Critical Assessment of a Structure-Based Screening Campaign for IDO1 Inhibitors: Tips and Pitfalls

Abstract

Over the last two decades, indoleamine 2,3-dioxygenase 1 (IDO1) has attracted wide interest as a key player in immune regulation, fostering the design and development of small molecule inhibitors to restore immune response in tumor immunity. In this framework, biochemical, structural, and pharmacological studies have unveiled peculiar structural plasticity of IDO1, with different conformations and functional states that are coupled to fine regulation of its catalytic activity and non-enzymic functions. The large plasticity of IDO1 may affect its ligand recognition process, generating bias in structure-based drug design campaigns. In this work, we report a screening campaign of a fragment library of compounds, grounding on the use of three distinct conformations of IDO1 that recapitulate its structural plasticity to some extent. Results are instrumental to discuss tips and pitfalls that, due to the large plasticity of the enzyme, may influence the identification of novel and differentiated chemical scaffolds of IDO1 ligands in structure-based screening campaigns.

Turning a Tumor Microenvironment Pitfall into Opportunity: Discovery of Benzamidoxime as PD-L1 Ligand with pH-Dependent Potency

“…Moreover, the detrimental effect of the methyl substituents in the ortho or meta position on the binding efficiency (BEI) is also evidenced by the binding modes of the relative compounds, with such groups not being involved in specific contacts with hydrophobic residues. Overall, this is in line with previous works suggesting that scoring functions are generally more capable of identifying the correct ligand binding poses than predicting binding affinities [ 35 , 36 , 37 ].…”

“…Compounds of the library showing a GScore lower than −4 kcal/mol, as well as engaging Asp122 and/or Lys124 with electrostatic and/or hydrogen bond interactions, were short-listed. Specifically, the energy cutoff of −4 kcal/mol was chosen on the basis of previous data suggesting this value as a good arbitrary threshold to identify hit compounds amid successfully docked ligands [ 37 ]. However, it should be stressed that GScore values do not generally correlate with the experimental binding data [ 35 , 36 , 37 ], pinpointing that further work is required to improve such a scoring function for reducing the number of false-positive virtual hits and/or the discard of active compounds.…”

“…Specifically, the energy cutoff of −4 kcal/mol was chosen on the basis of previous data suggesting this value as a good arbitrary threshold to identify hit compounds amid successfully docked ligands [ 37 ]. However, it should be stressed that GScore values do not generally correlate with the experimental binding data [ 35 , 36 , 37 ], pinpointing that further work is required to improve such a scoring function for reducing the number of false-positive virtual hits and/or the discard of active compounds. The presence of polar interactions with Asp122 and/or Lys124 was sought as a key feature of the selected hit compounds to increase the likelihood of having pH-dependent binding activity.…”

Progress of the “Molecular Informatics” Section in 2022

“…To collect high-quality research articles in all fields of molecular informatics applied to the understanding of the intricate world of biology and molecular research, the Topical Collection, "Feature Papers in Molecular Informatics", was created, and, looking only at 2022, 25 papers were published . The most recent developments in computer science applied to molecular modeling have been discussed as hot topics in the two collections entitled "State-of-the-Art Molecular Informatics in Italy", which was closed at the end of 2022 [26][27][28][29][30][31][32][33], and "Latest Review Papers in Molecular Informatics 2023", which was closed at the end of February 2023 [34,35]. In 2022, the number of closed SIs was 22, and the top SI based on the number of published articles and visualizations was the one dedicated to "Deep Learning and Machine Learning in Bioinformatics" .…”