Charakterisierung der antiviralen Aktivität von Rocaglamid-Derivaten und Validierung der humanen RNA-Helikase eIF4A als Target

The ongoing COVID-19 pandemic (coronavirus disease 2019) shows the urgency for the development of new antiviral drugs. In this context, an antiviral compound with broad-spectrum activity could help to combat newly emerging viral infections. For such an approach an appropriate target, as well as specific drugs that efficiently inhibit the target and thereby viral replication are required. Since RNA viruses are able to adapt to new environmental conditions within a very short time through mutations in their genome, they might develop resistance to the drugs. A suitable target could therefore be a host factor, such as the eukaryotic initiation factor 4A (eIF4A), which is used by a large number of viruses for their protein synthesis in infected host cells. In cancer research, eIF4A is already a well characterized target, but its validation as an antiviral target is still rather incomplete. A focus of this thesis was therefore the validation of the RNA helicase eIF4A as an antiviral target. During translation initiation, eIF4A unwinds secondary structures within the 5’-UTR of a mRNA to enable protein synthesis. Since many viral mRNAs have highly structured 5‘-UTRs, viral protein synthesis often depends on this helicase activity. Inhibition of eIF4A would consequently inhibit the synthesis of viral proteins and thus stop the viral replication in infected cells. Previous cell culture experiments have already shown that the natural eIF4A inhibitor Silvestrol has an antiviral activity at low nanomolar concentrations against several pathogenic RNA viruses. Therefore, it was another focus of this work to identify new specific eIF4A inhibitors and characterize their antiviral activity. For this purpose, the cytotoxicity of potential synthetic eIF4A inhibitors was determined in a human cancer cell line. Then a reporter assay (dual luciferase assay, DLA) was carried out to validate the eIF4A specificity of the compounds. In addition, the binding of the ligands to recombinantly purified human eIF4A was characterized using a thermal shift assay (TSA) established in this work. With this approach the synthetic rocaglate CR-31-B (-) could be identified as a potent eIF4A inhibitor with comparable antiviral activity to Silvestrol. In addition, the rocaglate HGW-3 showed an activity at medium nanomolar concentrations in the DLA. In order to predict a possible eIF4A dependency of viral protein synthesis based on the mRNA composition, various RNA structure and sequence elements were analyzed in the DLA. In addition to viral 5‘-UTRs, variations of viral hairpin structures or unstructured RNA motifs with specific sequence patterns were also investigated. Moreover, the natural rocaglate Silvestrol was compared to the synthetic rocaglate CR-31-B (-) in the DLA. It was found that for CR-31-B (-) a polypurine stretch within the investigated RNA substrate is required, while with Silvestrol also RNA substrates without polypurine sequences but with RNA hairpin structures can mediate eIF4A dependency. This suggests differences in the molecular interaction mechanism. To understand the mode of action of rocaglates in more detail, the interaction sites of eIF4A with the substrate RNA and the eIF4A inhibitors Silvestrol and CR-31-B (-) were investigated using hydrogen-deuterium exchange mass spectrometry (HDX-MS). Here, the interaction site that was postulated for RocA with eIF4A and the polypurine RNA could be confirmed. Nevertheless, no differences were found between Silvestrol and CR-31-B (-). Using a microscale thermophoresis assay (MST), a dissociation constant for the binding of the two ligands to eIF4A could be determined. Finally, based on a published crystal structure of eIF4A (PDB: 5ZC9), docking experiments could be performed and the binding mode and the binding affinity of potential eIF4A inhibitors could be predicted. A possible binding pocket consisting of arginine residues could be identified through which an additional interaction with Silvestrol might be formed. This would explain differences in the mode of action between Silvestrol and CR-31-B (-). Furthermore, the virtual mutation of eIF4A enabled the prediction of presumably rocaglate-sensitive and -resistant eIF4A variants, which can be found in various human pathogenic organisms.