As part of our studies to identify novel inhibitors for the immunosuppressive enzyme indoleamine 2,3-dioxygenase 1 (IDO1) for restoring tumor immunity, we investigated the ability of the enzyme's x-ray crystal structure to predict active molecules when used in molecular docking based virtual screening of fragment libraries.

The 352 compound Zenobia Therapeutics fragment library was screened by differential scanning fluorimetry and 74 compounds were identified that altered the thermal stability of the protein. Of these, biochemical analysis showed that only 14 inhibited the enzyme activity by >50% at 1 mM, and hydrazine containing fragments were the most potent inhibitors of recombinant human IDO1 (rhIDO1), with the best being a benzothiazole compound retrieving an IC50 of 8 µM. We recently showed that phenylhydrazine is a more potent inhibitor (IC50 0.3 µM), which provided a dataset for testing various docking strategies to discover fragment inhibitors.

The fragment library was docked into the IDO1 active site (PDB code 2D0T) using GOLD driven by the scoring functions ChemPLP, Goldscore, ASP or Chemscore. The three top scoring poses for each compound were kept then rescored by each scoring function. The combination of docking by Goldscore and rescoring by Chemscore to identify the top scoring pose gave the best outcome, an 8.3 fold enrichment of active compounds within the top 5% of the ranked library compared to random selection.

Based on this docking and scoring strategy we performed a virtual screening campaign on a 14,000 compound library containing commercially available phenylhydrazine derivatives. Our characterization of phenylhydrazine binding indicated that the hydrazine moiety interacts with the heme iron. Based on this, a set of compounds was selected from the top 30% of poses kept from the docked library for biochemical testing that had predicted hydrazine to heme iron interactions. Of 21 compounds chosen, 10 were tested and 9 inhibited IDO1 with IC50 of < 10 µM. Also, compounds predicted to make a hydrogen bond acceptor to heme iron interaction were selected and an additional 9 compounds chosen of which 6 were tested and only 1 inhibited rhIDO1 with IC50 of < 10 µM. For these latter ones the hydrazine was not predicted to interact with the heme iron.

This study indicates that the rhIDO1 structure would be useful for discovering inhibitors in a diverse fragment library, and an additional hydrazine to heme iron binding descriptor is useful for finding actives in a library of compounds containing the phenylhydrazine core.

Citation Format: Simon Fung, Jack U. Flanagan, Christopher J. Squire, Brian D. Palmer, Lai-Ming Ching. A molecular docking strategy for identifying fragment inhibitors of indoleamine 2,3-dioxygenase 1 (IDO1). [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 5372. doi:10.1158/1538-7445.AM2014-5372