Exploring structure-based discovery of antivirals targeting Human cytomegalovirus (HCMV)

Abstract

Human cytomegalovirus (HCMV) causes widespread infections globally, typically remaining asymptomatic and latent. However, under conditions of immunosuppression, HCMV can reactivate and cause severe disease. The United States Food and Drug Administration (USFDA) has approved six antivirals for prevention and/or treatment of HCMV infection: Ganciclovir, Valganciclovir, Cidofovir, Foscarnet, Letermovir (only prophylaxis), and Maribavir. Despite their efficacy, these antivirals face one or more of the following challenges: toxicity, efficacy, and the emergence of drug resistance. This highlights the urgent need for novel antivirals that are both effective and safer, targeting essential viral proteins to minimize resistance. This study aimed to extend the available HCMV drugs by establishing structure-based drug design pipeline for three distinct HCMV protein targets, namely the pUL94/pUL99 complex, pUL98 and pUL77. I worked towards the structural characterization of two HCMV targets, employing crystallization methods in combination with crystallization enhancing scaffolds as “crystallization chaperones” as well as alternative methodologies like cryo-EM single particle analysis. The resulting atomic structures will pave the way for virtual screening of small molecules to identify potential antiviral candidates. Additionally, we aimed to establish a semi-automated virtual screening workflow and screen for protein-protein interaction inhibitors against HCMV pUL77. I initially expressed pUL98 in E.coli and purified it to homogeneity. Since initial robotic crystallization experiments remained unsuccessful, I employed sybodies and megabodies as “crystallization chaperones” for pUL98. Unfortunately, crystallizing pUL98 with sybodies or megabodies did not yield diffraction-quality crystals. For the pUL94/pUL99 complex sybodies were already available at the beginning of this thesis, and as crystallization also remained unsuccessful, we started a collaboration with the group of Maya Topf at CSSB; Hamburg, to characterize the complex structure using cryo-EM single particle analysis. To increase particle contrast in cryo-EM, I used the existing sybodies to engineer legobodies binding pUL94/pUL99, but the complex showed significant compositional and conformational heterogeneity, limiting the obtained resolution to ~6Å. Finally, I used the reported cryo-EM structure of the pUL77 pentamer to establish a consensus virtual screening workflow for pUL77 using multiple scoring functions, docking tools and additional screening parameters. This allowed us to identify consensus poses from two docking tools and attempt to improve probability of true-positive hits. Clustering algorithms were used to scrutinize the molecules further, enhancing the hit identification rate and lead selection. The shortlisted molecules from this screening workflow have been ordered and will be tested by our collaboration partners in cell culture models for HCMV infection. In conclusion, this work established a workflow for virtually screening molecules with improved hit rates, moving beyond reliance on docking scores alone. This workflow aims to enhance the efficiency of structure-based antiviral discovery targeting HCMV infection.