NMR reveals the allosteric opening and closing of Abelson tyrosine kinase by ATP-site and myristoyl pocket inhibitors

Successful treatment of chronic myelogenous leukemia is based on inhibitors binding to the ATP site of the deregulated breakpoint cluster region (Bcr)–Abelson tyrosine kinase (Abl) fusion protein. Recently, a new type of allosteric inhibitors targeting the Abl myristoyl pocket was shown in preclinical studies to overcome ATP-site inhibitor resistance arising in some patients. Using NMR and small-angle X-ray scattering, we have analyzed the solution conformations of apo Abelson tyrosine kinase (c-Abl) and c-Abl complexes with ATP-site and allosteric inhibitors. Binding of the ATP-site inhibitor imatinib leads to an unexpected open conformation of the multidomain SH3-SH2-kinase c-Abl core, whose relevance is confirmed by cellular assays on Bcr-Abl. The combination of imatinib with the allosteric inhibitor GNF-5 restores the closed, inactivated state. Our data provide detailed insights on the poorly understood combined effect of the two inhibitor types, which is able to overcome drug resistance.Protein kinases, such as Abelson tyrosine kinase (c-Abl), control numerous cellular signal pathways, and therefore require tight regulation (1). Modulation of c-Abl activity is achieved by an autoinhibitory mechanism involving the N-terminal region of the protein (1, 2). A myristoyl moiety covalently bound to Gly2 occupies a deep pocket located in the C-lobe of the kinase domain and stabilizes docking of the SH3 and SH2 domains to the kinase domain (3–5) (Fig. 1). In the oncogenic fusion protein Bcr-Abl, this crucial autoinhibition is lost as a result of chromosomal translocation, leading ultimately to chronic myelogenous leukemia (CML) or acute lymphoblastic leukemia (6, 7).Open in a separate windowFig. 1.Domain definition and structure of the c-Abl fragment c-Abl^83–534. (A) Crystal structure of c-Abl in its fully assembled state (PDB ID code 2FO0). The inhibitor PD166326 (orange) is bound to the ATP site, whereas myristic acid (gray) occupies a deep pocket in the C-lobe of the kinase domain. Domain coloring follows that in B. Residues 2–82, which form the disordered N-terminal cap, are schematically represented by a gray dashed line. (B) Amino acid sequence and schematic representation of secondary structure elements of c-Abl^83–534. Color coding is as follows: SH3 domain (green), SH2 domain (yellow), kinase domain (blue), interdomain linkers (red), and activation loop (magenta). (C) Chemical structures of the ATP-competitive inhibitor imatinib (STI-571/Gleevec; Novartis) and the allosteric inhibitor GNF-5.The ATP-binding site inhibitors imatinib (STI-571/Gleevec), nilotinib (AMN-107/Tasigna), and dasatinib (Sprycel) constitute the front-line therapy against CML (8–11). However, spontaneous point mutations render these inhibitors ineffective and cause clinical relapse in advanced-phase patients (12, 13). Although nilotinib and dasatinib retain their efficacy against many of the imatinib-resistant mutants, the “gatekeeper” T334I mutation (T315I in Abl 1a numbering) abrogates the binding of all three inhibitors (12). Emergence of this multidrug-resistant mutant, which occurs in ∼15% of patients with resistance to imatinib, has stimulated the search for new therapeutics (14). Recently, several new ATP-competitive inhibitors that are active against the T334I mutant (15–17) have been developed, and one of these, ponatinib (Iclusig) (15), has received US Food and Drug Administration approval. An alternative approach has resulted from the discovery of allosteric inhibitors, which bind to the myristoyl-binding pocket of c-Abl (18, 19). Subsequent studies have revealed that combining allosteric inhibitors with ATP-competitive inhibitors overcomes T334I-related resistance in an in vivo model and may be a relevant therapeutic strategy (20). The importance of the myristoyl-binding pocket is further supported by the discovery of small-molecule c-Abl activators that bind to this site (21, 22).c-Abl and other tyrosine kinases are regulated by complicated allosteric interactions between their constituent domains (5). Whereas crystallographic structures have laid the foundation for our current understanding of c-Abl regulation, the vast majority of solved structures represent the isolated kinase domain in complex with small molecules. Only two reports by Kuriyan and coworkers (3, 5) provide structures of the entire minimal autoregulatory fragment of c-Abl, which comprises the SH3, SH2, and kinase (also termed SH1) domains. In both cases, the protein was complexed with an ATP-site inhibitor and a myristoyl chain attached covalently to Gly2 or added in trans. Three structural features were identified as requirements for the assembly of a “closed,” inactive state (5) (Fig. 1A): (i) docking of the SH3 domain to a polyproline helix in the SH2-kinase linker, (ii) docking of the SH2 domain to the kinase domain facilitated by the binding of the myristoyl moiety, and (iii) the clamp formed by the N-terminal cap region. The removal of these “linchpins” resulted in an “activated” c-Abl mutant with the SH3-SH2-kinase domains arranged into an elongated structure with a direct contact between the SH2 domain and the N-lobe of the kinase (5). This has inspired further studies on the role of the SH2 domain in the regulation of c-Abl and other kinases (23–25).Crystal structures represent frozen snapshots of protein states that may not reflect all physiologically relevant conformations. In particular, it is expected that the active forms of c-Abl and other kinases undergo dynamic exchange, which makes them difficult to crystallize. Indeed, for example, the apo form of c-Abl has resisted crystallization so far. In principle, solution NMR can provide much of the missing dynamic information to understand protein function (26). However, its applicability to larger systems is restricted by its inherent size limit, its low sensitivity, and the need for isotope labeling. Thus, so far, solution conformations and dynamics have been analyzed by NMR for smaller fragments of protein kinases comprising the catalytic and/or adjacent domains, for example, of protein kinase A (27, 28), MAP kinase p38 (29, 30), and Eph receptor tyrosine kinase (31), as well as c-Abl (32) (c-Abl^248-519; throughout this report the amino acid numbering follows the 1b isoform). Here, we have determined the solution conformation and domain motions of the considerably larger autoregulatory fragment c-Abl^83–534 (designated as c-Abl in the following when clear from the context), which comprises the SH3, SH2, and kinase domains (Fig. 1B), by advanced NMR experiments in combination with small-angle X-ray scattering (SAXS). The data provide the first structural information on the apo form of c-Abl in the absence of inhibitors, which is shown to adopt the “closed” conformation. Unexpectedly, the addition of the catalytic site inhibitor imatinib induces a large structural rearrangement characterized by the detachment of the SH3-SH2 domains from the kinase domain and the formation of a dynamic “open” inactive state, which is inhibited in the ATP site. As a consequence of this opening, Tyr245 within the SH2-kinase linker becomes exposed. The in vivo relevance of this unexpected opening is corroborated by an observed increase in Tyr245 phosphorylation at moderate imatinib concentrations in cellular assays on full-length Bcr-Abl. In contrast to imatinib, addition of the myristoyl pocket inhibitor GNF-5 (20) to apo c-Abl induces only local changes around the myristoyl-binding pocket and keeps the protein in the “closed” state. However, addition of GNF-5 to the “open” c-Abl•imatinib complex restores the “closed,” inactive conformation. These findings on the allosteric actions of ATP and myristoyl pocket inhibitors reveal molecular details of their recently reported synergy to overcome drug resistance and may help to devise new strategies for drug development.