Biochemical, Biophysical, and Computational Characterization of RAF Dimer Inhibition and Paradoxical Activation by Diverse RAF Inhibitors

Abstract

The RAF family (ARAF, BRAF, and CRAF) of serine/threonine kinases are some of the most frequently mutated proteins in the human genome, and RAF alteration is frequently associated with a number of cancers and other developmental syndromes. The BRAFV600E mutation is particularly prevalent clinically, and, currently, the treatment of BRAFV600E cancers is co-administration of a BRAFV600E specific inhibitor with a MEK inhibitor, though this treatment regimen is only effective against BRAFV600E driven disease states, and even there, is not without issue. When given clinically, BRAFV600E inhibitors often lead to the development of secondary skin lesions, due to a phenomenon called paradoxical activation, where the inhibitors in question paradoxically activate RAF signaling, causing aberrant cell growth, and the aforementioned skin lesions. RAF inhibitors are now co-administered with a MEK inhibitor to more completely inhibit MAPK signaling and mitigate this effect. Here, I developed a reconstituted biochemical system to study RAF activity and RAF response to various inhibitors in vitro, in particular focusing on characterizing paradoxical activation. I find that specific classes of RAF inhibitor have specific isoform preferences for inhibition, and that type II RAF inhibitors inhibit RAF dimers with positive cooperativity, exhibiting tighter binding to the second RAF protomer than the first. We also report on crystal structures of BRAF bound to naporafenib and tovorafenib, two of these type II inhibitors. Notably, there is no evidence of paradoxical activation of these RAF dimers, regardless of inhibitor. I find that while RAF dimers do not exhibit paradoxical activation in this system, two of these same classes of inhibitor (I and II) are able to induce paradoxical activation of RAF monomers, in an isoform dependent fashion. In line with with the prevailing model for paradoxical activation, I find that it begins with inhibitor induced formation of active RAF dimers, from otherwise monomeric RAFs. However, my findings diverge from a crucial aspect of the prevailing model, which posits that the resulting inhibitor-induced RAF dimers escape inhibition due to negative cooperativity of inhibition: binding to one site of the dimer induces a kinase active but inhibitor resistant conformation in the other side of the dimer, which results in a buildup of half occupied RAF dimers that cause the inadvertent activation. I show here, via computational modeling of RAF dimerization, activation, and inhibition, that paradoxical activation can theoretically be induced by any ATP competitive inhibitor, regardless of cooperativity, but that the mechanism of activation can be affected by the type of inhibition cooperativity. Considering type II inhibitors exhibit positive cooperativity against RAF dimers, this computational modeling, in combination with my biochemical results, suggests that paradoxical activation by type II inhibitors is caused by inhibitor-free RAF dimers, which themselves arise because inhibitor induced dimers dissociate back into monomers more slowly than inhibitor itself dissociates from those aforementioned dimers. These findings deepen our understanding of RAF regulation, inhibition, and activation, granting valuable insights into what may be done to circumvent paradoxical activation entirely.