Garcinolic Acid Distinguishes Between GACKIX Domains and Modulates Interaction Networks

Abstract

Natural products are often uniquely suited to modulate protein-protein interactions (PPIs) due to their architectural and functional group complexity relative to synthetic molecules. Here we demonstrate that the natural product garcinolic acid allosterically blocks the CBP/p300 KIX PPI network and displays excellent selectivity over related GACKIX motifs. It does so via a strong interaction (K_D 1 μM) with a non-canonical binding site containing a structurally dynamic loop in CBP/p300 KIX. Garcinolic acid engages full-length CBP in the context of the proteome and in doing so effectively inhibits KIX-dependent transcription in a leukemia model. As the most potent small-molecule KIX inhibitor yet reported, garcinolic acid represents an important step forward in the therapeutic targeting of CBP/p300.

Keywords: CBP/p300; acute myelogenous leukemia; cMyb; natural product; protein-protein interaction inhibitor.

Figure 1.

A) Structure of GACKIX domains tested in this study; CBP KIX, cgMED15 KIX, scMED15 KIX, and ARC105 KIX derived from Protein Data Bank entries 2AGH, 4D7X, 2K0N, and 2GUT respectively. Flexible loop between α1 and α2 is highlighted in each domain. Structures of B) garcinolic and C) gambogic acid. D) Garcinolic acid inhibition of CBP KIX binding to the MLL (black circles), Myb (black squares), and pKID (black triangles) Transcriptional Activation Domains (TADs) and ARC105 KIX binding to the MLL (gray) TAD as measured by competitive fluorescence polarization. Error bars are standard deviation of the mean as measured from three independent experiments.

Figure 2.

A) Stopped-flow fluorescence association data between Y631W CBP KIX and garcinolic acid. All data points are averaged from 35–40 individual traces with the indicated error (SD). B) ¹⁹F-NMR data of 3FY/4FF-labeled CBP KIX (15 μM, bottom) exhibiting lowering of fluorine-labeled Y631 residue signal upon garcinolic acid (30 μM, top) binding as measured by PrOF NMR. C) Schematic of CBP KIX (dark blue) highlighting unaffected fluorine-labeled phenylalanine and tyrosine residues (light blue) and perturbed Y631 residue (lime green) as measured by PrOF NMR. See Figures S10–S15 for additional details. D) ¹H-¹⁵N HSQC NMR data exhibiting perturbation of CBP KIX (100 μM) residues V608, I611, and F612 in the presence of 0 (blue), 0.25 (yellow) and 0.5 eq of garcinolic acid (green).

Figure 3.

A) DSF data for CBP KIX complexed with garcinolic acid and MLL and Myb peptides. All complexes of CBP KIX had an increased melting temperature over free CBP KIX. Error bars are the standard deviation as calculated from three replicates. B) Thermal shift of endogenous CBP in MV4–11 nuclear extracts treated with 25 μM garcinolic acid. A Western blot of the input and and treated samples was probed with anti-CBP antibody. C) Impact of garcinolic acid on CREB-dependent cyclin genes in MV4–11 acute myeloid leukemia cells over 8 h. Error bars are standard deviation calculated from technical triplicates of biological duplicate experiments. D) Garcinolic acid decreases viability of MV4–11 (black) and HL-60 (grey) acute myeloid leukemia cells over 48 h. Error bars are standard deviation calculated from biological triplicates.