Featured Article - June 2015
Short description: Reconstructed ancient ancestors of Src and Abl shed light on Gleevec's specificity.
Kinases have a strongly conserved catalytic domain structure, yet the cancer drug Gleevec shows high selectivity towards Abl kinase, with 103-lower affinity for closely related kinases such as Src. It has been proposed that Gleevec binding induces a global conformational change that differs between Abl and Src, but both modeling- and domain-swapping experiments have failed to identify the basis of specificity.
As protein function reflects not only its sequence and structure but also its energy landscape, Kern and colleagues decided to take an orthogonal approach. Reasoning that both Abl and Src kinases share a common ancestor, they postulated that not just specific residues, but also the surrounding amino acids would be important in sculpting the distinct energy landscapes of the kinases during evolution. They therefore used a Bayesian phylogenetic approach to design four potential ancestral proteins of Abl and Src. These proteins were constructed and shown to have intermediate Gleevec-binding affinities. Overall, the kinetic scheme of the ancestral proteins is the same as that of modern kinases, but the conformational steps (i.e. kinetic parameters) are altered in concert with the affinity changes. This reflects a progressive shift in conformational equilibrium between two bound states; differences in this dynamic process of the kinase/drug complex, but not binding/dissociation rates, are primarily responsible for selectivity.
Structural analysis of the last common ancestor, ANC-AS, helped identify globally distributed amino acid changes responsible for selectivity. In particular, modifying 15 amino acids in the N-terminal domain of ANC-AS (to generate ANC-AS(+15)) increased the Gleevec affinity to equal that of Abl; those amino acid residues underlie most of the change in the equilibrium towards Abl's inhibitor-bound, induced-fit state. Mapping those 15 amino acid changes onto the crystal structure of ANC-AS emphasizes why the basis of selectivity had been so difficult to determine: many of the 15 residues are removed from the binding pocket and form a hydrogen bond network in ANC-AS that cannot occur in ANC-AS(+15) or Abl. The authors speculated that in the absence of this network, the conserved P-loop has mobility and can close over the drug, whereas with the network, the P-loop is stabilized in a different conformation. The concept that the P-loop may be involved in discrimination was previously tested in domain-swapping experiments, but those experiments did not include these 15 residues that were identified from the ancestral reconstruction.
In conclusion, this study illustrates not only the important role that evolutionary reconstruction can offer in identifying key elements that allow for atomistic discrimination of ligands, but also indicates that energy landscapes are a driving force in evolution.
C. Wilson et al. Using ancient protein kinases to unravel a modern cancer drug's mechanism.
Science. 347, 882-6 (2015). doi:10.1126/science.aaa1823