis is poor (Cortes et al., 13), highlighting the need to identify therapeutic strategies that minimize mutational escape in Ph + leukemia. In addition, our findings are relevant to other cancers in which compound mutations are a predicted mechanism of therapy escape, including acute myeloid leukemia (Smith et al., 13) and nonsmall cell lung cancer (Awad et al., 13; Davare et al., 13). Development of therapeutic strategies to control and target compound mutation-based resistance in Ph + leukemia will also provide a blueprint for similar discovery in other cancers. EXPERIMENTAL PROCEDURES Inhibitors Inhibitors were prepared as 1. mm stock solutions in phosphate-buffered saline (imatinib) or DMSO and stored at C. Serial dilutions of stock solutions were carried out before each experiment. Cellular Proliferation Assays Ba/F3 BCR-ABL1-expressing cells were plated in 96-well plates ( cells/ well) and incubated in 2-fold escalating concentrations of dasatinib, ponatinib ( 768 nm), imatinib, nilotinib, rebastinib, or bosutinib ( 1,2 nm) for 72 hr. Proliferation was assessed by methanethiosulfonate-based viability assay (CellTiter 96 AQueous One; Promega). IC 5 values are reported as the mean of three independent experiments performed in quadruplicate. See also Supplemental Experimental Procedures. Isolation of Primary Ph + Leukemia Cells from Blood or Bone Marrow All patients were consented in accordance with the Declaration of Helsinki and the Belmont Report, and University of Utah Institutional Review Board approved all studies with human specimens. Mononuclear cells (MNCs) were isolated from primary patient peripheral blood or bone marrow specimens by Ficoll-separation. CD34 + cells were enriched by magnetic column separation using a CD34 human microbead kit and the POSSELDS program (AutoMACS; Miltenyi). Purity of the CD34 + fraction was determined to be >9% by fluorescence-activated cell sorting. If MNC yield was limiting (< cells), the RNA isolation described below was done with an aliquot of MNCs. See also Supplemental Experimental Procedures. Conventional and Clonal Sequencing of the BCR-ABL1 Kinase Domain RNA obtained from primary Ph + leukemia cell lysates (QIAGEN RNeasy Mini Kit) served as template for cdna synthesis (BioRad iscript cdna Synthesis Kit) as recommended by the manufacturer. Amplification of the BCR-ABL1 kinase domain was done by two-step PCR to exclude amplification of normal ABL1 (Khorashad et al., 6; Shah et al., 7). PCR products were electrophoresed on a 2% agarose gel to confirm amplification, purified (QIAquick PCR Purification Kit; QIAGEN), and subjected to (1) conventional Sanger sequencing in both directions using BigDye terminator chemistry on an ABI373 instrument (Khorashad et al., 6), and (2) cloning and sequencing of amplified fragments introduced into E. coli TOP1 cells (TOPO cloning system; Invitrogen) (Khorashad et al., 13). For cloning and sequencing, individual bacterial colonies (average: 85/specimen; range: 23 ), each carrying a recombinant plasmid with a single BCR-ABL1 kinase domain amplicon inserted, were subjected to BCR-ABL1 kinase domain amplification and Sanger sequenced in both directions (Beckman Coulter Genomics). DNA sequence analysis was done with Mutation Surveyor software (SoftGenetics) (O Hare et al., 9). Immunoblot Analysis of BCR-ABL1 Tyrosine Phosphorylation Ba/F3 cells expressing native or compound mutant BCR-ABL1 were cultured for 4 hr in standard medium alone or with escalating concentrations of TKI, followed by boiling for 1 min in SDS-PAGE loading buffer. Lysates were separated on 4% 15% Tris-glycine gels, transferred, and immunoblotted with antibodies for the BCR N terminus (392; Cell Signaling) and phospho-abl1 (Y393 [1a numbering]; Cell Signaling). Molecular Dynamics Simulations Mutant conformations of the ABL1 kinase were prepared using standard methods to generate ABL1 Y253H/E255V, ABL1 E255V/, and ABL1 I315M. For each mutant, both the active (Protein Data Bank [PDB] entry 2GQG; Tokarski et al., 6) and inactive (PDB entry 2HYY; Manley et al., 5) conformations of ABL1 kinase were created. The Nanoscale Molecular Dynamics simulation package was used for molecular dynamics simulation, and the Amber ff12sb force field was employed for standard protein parameters. See also Supplemental Experimental Procedures. Docking Simulations The Schrödinger suite of programs (Suite 12: Maestro, version 9.3) was used for docking studies. In the final 5 ns of the simulation, 5 conformations were extracted as docking receptors. Selected conformations were prepared using Protein Preparation Wizard. Ligands (ponatinib and dasatinib) were prepared (Suite 12: LigPrep, version 2.5) and initial docking simulation was performed using the GlideXP module (version 5.7) of the Schrödinger program. To enhance binding conformations and allow receptor flexibility, docked conformations were subjected to induced fit simulations. Docking scores were computed using the GlideXP module. See also Supplemental Experimental Procedures. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, four figures, and nine tables and can be found with this article online at AUTHOR CONTRIBUTIONS M.S.Z. and C.A.E. are co-first authors. M.S.Z. carried out all in vitro experiments, analyzed data, prepared display items, and assisted in writing the manuscript. C.A.E. analyzed data, organized and tabulated the sequencing data, prepared display items, and assisted in writing the manuscript. M.S.Z. and C.A.E. made conceptual contributions to the design of experiments. ACKNOWLEDGMENTS Oregon Health & Science University (OHSU) and B.J.D. have a financial interest in MolecularMD. Technology used in this research has been licensed to MolecularMD. This potential conflict of interest has been reviewed and managed by the OHSU Conflict of Interest in Research Committee and Integrity Oversight Council. OHSU also has clinical trial contracts with Novartis and Bristol-Myers Squibb (BMS) to pay for patient costs, nurse and data manager salaries, and institutional overhead. B.J.D. does not derive salary, nor does his laboratory receive funds, from these contracts. M.W.D. served on advisory boards and as a consultant for BMS, ARIAD, and Novartis and receives research funding from BMS, Celgene, Novartis, and Gilead. H.M.K. receives research funding from ARIAD. E.J.J. receives consultancy fees from ARIAD. S.S. is a consultant for BMS, Novartis, and ARIAD. The authors acknowledge the mentorship of Professor John M. Goldman ( ), Imperial College London, and dedicate this paper to his legacy. We thank ARIAD Pharmaceuticals for nonfinancial support and permission to perform an investigator-initiated companion study (P.I.: M.D.) to study mechanisms of resistance to ponatinib in patients treated on the PACE trial. We thank Qian Yu, David J. Anderson, and Ira L. Kraft for technical assistance. H.J.K. thanks the Georgia Cancer Coalition for a tissue bank-supporting grant. We acknowledge support in conjunction with grant P3 CA414 awarded to the Huntsman Cancer Institute (T.O.). T.O. is supported by the NIH/NCI (R1 CA178397). S.K.T is a recipient of 13 Research Training Award for Fellows 4 Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

14 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia from the American Society of Hematology. A.M.E. is currently a Fellow of the Leukemia & Lymphoma Society (59-12). This work was supported by Howard Hughes Medical Institute and NIH/NCI MERIT award R37CA65823 (B.J.D.). M.W.D. is supported by the NIH (HL , 5 P1 CA and R1 CA178397), was a Leukemia & Lymphoma Society (LLS) Scholar in Clinical Research (736-1), and is an investigator on LLS SCOR R.B. acknowledges funding from the University of Utah Department of Medicinal Chemistry, a computing allocation at the XSEDE supercomputers (award TG-CHE186), and the Director s Discretionary Program (Epigenetics), which used resources of the Argonne Leadership Computing Facility, supported by the Office of Science of the U.S. Department of Energy (DE-AC2-6CH11357). Received: March 1, 14 Revised: April 3, 14 Accepted: July 1, 14 Published: August 14, 14 REFERENCES Apperley, J.F. (7). Part I: mechanisms of resistance to imatinib in chronic myeloid leukaemia. Lancet Oncol. 8, Awad, M.M., Katayama, R., McTigue, M., Liu, W., Deng, Y.L., Brooun, A., Friboulet, L., Huang, D., Falk, M.D., Timofeevski, S., et al. (13). Acquired resistance to crizotinib from a mutation in CD74-ROS1. N. Engl. J. Med. 368, Azam, M., Latek, R.R., and Daley, G.Q. (3). Mechanisms of autoinhibition and STI-571/imatinib resistance revealed by mutagenesis of BCR-ABL. Cell 112, Azam, M., Seeliger, M.A., Gray, N.S., Kuriyan, J., and Daley, G.Q. (8). Activation of tyrosine kinases by mutation of the gatekeeper threonine. Nat. Struct. Mol. Biol. 15, Bradeen, H.A., Eide, C.A., O Hare, T., Johnson, K.J., Willis, S.G., Lee, F.Y., Druker, B.J., and Deininger, M.W. (6). Comparison of imatinib mesylate, dasatinib (BMS ), and nilotinib (AMN17) in an N-ethyl-N-nitrosourea (ENU)-based mutagenesis screen: high efficacy of drug combinations. Blood 18, Burgess, M.R., Skaggs, B.J., Shah, N.P., Lee, F.Y., and Sawyers, C.L. (5). Comparative analysis of two clinically active BCR-ABL kinase inhibitors reveals the role of conformation-specific binding in resistance. Proc. Natl. Acad. Sci. USA 12, Chan, W.W., Wise, S.C., Kaufman, M.D., Ahn, Y.M., Ensinger, C.L., Haack, T., Hood, M.M., Jones, J., Lord, J.W., Lu, W.P., et al. (11). Conformational control inhibition of the BCR-ABL1 tyrosine kinase, including the gatekeeper mutant, by the switch-control inhibitor DCC-36. Cancer Cell 19, Corbin, A.S., Buchdunger, E., Pascal, F., and Druker, B.J. (2). Analysis of the structural basis of specificity of inhibition of the Abl kinase by STI571. J. Biol. Chem. 277, Cortes, J.E., Kantarjian, H.M., Brümmendorf, T.H., Kim, D.W., Turkina, A.G., Shen, Z.X., Pasquini, R., Khoury, H.J., Arkin, S., Volkert, A., et al. (11). Safety and efficacy of bosutinib (SKI-6) in chronic phase Philadelphia chromosome-positive chronic myeloid leukemia patients with resistance or intolerance to imatinib. Blood 118, Cortes, J.E., Kantarjian, H., Shah, N.P., Bixby, D., Mauro, M.J., Flinn, I., O Hare, T., Hu, S., Narasimhan, N.I., Rivera, V.M., et al. (12). Ponatinib in refractory Philadelphia chromosome-positive leukemias. N. Engl. J. Med. 367, Cortes, J.E., Kim, D.W., Pinilla-Ibarz, J., le Coutre, P., Paquette, R., Chuah, C., Nicolini, F.E., Apperley, J.F., Khoury, H.J., Talpaz, M., et al.; PACE Investigators (13). A phase 2 trial of ponatinib in Philadelphia chromosome-positive leukemias. N. Engl. J. Med. 369, Dai, Y., Rahmani, M., Corey, S.J., Dent, P., and Grant, S. (4). A Bcr/Ablindependent, Lyn-dependent form of imatinib mesylate (STI-571) resistance is associated with altered expression of Bcl-2. J. Biol. Chem. 279, Davare, M.A., Saborowski, A., Eide, C.A., Tognon, C., Smith, R.L., Elferich, J., Agarwal, A., Tyner, J.W., Shinde, U.P., Lowe, S.W., and Druker, B.J. (13). Foretinib is a potent inhibitor of oncogenic ROS1 fusion proteins. Proc. Natl. Acad. Sci. USA 11, Donato, N.J., Wu, J.Y., Stapley, J., Gallick, G., Lin, H., Arlinghaus, R., and Talpaz, M. (3). BCR-ABL independence and LYN kinase overexpression in chronic myelogenous leukemia cells selected for resistance to STI571. Blood 11, Druker, B.J., Guilhot, F., O Brien, S.G., Gathmann, I., Kantarjian, H., Gattermann, N., Deininger, M.W., Silver, R.T., Goldman, J.M., Stone, R.M., et al.; IRIS Investigators (6). Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N. Engl. J. Med. 355, Eide, C.A., Adrian, L.T., Tyner, J.W., Mac Partlin, M., Anderson, D.J., Wise, S.C., Smith, B.D., Petillo, P.A., Flynn, D.L., Deininger, M.W., et al. (11). The ABL switch control inhibitor DCC-36 is active against the chronic myeloid leukemia mutant BCR-ABL and exhibits a narrow resistance profile. Cancer Res. 71, Gibbons, D.L., Pricl, S., Posocco, P., Laurini, E., Fermeglia, M., Sun, H., Talpaz, M., Donato, N., and Quintás-Cardama, A. (14). Molecular dynamics reveal BCR-ABL1 polymutants as a unique mechanism of resistance to PAN- BCR-ABL1 kinase inhibitor therapy. Proc. Natl. Acad. Sci. USA 111, Gorre, M.E., Mohammed, M., Ellwood, K., Hsu, N., Paquette, R., Rao, P.N., and Sawyers, C.L. (1). Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification. Science 293, Griswold, I.J., MacPartlin, M., Bumm, T., Goss, V.L., O Hare, T., Lee, K.A., Corbin, A.S., Stoffregen, E.P., Smith, C., Johnson, K., et al. (6). Kinase domain mutants of Bcr-Abl exhibit altered transformation potency, kinase activity, and substrate utilization, irrespective of sensitivity to imatinib. Mol. Cell. Biol. 26, Hochhaus, A., Kreil, S., Corbin, A.S., La Rosée, P., Müller, M.C., Lahaye, T., Hanfstein, B., Schoch, C., Cross, N.C., Berger, U., et al. (2). Molecular and chromosomal mechanisms of resistance to imatinib (STI571) therapy. Leukemia 16, Hochhaus, A., Saglio, G., Larson, R.A., Kim, D.W., Etienne, G., Rosti, G., De Souza, C., Kurokawa, M., Kalaycio, M.E., Hoenekopp, A., et al. (13). Nilotinib is associated with a reduced incidence of BCR-ABL mutations vs imatinib in patients with newly diagnosed chronic myeloid leukemia in chronic phase. Blood 121, Kantarjian, H., Giles, F., Wunderle, L., Bhalla, K., O Brien, S., Wassmann, B., Tanaka, C., Manley, P., Rae, P., Mietlowski, W., et al. (6). Nilotinib in imatinib-resistant CML and Philadelphia chromosome-positive ALL. N. Engl. J. Med. 354, Kastner, R., Zopf, A., Preuner, S., Pröll, J., Niklas, N., Foskett, P., Valent, P., Lion, T., and Gabriel, C. (14). Rapid identification of compound mutations in patients with Philadelphia-positive leukaemias by long-range next generation sequencing. Eur. J. Cancer 5, 793. Khorashad, J.S., Anand, M., Marin, D., Saunders, S., Al-Jabary, T., Iqbal, A., Margerison, S., Melo, J.V., Goldman, J.M., Apperley, J.F., and Kaeda, J. (6). The presence of a BCR-ABL mutant allele in CML does not always explain clinical resistance to imatinib. Leukemia, Khorashad, J.S., Milojkovic, D., Mehta, P., Anand, M., Ghorashian, S., Reid, A.G., De Melo, V., Babb, A., de Lavallade, H., Olavarria, E., et al. (8). In vivo kinetics of kinase domain mutations in CML patients treated with dasatinib after failing imatinib. Blood 111, Khorashad, J.S., Kelley, T.W., Szankasi, P., Mason, C.C., Soverini, S., Adrian, L.T., Eide, C.A., Zabriskie, M.S., Lange, T., Estrada, J.C., et al. (13). BCR- ABL1 compound mutations in tyrosine kinase inhibitor-resistant CML: frequency and clonal relationships. Blood 121, Kim, D., Kim, D.W., Cho, B.S., Goh, H.G., Kim, S.H., Kim, W.S., Lee, J., Kweon, I.Y., Park, S.H., Yoon, J.H., et al. (9). Structural modeling of V299L and E459K Bcr-Abl mutation, and sequential therapy of tyrosine kinase inhibitors for the compound mutations. Leuk. Res. 33, Kim, W.S., Kim, D., Kim, D.W., Kweon, I.Y., Kim, S.H., Goh, H.G., Park, S.H., and Lee, J. (1). Dynamic change of BCR-ABL kinase domain Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc. 441

15 Cancer Cell Compound Mutants Cause TKI Failure in Ph + Leukemia mutation in Korean chronic myeloid leukaemia patients during treatment with Abl tyrosine kinase inhibitors. Hematol. Oncol. 28, Manley, P.W., Cowan-Jacob, S.W., and Mestan, J. (5). Advances in the structural biology, design and clinical development of Bcr-Abl kinase inhibitors for the treatment of chronic myeloid leukaemia. Biochim. Biophys. Acta 1754, O Hare, T., Shakespeare, W.C., Zhu, X., Eide, C.A., Rivera, V.M., Wang, F., Adrian, L.T., Zhou, T., Huang, W.S., Xu, Q., et al. (9). AP24534, a pan- BCR-ABL inhibitor for chronic myeloid leukemia, potently inhibits the mutant and overcomes mutation-based resistance. Cancer Cell 16, O Hare, T., Zabriskie, M.S., Eiring, A.M., and Deininger, M.W. (12). Pushing the limits of targeted therapy in chronic myeloid leukaemia. Nat. Rev. Cancer 12, Peng, B., Hayes, M., Resta, D., Racine-Poon, A., Druker, B.J., Talpaz, M., Sawyers, C.L., Rosamilia, M., Ford, J., Lloyd, P., and Capdeville, R. (4). Pharmacokinetics and pharmacodynamics of imatinib in a phase I trial with chronic myeloid leukemia patients. J. Clin. Oncol. 22, Ray, A., Cowan-Jacob, S.W., Manley, P.W., Mestan, J., and Griffin, J.D. (7). Identification of BCR-ABL point mutations conferring resistance to the Abl kinase inhibitor AMN17 (nilotinib) by a random mutagenesis study. Blood 19, Redaelli, S., Piazza, R., Rostagno, R., Magistroni, V., Perini, P., Marega, M., Gambacorti-Passerini, C., and Boschelli, F. (9). Activity of bosutinib, dasatinib, and nilotinib against 18 imatinib-resistant BCR/ABL mutants. J. Clin. Oncol. 27, Senior, M. (14). FDA halts then allows sales of Ariad s leukemia medication. Nat. Biotechnol. 32, Shah, N.P., Tran, C., Lee, F.Y., Chen, P., Norris, D., and Sawyers, C.L. (4). Overriding imatinib resistance with a novel ABL kinase inhibitor. Science 35, Shah, N.P., Skaggs, B.J., Branford, S., Hughes, T.P., Nicoll, J.M., Paquette, R.L., and Sawyers, C.L. (7). Sequential ABL kinase inhibitor therapy selects for compound drug-resistant BCR-ABL mutations with altered oncogenic potency. J. Clin. Invest. 117, Skaggs, B.J., Gorre, M.E., Ryvkin, A., Burgess, M.R., Xie, Y., Han, Y., Komisopoulou, E., Brown, L.M., Loo, J.A., Landaw, E.M., et al. (6). Phosphorylation of the ATP-binding loop directs oncogenicity of drug-resistant BCR-ABL mutants. Proc. Natl. Acad. Sci. USA 13, Smith, C.C., Lasater, E.A., Zhu, X., Lin, K.C., Stewart, W.K., Damon, L.E., Salerno, S., and Shah, N.P. (13). Activity of ponatinib against clinically-relevant AC2-resistant kinase domain mutants of FLT3-ITD. Blood 121, Soverini, S., Gnani, A., Colarossi, S., Castagnetti, F., Abruzzese, E., Paolini, S., Merante, S., Orlandi, E., de Matteis, S., Gozzini, A., et al. (9). Philadelphiapositive patients who already harbor imatinib-resistant Bcr-Abl kinase domain mutations have a higher likelihood of developing additional mutations associated with resistance to second- or third-line tyrosine kinase inhibitors. Blood 114, Soverini, S., De Benedittis, C., Machova Polakova, K., Brouckova, A., Horner, D., Iacono, M., Castagnetti, F., Gugliotta, G., Palandri, F., Papayannidis, C., et al. (13). Unraveling the complexity of tyrosine kinase inhibitor-resistant populations by ultra-deep sequencing of the BCR-ABL kinase domain. Blood 122, Stagno, F., Stella, S., Berretta, S., Massimino, M., Antolino, A., Giustolisi, R., Messina, A., Di Raimondo, F., and Vigneri, P. (8). Sequential mutations causing resistance to both Imatinib Mesylate and Dasatinib in a chronic myeloid leukaemia patient progressing to lymphoid blast crisis. Leuk. Res. 32, Talpaz, M., Shah, N.P., Kantarjian, H., Donato, N., Nicoll, J., Paquette, R., Cortes, J., O Brien, S., Nicaise, C., Bleickardt, E., et al. (6). Dasatinib in imatinib-resistant Philadelphia chromosome-positive leukemias. N. Engl. J. Med. 354, Tanneeru, K., and Guruprasad, L. (13). Ponatinib is a pan-bcr-abl kinase inhibitor: MD simulations and SIE study. PLoS ONE 8, e Tokarski, J.S., Newitt, J.A., Chang, C.Y., Cheng, J.D., Wittekind, M., Kiefer, S.E., Kish, K., Lee, F.Y., Borzillerri, R., Lombardo, L.J., et al. (6). The structure of Dasatinib (BMS ) bound to activated ABL kinase domain elucidates its inhibitory activity against imatinib-resistant ABL mutants. Cancer Res. 66, Weisberg, E., Manley, P.W., Breitenstein, W., Brüggen, J., Cowan-Jacob, S.W., Ray, A., Huntly, B., Fabbro, D., Fendrich, G., Hall-Meyers, E., et al. (5). Characterization of AMN17, a selective inhibitor of native and mutant Bcr-Abl. Cancer Cell 7, Zhou, T., Commodore, L., Huang, W.S., Wang, Y., Thomas, M., Keats, J., Xu, Q., Rivera, V.M., Shakespeare, W.C., Clackson, T., et al. (11). Structural mechanism of the Pan-BCR-ABL inhibitor ponatinib (AP24534): lessons for overcoming kinase inhibitor resistance. Chem. Biol. Drug Des. 77, Cancer Cell 26, , September 8, 14 ª14 Elsevier Inc.

16 Cancer Cell, Volume 26 Supplemental Information BCR-ABL1 Compound Mutations Combining Key Kinase Domain Positions Confer Clinical Resistance to Ponatinib in Ph Chromosome-Positive Leukemia Matthew S. Zabriskie, Christopher A. Eide, Srinivas K. Tantravahi, Nadeem A. Vellore, Johanna Estrada, Franck E. Nicolini, Hanna J. Khoury, Richard A. Larson,8 Marina Konopleva,9 Jorge E. Cortes, Hagop Kantarjian, Elias J. Jabbour, Steven M. Kornblau, Jeffrey H. Lipton, Delphine Rea, Leif Stenke, Gisela Barbany, Thoralf Lange, Juan-Carlos Hernández-Boluda, Gert J. Ossenkoppele, Richard D. Press, Charles Chuah, Stuart L. Goldberg, Meir Wetzler, Francois-Xavier Mahon, Gabriel Etienne, Michele Baccarani, Simona Soverini, Gianantonio Rosti, Philippe Rousselot, Ran Friedman, Marie Deininger, Kimberly R. Reynolds, William L. Heaton, Anna M. Eiring, Anthony D. Pomicter, Jamshid S. Khorashad, Todd W. Kelley, Riccardo Baron, Brian J. Druker, Michael W. Deininger, and Thomas O Hare

17 SUPPLEMENTAL DATA

18 Figure S1, related to Figure 1. Pairing among key BCR-ABL1 kinase domain positions in compound mutations reported to date. (A) Circos diagrams for each of the 12 key BCR-ABL1 kinase domain positions. Ribbons extending from the position shown in bold depict connectivities that have been observed to date in clinically reported two-component compound mutations. Lettering at the outermost edge indicates the specific substitutions observed for each position. Absence of a connectivity ribbon does not rule out the possibility that the compound mutation will be reported in the future. Ribbon thickness represents the relative number of clinical cases of the pairing reported to date. For pairings represented among the panel of tested Ba/F3 BCR-ABL1 compound mutant cell lines in Figure 2, relative sensitivity to ponatinib in vitro is indicated by ribbon color. Ribbons for pairings not tested in vitro are dark grey. (B) Distribution of clinically reported BCR-ABL1 kinase domain compound mutations pairing the 12 key BCR-ABL1 positions. Each of the possible 66 pairings is listed and bars represent the number of cases reported to date (n=89).

19 Table S1, related to Figure 1. Two-component compound mutations reported at end of treatment in the PACE trial 1. 1 At a median follow-up of 15 months, adapted from (Cortes et al., 13). 2 Imatinib: I, nilotinib: N, dasatinib: D, bosutinib: B. 3 Mutations with a reported signal abundance value of <% (shown in gray) were not taken into consideration; the typical detection limit of conventional sequencing is ~%. Only candidate twocomponent compound mutations in which one component was present at an estimated frequency of % are included in the table. 4 PD: progressive disease, AE: adverse event.

20 Table S2, related to Figure 2. Cell proliferation IC 5 values of TKIs in parental Ba/F3 cells and Ba/F3 cells expressing native or mutant BCR-ABL1 1. Imatinib Nilotinib Dasatinib Ponatinib Rebastinib Bosutinib parental >12. >12. >768 > ± ± 11.3 Native ± ± ± ± ± ± 12.3 Single Mutants M244V ± ± ± ± ± ± 7.5 G25E ± ± ± ± ± ± 41.1 Q252H ± ± ± ± ± ± 12.8 Y253H > ± ± ± ± ± 6.2 E255V > ± ± ± ± ± 46.7 V299L ± ± ± ± ± ± 77.4 F311I ± ± ± ± ± ± 19.5 >12. >12. > ± ± ± 79.4 I315M >12. >12. > ± ± ± F317L ± ± ± ± ± ± 45.2 M351T ± ± ±.4 9. ± ± ± ± ± ± ± ± ± 9. H396R 1.3 ± ± ±.4.1 ± ± ± 8.6 Compound Mutants, -Inclusive M244V/ ± ± > ± ± ± G25E/ >12. >12. > ± ± ± Q252H/ >12. >12. > ± ± ± Y253H/ >12. >12. > ± ± ± 12. E255V/ >12. >12. > ± ± ± F311I/ >12. >12. > ± ± ± 167. /M351T >12. >12. > ± ± ± / >12. >12. > ± ± ± /H396R >12. >12. > ± ± ± 138. /E453K >12. >12. > ± ± ± Compound Mutants, Non- G25E/V299L ± ± ± ± ± ± Y253H/E255V >12. > ± ± ± ± 18.5 Y253H/F317L >12. > ± ± ± ± 25.1 E255V/V299L > ± ± ± ± ± V299L/F317L ± ± ± ± ± ± V299L/M351T ± ± ± ± ± ± 39.5 V299L/ 27.7 ± ± ± ± ± ± F317L/ ± ± ± ± ± ± All values are the mean of three independent experiments performed in quadruplicate. Standard error of the mean for three independent experiments is shown.

21 A B Residue Y253H/E255V E255V/ I315M Native System Active conformation Inactive conformation Native 67,987 56,192 Y253H/E255V 68,122 56,194 E255V/ 68,131 56,3 I315M 68,14 56,195 Probability % % Figure S2, related to Figure 3. (A) Probability distribution of selected residues lining the ponatinib site in native BCR-ABL1 coming into unfavorable proximity (within 2 Å) of ponatinib in the Y253H/E255V, E255V/ and I315M mutants. (B) System size (number of atoms) for active and inactive ABL1 kinase considered for explicit molecular dynamics simulations.

22 Table S3, related to Table 1. Characteristics of patients with a mutation detected at baseline prior to ponatinib therapy for which BCR-ABL1 cloning and sequencing was performed. 1 Imatinib: I, nilotinib: N, dasatinib: D, rebastinib: R, bosutinib: B. 2 Major hematologic response: MaHR, complete hematologic response: CHR, partial cytogenetic response: pcyr, complete cytogenetic response: CCyR, major molecular response: MMR, complete molecular response: CMR. 3 Pre-ponatinib treatment: Pre-Tx. 4 Only point mutations are listed; for full details, including insertions/deletions, see Tables S6-S9. 5 Stem cell transplant: SCT.

23 Table S4 related to Table 1. Characteristics of patients with a mutation other than detected at baseline prior to ponatinib therapy for which BCR-ABL1 cloning and sequencing was performed. 1 Imatinib: I, nilotinib: N, dasatinib: D, bosutinib: B. 2 Major hematologic response: MaHR, complete hematologic response: CHR, partial cytogenetic response: pcyr, complete cytogenetic response: CCyR, major molecular response: MMR, molecular response corresponding to 4.5 log reduction in BCR-ABL transcripts: MR Pre-ponatinib treatment: Pre-Tx. 4 Only point mutations are listed; for full details, including insertions/deletions, see Tables S6-S9.

24 Table S5, related to Table 1. Characteristics of patients with no key position mutation detected at baseline prior to ponatinib therapy for which BCR-ABL1 cloning and sequencing was performed. 1 Imatinib: I, nilotinib: N, dasatinib: D, rebastinib: R, bosutinib: B. 2 Major hematologic response: MaHR, complete hematologic response: CHR, partial cytogenetic response: pcyr, complete cytogenetic response: CCyR, major molecular response: MMR, complete molecular response: CMR. 3 Pre-ponatinib treatment: Pre-Tx. 4 Only point mutations are listed; for full details, including insertions/deletions, see Tables S6-S9. 5 Stem cell transplant: SCT, adverse event: AE.

25