Targeting the Human Influenza a Virus: The Methods, Limitations, and Pitfalls of Virtual Screening for Drug-like Candidates Including Scaffold Hopping and Compound Profiling

Abstract

In this study, we describe the input data and processing steps to find antiviral lead compounds by a virtual screen. Two-dimensional and three-dimensional filters were designed based on the X-ray crystallographic structures of viral neuraminidase co-crystallized with substrate sialic acid, substrate-like DANA, and four inhibitors (oseltamivir, zanamivir, laninamivir, and peramivir). As a result, ligand–receptor interactions were modeled, and those necessary for binding were utilized as screen filters. Prospective virtual screening (VS) was carried out in a virtual chemical library of over half a million small organic substances. Orderly filtered moieties were investigated based on 2D- and 3D-predicted binding fingerprints disregarding the “rule-of-five” for drug likeness, and followed by docking and ADMET profiling. Two-dimensional and three-dimensional screening were supervised after enriching the dataset with known reference drugs and decoys. All 2D, 3D, and 4D procedures were calibrated before execution, and were then validated. Presently, two top-ranked substances underwent successful patent filing. In addition, the study demonstrates how to work around reported VS pitfalls in detail.

Keywords

• influenza
• neuraminidase inhibitors
• noncompetitive inhibition
• virtual screening
• ligand docking
• screening pitfalls