Effect of type I and type II RAF inhibitors on RAF monomer complex activity.

A, Domain schematic of full-length RAF proteins, with the N-terminal acidic (NtA) motif shaded purple. Conserved 14-3-3 binding sites are indicated, with residue numbers corresponding to human BRAF. Co-expression of the RAF kinase domains with MEK1 allows for purification of the RAF/MEK1 monomer complex depicted schematically on the right and referred to as ARAF, BRAF, or CRAF in B. B, Representative concentration-response curves for the indicated type I, type I.5, and type II inhibitors with ARAF, BRAF, and CRAF monomer complexes are plotted as mean ± SD from one independent experiment performed in triplicate (n≥3).

AC50, IC50, and Hill slopes (nH) for inhibitor titrations of ARAF, BRAF, and CRAF.*

Effect of type I and type II inhibitors in an in vitro MAP kinase pathway reconstitution with BRAF, MEK1 and ERK2.

A, Paradoxical activation of BRAF as measured by changes in ERK phosphorylation. B, Paradoxical activation of BRAF as measured by changes in MEK phosphorylation. Phospho-MEK and Phospho-ERK were measured in parallel reactions. Data are plotted as mean ± SD from one of three independent experiments performed in triplicate (n=3).

Phosphomimetic mutations of the NtA motif prime ARAF and CRAF monomer complexes for paradoxical activation.

A, Concentration-response curves for titration of ARAFSSDD/MEK1 (ARAFSSDD) and CRAFSSDD/MEK1 (CRAFSSDD) with type I inhibitor GDC0879. B, Concentration-response curves for titration of ARAFSSDD and CRAFSSDD with type II inhibitors tovorafenib and LY3009120. Data for the wild-type ARAF and CRAF are reproduced from Figure 1 to facilitate comparison and are plotted in gray. Curves for additional type I and type II inhibitors are shown in Supplemental Figure 3. Data are plotted as mean ± SD from one independent experiment performed in triplicate.

AC50, IC50, and Hill slopes (nH) for inhibitor titrations of ARAFSSDD and CRAFSSDD.*

Effect of ATP concentration on paradoxical activation of BRAF by type I and type II inhibitors.

A-D, Concentration-response curves measuring MEK phosphorylation upon treatment of the BRAF monomer complex with increasing concentrations of the indicated inhibitor at ATP concentrations of 10 μM (blue), 100 μM (red), and 1000 μM (green). MEK phosphorylation activity is plotted as the raw FRET ratio in the panels on the left, and is replotted on the right as percent activity, normalized to the activity in the absence of inhibitor at each ATP concentration. Data are plotted as mean ± SD from one independent experiment performed in triplicate (n=3).

Mass photometry analysis of the oligomeric state of BRAF and MEK1.

A, MP mass distributions of the BRAF/MEK1 complex (32 nM) in the presence 10 μM ATPγS and without addition of inhibitor. B, “Single-shot” binding affinities of BRAF and MEK1 interactions in the absence of inhibitor, calculated as mean ± SD from five independent MP experiments (n = 5). A representative experiment is shown in panel A. C, Mass distributions of the BRAF/MEK1 complex (32 nM) in ATPγS (10 μM) at increasing concentrations of inhibitor GDC0879. D, Mass distributions of the BRAF/MEK1 complex (32 nM) in ATPγS (10 μM) at increasing concentrations of inhibitor vemurafenib. E, Mass distributions of the BRAF/MEK1 complex (32 nM) in ATPγS (10 μM) at increasing concentrations of inhibitor naporafenib. For experiments in C, D, and E, inhibitor concentrations from 10-2 nM to 105 nM were tested in three independent experiments (n=3) per inhibitor concentration, but only representative experiments from 100 nM to 104 nM are illustrated here. See also Supplemental Tables 2 and 3. F, BRAF monomer depletion plotted across inhibitor concentrations, as determined by relative distributions of counts from C, D, and E. G, BRAF dimer formation plotted across inhibitor concentrations, as determined by relative distributions of counts from C, D, and E. In F and G, the 45 kDa and 80 kDa peaks are summed to obtain the percent BRAF monomer, and the 115 kDa and 160 kDa peaks are summed to obtain percent BRAF dimer.

MP analysis of BRAF (R) and MEK1 (M) interactions at 10 μM inhibitor.*

A general model for inhibitor-induced paradoxical activation of RAF signaling.

In the resting state, autoinhibitory interactions between the RAS binding domain (RBD), cysteine-rich domain (CRD), and RAF kinase domain (KD) maintain RAF/MEK/14-3-3 complexes as a monomer in the cytosol. We find that this species is resistant to inhibitor-induced activation. RAS-driven recruitment to the membrane promotes opening of this complex, forming an “open” monomer that is not yet dimerized and active, but, unlike the autoinhibited monomer is susceptible to paradoxical activation (PA). Inhibitor binding to this open monomer causes paradoxical activation by displacing ATP to promote RAF dimerization, leading to the formation of signaling competent dimers. Dissociation of inhibitor from one or both sides of the dimer, or formation of a dimer with an inhibitor-free RAF, allows phosphorylation and activation of MEK. In this model, the open monomer is the target for activation, while the 14-3-3-bound dimer is the target for inhibition. Differing affinities of these distinct species for drug and for ATP can be expected to leave a window in which paradoxical activation is observed for any inhibitor that is more potent at displacing ATP from the open monomer than from the active dimer.