Skip to main content

Improving the selectivity of kinase inhibitors: Characterizing binding mechanisms of inhibitors targeting inactive states and allosteric sites

Final Report Summary - KIBINDING (Improving the selectivity of kinase inhibitors: Characterizing binding mechanisms of inhibitors targeting inactive states and allosteric sites)

Protein kinases play a key role in cell signaling, controlling a number of cellular processes including cell growth, proliferation and differentiation. It is therefore unsurprising that dysfunction of kinases has been implicated in a number of diseases, most notably cancer, but also diabetes and inflammation. A crucial part of their regulatory processes is the tightly controlled transition between a catalytically active conformation and one or more inactive conformations, so that the kinase is “on” only when needed. Since many oncogenic (cancer-causing) mutations result in a constitutively active conformation, developing kinase inhibitors has been a crucial step in the treatment of cancer as shown the most successful of the drug Gleevec in the treatment of chronic myelogenous leukemia.

One target of interest for the treatment of cancer is B-Raf, part of the RAS-RAF-MEK pathway. While B-Raf is mutated in about 8% of all cancers, B-Raf has been targeted for melanoma treatment since 80% of all malignant melanomas have a V600E mutation which results in a constitutively active kinase. Two drugs, vemurafenib and dabrafenib, which target B-Raf with the V600E mutation have been approved to treat melanoma by the FDA. Despite the success of targeting B-Raf V600E, patients taking the drugs can develop secondary tumors, which when analyzed are found to have wild-type B-Raf and not mutant B-Raf. The ability of B-Raf inhibitors to trans-activate Raf in a dimerization-dependent mechanism is known as “paradoxical activation”. A number of experiments have confirmed that B-Raf inhibitors cause paradoxical activation by binding to the wild-type protomer. Wild-type B-Raf requires dimerization for full activation, and it is believed that binding of an inhibitor to one of the protomers can transactivate the other protomer, which is thus responsible for the unregulated catalytic activity. This theory has additional support because some oncogenic mutations (e.g. G466V) inactivate B-Raf.

The objective of this project was two-fold: determine the effect of the V600E mutation on the conformational landscape of the B-Raf monomer and determine the molecular mechanism by which B-Raf inhibitors result in transactivation at the molecular level. Due to the difficulty in answering these questions experimentally, we adopted a computational approach using molecular dynamics (MD) simulations and a state-of-the-art enhanced sampling method, parallel-tempering metadynamics (PT-metaD).

Despite the availability of more than 30 crystal structures of B-Raf, the effect of the mutation on the dynamics of the transition between the active and inactive states remains unclear. PT-metaD was used to calculate the free energy surfaces (FES) of the wild-type and mutant B-Raf monomers. In agreement with experimental results, we find that the active state of the mutant is stabilized by the mutation. The results, recently accepted for publication on the J. Am. Chem. Soc. DOI: 10.1021/jacs.5b01421 show a significant effect of the mutation of the conformation of B-Raf in the active state due to increased flexibility of the activation loop. This is turn leads to an increased ATP-binding site volume which could increase the rate of ADP-unbinding which is thought to be the rate-limiting step in the phosphorylation cycle. The free energy surfaces also revealed that the mutation increases the barrier for the active to inactive transition trapping the kinase in the active state. Further a possible intermediate in the wild-type active to inactive transition was identified. This conformation could be a target in the design of inhibitors for the treatment of cancers due to overactive RAS.

The second objective was split into two tasks. First, long unbiased MD simulations were performed on the wild-type dimer, the V600E mutant dimer, and the wild-type dimer with an inhibitor bound in one of the active sites. Principal component analysis and elastic network analysis were performed to identify key residues involved in allosteric communication within the dimer. While the results are still being prepared for publication, the results show that the dimer interface is very dynamic and inhibitor binding is correlated cross-interface motions. Another open question we are trying to answer is the effect of the mutation on dimerization. While the mutant is active as a monomer, it forms more stable dimers than the wild-type. The results from our study of the monomer showed no significant structural differences in the residues which make up the dimer interface.

While the MD trajectories were 500 ns long, PT-metaD simulations are required to capture conformational changes between the inactive and active states of the dimer. These simulations which should are currently being performed and will show how the bound inhibitor affects the free energy barrier between and the conformational landscape of this transition.