Investigation of the efficacy of designed peptides targeting the inhibition of SARS-CoV-2 Spike–ACE2 interaction using molecular modeling methods

Viruses Viruses are infectious agents that can only replicate within living (host) cells. In this context, the SARS-CoV-2 virus causes COVID-19 by entering host cells through binding of its surface Spike (S) protein to the angiotensin-converting enzyme 2 (ACE2) receptor. Since ACE2 is highly expressed in tissues such as the heart and lungs, disruption of its function by the virus has contributed to the death of over 7 million people globally during the COVID-19 pandemic, according to data from the World Health Organization. The high mutation rate observed particularly in the region of the SARS-CoV-2 genome encoding the Spike protein increases the virus's transmissibility and its ability to evade the immune system, thereby enhancing its lethality. As a result, the Spike–ACE2 interaction has become a major target in drug development against the virus, with increasing focus on inhibiting this interaction using peptides. Compared to small molecules, peptides offer significant advantages in targeting protein–protein interactions due to their flexible structures, specific binding capabilities, and low toxicity. However, the high cost and lengthy timelines of traditional drug development approaches have necessitated the use of computer-aided drug design (CADD) strategies in the early-stage screening of peptide candidates. Despite advances in this field, discrepancies have been observed between computational and experimental data evaluating the efficacy of peptides developed to inhibit the SARS-CoV-2 Spike–ACE2 interaction. Inhibitor design efforts targeting this interaction have involved structural prediction and molecular docking using programs such as HADDOCK and AutoDock CrankPEP (ADCP), as well as molecular modeling tools like AlphaFold2 (AF2), AlphaFold2-Multimer (AF2M), PepFold3, and ESMFold. Among these, AF2M demonstrated superior performance in accurately predicting both the binding site and the post-binding conformation of the peptide, compared to other methods. Consequently, the effects of different scoring functions such as Rosetta Ref2015, Rosetta ddG, Prodigy, and PyDock were analyzed on AF2M-modeled structures. It was observed that ranking the AF2M-modeled structures based on the Rosetta Ref2015 score yielded meaningful results in terms of peptide positioning on the receptor Spike protein. However, correct positioning at the binding site does not necessarily confirm inhibition of the interaction. Therefore, it has been proposed that future studies should expand the dataset and integrate multiple scoring functions. Moreover, conducting molecular dynamics (MD) simulations at the microsecond scale and calculating binding free energies would be appropriate to evaluate the biological relevance of SAPx peptide–Spike RBD complexes. In conclusion, this thesis demonstrates that employing accurate scoring strategies in AI-assisted modeling approaches for peptide-based drug development targeting the SARS-CoV-2 Spike–ACE2 interaction yields promising results for testing design accuracy. The use of correct scoring strategies in computational methods is considered to contribute positively to solving this complex problem. Keywords: SARS-CoV-2 virus, ACE2 receptor protein, Peptide-protein interactions, Peptide-structure prediction, Molecular docking, Molecular modeling, Molecular dynamics simulations