I have now completed a 10.2 million compound virtual screen of FABP5, an incredibly important drug target for anti-inflammatory, anti-nociceptive, and anti-cancer medicinal development. I’m actually not a fan of this FABP5 model for several reasons. Firstly, the protein crystallized as a multimer with different conformations and binding modes of the ligand SB FI 26 in different sections of the X-ray crystal. I can easily overlook this if there was more consistency across the different chains, but there really isn’t. Secondly, the ligand that was used for crystallization was not enantiomerically pure. This is a huge error, and the “discovery” of both enantiomers binding at the portal site but only one enantiomer binding deeper in the pocket is totally ridiculous. Although binding experiments later demonstrated that both enantiomers bind with almost equal affinity, using a mixture for an X-ray crystallization is, in my opinion, a waste of time. Lastly, the conformation of the portal site beta sheet of FABP5 is entirely open. As a ligand enters the FABP binding site, the beta sheet should close in. By modelling with an open conformation of the beta sheet, certain steric clashes might not be accurately predicted, and more favorable interactions may also not be modelled accurately. Nevertheless, since there are few X-ray co-crystal structures of FABP5, I chose to go ahead and perform virtual screening with the structure (PDB ID 5UR9).
As I previously mentioned, I split up the molecular profile of the docked ligands into two major segments; a more polar, lighter molecular weight tranche and a heavier, more hydrophobic tranche. The figure below demonstrates a few different predicted hits from the tranches modelled in the FABP5 binding. Within the videos, one can visualize how the difference in chemical profile affects the molecular interactions with the FABP binding site. Interestingly, the carboxylic acid penicillin motifs emerged as promising potential ligands of FABP5. It’s always fascinating to see how existing drugs can be repurposed, and perhaps there is potential for penicillin analogs as FABP5 inhibitors. Another interesting set of molecules included the sulfonated glucose analogs. The multiple sulfate groups are predicted to form incredibly strong electrostatic interactions within the binding site. The best scoring compound from the hydrophobic tranche, interestingly, has a naphthyl group like SB FI 26. Since the FABP5 binding site is so hydrophobic, this observation is not surprising. With a binding score over double the cognate ligand, I am very confident this compound will bind with submicromolar affinity to FABP5. If you are interested in analyzing more of these results, especially if you’d like to test some of these compounds, send me an email. Until next time, cheers.
