Generation and Verification of a Three-Dimensional Model of the ALK Kinase Domain: ALK - a “Supermutant” of Abl.

Anaplastic Lymphoma Kinase (ALK) is a receptor tyrosine kinase that plays a key role in oncogenesis in a subset of non-Hodgkin’s lymphomas, through the generation of fusion proteins containing its kinase domain (KD), such as NPM/ALK (N/A). A three-dimensional (3D) model of the ALK KD was generated by homology modelling, using the crystal structures of the inactive conformations of Abl and insulin receptor kinase as templates. Sequence alignment of ALK and Abl, which share 34% identity and 61% homology in their KDs indicated that sequences diverged at positions known to be critical for imatinib binding to Abl.

ALK sequence differs from Abl at 13 of the 18 amino acids known to be mutated in the Abl KD of imatinib resistant CML patients, with 3 of these ALK amino acids being identical to mutations observed in patients (Y253F, F317L and V379I). Hence, ALK could be considered as a ‘supermutant’ of Abl that should be resistant to Abl inhibitors. Mutation of ALK to an Abl-like sequence should be able to ‘restore’ sensitivity of ALK to Abl inhibitors. In Bcr/Abl, a key determinant in imatinib sensitivity is threonine³¹⁵ (T315), which is often mutated to a bulky isoleucine residue in relapsed CML patients. In ALK, the corresponding amino acid is a leucine (L256 of N/A), which similarly to isoleucine, causes a steric hindrance in the ATP binding pocket and eliminates the possibility of hydrogen bond formation. Therefore, we hypothesised that a L256T mutation in N/A might render this kinase sensitive to Abl inhibitors. Molecular docking of Abl inhibitors to ALK wild type (WT) and L256T 3D models predicted that WT-ALK should be insensitive to three Abl inhibitors (compound 583, compound 683, imatinib), while L256T-ALK should be inhibited by compound 583 and 683, but not imatinib, which would require further mutagenization. To test this hypothesis we generated BaF3 cell lines stably transfected with WT- and L256T-N/A. We observed that L256T-BaF3 cells were 5 times more sensitive to growth arrest induced by compound 583 than WT-BaF3 cells as determined by ³[H]-thymidine uptake (IC₅₀: 0.2±0.02 v 1.0±0.07 μM [mean ± sem]), and 7 and 3.5 times more sensitive to cell death induced by compounds 583 (70 v 10% after 48h) and 683 (70 v 20% after 16h) respectively, as determined by trypan blue exclusion. No difference in sensitivity to imatinib was observed in the 2 cell lines.

In addition, autophosphorylation activity of L256T-N/A, assessed by anti-phosphotyrosine immunoblotting and in vitro radioactive kinase assays, was inhibited by compounds 583 and 683 with IC₅₀s of approximately 0.2 and 3 μM respectively, whereas autophosphorylation activity of WT-N/A was not affected by either inhibitor. Neither L256T- nor WT-N/A kinase activities were inhibited by imatinib. Summarizing, L256T-N/A was significantly more sensitive to compounds 583 and 683 compared with WT-N/A. These data provide experimental evidence supporting our modelling predictions, thereby verifying the reliability of the ALK 3D model. The sensitisation of ALK to compounds 583 and 683, but not imatinib, by the L256T mutation is in agreement with data describing the activities of these inhibitors on Abl mutants, excluding the T315I, and supports the ‘supermutant’ hypothesis.