Fig. 6: MSI2 small molecule inhibitors identified in 3D structure models of MSI2-RNA interactions reduce MYC expression.

A Local quality estimate. Human MSI2 structure was simulated by using the template of 2mssA and 2lyvA. B Comparison with non-redundant set of PDB structures. The simulated structure has three potential binding sites searched by Van der Waals radii over the molecular surface. C For RNA AUUGG, each binding site was simulated 1500 times with redundant radius searching for the lowest energy of interaction. These results were then combined to return the lowest energy from these parameters. D Cavity 2 which had the lowest energy was chosen as the RNA docking site. E The RNA interacting residues from MSI2 were found based on hydrophilic, hydrophobic and charge-π interactions within 4 Å of the RNA binding site (Top panel). The interacting residues from (E, Top) are: Pro103, Lys104, Lys111, Phe113, Met 140, Met 142, Phe153, Phe155, Lys184, Ala185, Gln186, Pro187, Lys188, Val190, Met191, Phe 192 and Pro193. Each compound was initially tested by energy minimized force field analysis and then docked to Cavity 2 individually. The shared interacting residues on MSI2 (for all compounds tested with RNA): Pro103 Lys111 Phe113 Met 142, Phe153, Phe155, Lys184, Ala185, Gln186, Pro187, Lys188 and Met191 (Middle panel). The RNA (5′AAUUCCAGCGAGAG3′) molecular structure was simulated by force field using Monte Carlo analysis for sampling various conformations; the minimized lowest energy conformation was chosen. The RNA was docked to MSI2 protein via template-based modeling. MSI2 is shown in green, whereas the RNA molecule is shown in the cartoon model as blue and brown. The interacting residues from MSI2 to RNA were selected based on polar, charge, hydrophobic or polar-hydrophobic interactions, within the range of 4 Å: Phe24, Asp55, Pro56, Arg62, Phe64, Lys94, Arg100, Ala101, Gln102, Pro103, Lys104, Val106, Thr107, Gln186, Val190, Met191, Phe192, Pro193 (Bottom panel). F Gossypol binds to Cavity 2 by torsion of the central C–C bond. H Oleic acid binding to Cavity 2. G For binding of MP-Gr, the torsion occurs across the methanimine linkage. Also, the methoxyl oxygen forms additional polar contacts. Red is used for labeling the key binding structures besides interactions with surrounding residues. For the rest of the compounds, they all take similar conformations to mimic the RNA which goes around this loop. The RNA ligand surrounding the loop of MSI2 is docked to polar residues outside as this was the best molecular interaction with the lowest energy. I Simvastatin: Due to its small structure and relatively inflexible rings, Simvastatin resides right on the RNA binding site. The polar interaction groups with the MSI2 are shown in blue. The residues interacting with Simvastatin from MSI2 are: Gln102, Pro103, Met105, Lys111, Phe113’, Met142, Phe153, Phe155, Lys183, Lys184, Ala185, Gln186, Pro187, Lys188, Met191. J Idarubicin: The 3-ring structure occupies the RNA binding groove with the ether linkage of Idarubicin crossing over the loop from MSI2. Polar groups that further improve binding stability are shown in blue. The residues interacting with Idarubicin from MSI2 are: Gln102, Pro103, Lys104, Met105, Phe113, Phe153, Lys183, Lys184, Ala185, Gln186, Met191. K GW7647 docking structure with MSI2, which is also another predicted good inhibitor with lower energy. The potential interacting site is also suitable for long alkyl chain fatty acids, e.g. oleic acid (Top panel). Any compound that could bend over this loop and pair with polar groups (the interacting red loop from MSI2 contains Lys184, Ala185, Gln186, Pro187 and Lys188) towards the end or along the side could be potential good inhibitors. The long-carbon-chain acyl compounds can form this type of “C” shape with potential polar interactions, therefore they are expected to be potential inhibitors. If their carbon chains at the C curve possess negatively charged/polar residues (such as in Gossypol), these would be potentially the better candidates to bind to the lysine residues and inhibit MSI2-binding ability, which mimic the phosphates from the RNA molecule (Bottom panel). L Screening for selective inhibitors of MSI2-MYC IRES interaction. Wild type and mutant MYC IRES-binding RNA sequences were labeled with FITC and tested in fluorescence polarization assays for inhibition of MSI2 binding activity. (Right panel) Diagram of drug screening by fluorescence polarization assays. M Results of drug inhibition of spheroid formation using Huh7 cells. N Effect of drug panel on MYC expression in Huh7 cells. The immunoblot is shown; β-ACTIN included as loading control. Expression of MSI2 promoted self-renewal via increased MYC expression. O Microscale thermophoresis (MST) assay of GW7647 was performed with full-length wild-type MSI2 with other RNA-binding proteins and non-RNA-binding protein (GST). P Microscale Thermophoresis (MST) assay was performed with full-length wild type MSI2: K184A, P187A and K188A mutants. Q Representative images of EMSA with GW7647 in the presence of MSI2 and IRES RNA probe (Left panel). Microscale Thermophoresis (MST) assay was performed with full-length wild type MSI2 in the presence of different concentration of GW7647 (Right panel). A ligand binding curve of the drugs is shown (Right panel). Two-tailed paired t-test: *P < 0.05.