Search for new antiviral compounds using fragment screening methodology

Abstract

[spa] Los Picornaviridae son una de las familias de virus más diversas y conocidas desde hace más tiempo. Esta familia incluye importantes agentes patógenos que afectan a humanos y a animales. Los Picornaviridae son virus pequeños, icosaédricos, de ARN de cadena sencilla de sentido positivo y causan una gran variedad de enfermedades, tales como encefalitis y poliomielitis. Se dispone de vacunas para el poliovirus, el virus de la hepatitis A y el virus de la fiebre aftosa, pero no se ha implementado ninguna profilaxis efectiva para otros picornavirus. Hasta ahora, la investigación antiviral se ha centrado en la cápside, mientras que los inhibidores dirigidos a proteínas no estructurales (como las proteasas, las helicasas y las polimerasas) están todavía por explorar. Este proyecto se centró en la caracterización estructural y bioquímica de la proteasa 3C de enterovirus B93 (EV-B93) y de los complejos de esta proteasa con varios inhibidores covalentes. El segundo objetivo fue conseguir inhibidores no covalentes de la proteasa EV-B93 3C y realizar una caracterización bioquímica, antiviral y estructural de los complejos de EV-B93 3C con estos inhibidores.

[eng] Picornaviridae are among the most diverse and oldest known viral families that include many important pathogens of humans and animals. They are small, icosahedral (+)ssRNA viruses, causing a variety of diseases, such as encephalitis, and poliomyelitis. Vaccines are available for poliovirus, hepatitis A virus and foot-and mouth disease virus, but no effective prophylaxis is implemented for other picornaviruses. Thus far, anti-viral research has focused on the capsid, whereas inhibitors targeting non-structural proteins (i.e. proteases, helicases, polymerases) have remained largely unaddressed. The project was focused on structural and biochemical characterization of the enterovirus-B93 (EVB93) 3C protease alone and in complex with several covalent inhibitors. The second objective was to identify the first non-covalent potent inhibitors of the EV-B93 3C protease and their further biochemical, antiviral, and structural evaluation. This work studied the in-vitro proteolytic activity of the EV-B93 3C protease, alone and in the presence of two known covalent inhibitors - rupintrivir and compound 1, as well as three low molecular weight covalent inhibitors - NZO, NZN and DB5_60. The crystal structures of the EV-B93 3C protease alone and in complex with rupintrivir, compound 1, and NZN molecule were solved at high resolution (1.57, 1.50, 1.32, and 1.73 Å, respectively). The structures revealed that the protein adapts a chymotrypsinlike fold similarly to other picornavirus 3C proteases and possesses His-40, Glu-71 and Cys-147 as a catalytic triad. The STD NMR-based fragment screening was performed to select non-covalent binders of the EV-B93 3C protease. Validation and profiling of the most promising non-covalent hits were done using thermal shift assay (TSA), surface plasmon resonance (SPR), and proteolytic activity assay. 44 analogs of the most potent molecule were evaluated in the in-vitro proteolytic activity assay. The most active compound displayed IC50 value of 5 flM. Further chemical optimization was performed resulting in more efficient inhibitor with similar IC50 value. Selected analogs were tested in the in-vitro proteolytic assay against analogous 3C proteases from the following viruses: human rhinovirus-A49, enterovirusD68, aichivirus A, porcine sapelovirus, and equine rhinitis B virus. All compounds exhibited good inhibitory activity against three of the tested proteases. Furthermore, in a cell-based proteolytic assay and an antiviral assay the compounds did not exhibit either proteolytic or antiviral activity, which may be explained by several factors such as lack of cell permeability, low solubility and/or high toxicity. Extensive co-crystallization and soaking trials were performed to obtain crystal structures of noncovalent complexes of the EV-B93 3C protease with the most potent compounds. Regrettably, no additional electron density was identified in the proteolytic active site. Bioinformatics docking simulations suggested potential binding mode of the optimized compound. These pointed to the presumed pockets occupied by the compound that interact with the two conserved residues from the catalytic triad. Since the most potent compound is a relatively large and rigid molecule, it is unable to bind to the protease without its previous rearrangement, which is unfavorable in the crystalline state of the protein. This observation may explain the inability of the non-covalent molecules to co-crystallize with EV-B93 3C protease. The results obtained in this study may aid the design of potent, noncovalent antivirals targeting enteroviral 3C proteases.