To understand the molecular components of biological systems, computational and molecular modeling methods have evolved into a close substitute for experimentation. To find new hits for various therapeutic targets, computational techniques including homology modeling, molecular docking, quantitative structure-activity relationships (QSAR), and molecular dynamics (MD) are frequently employed. Recently there has been significant interest in the design of drugs forming a covalent bond with the target protein, with nearly 30% of the marketed drugs targeting enzymes known to act by covalent inhibition. Both non-covalent interactions and the development of a covalent bond between the inhibitor and the target protein give rise to the activity of these kinds of inhibitors. Because of the covalent bond formed between the ligand (which is electrophilic) and the target (which is nucleophilic), covalent drugs typically have much stronger binding affinities with their targets. This leads to stronger potency while maintaining a pharmaceutically preferred small molecule size.
The biological effect lasts longer when there is a covalent interaction with the target protein. However, due to the difficulties of disassociation in the incidence of off-target binding, these inhibitors frequently have a toxic profile. As a result, covalent drugs’ precise selectivity profiles are needed. According to reports, about 33% of covalent medications on the market are anti-infectives, 20% treat cancer, 15% treat gastrointestinal disorders, and 15% are used to treat indications related to the central nervous system and the cardiovascular system.
Many covalent drug design scenarios still heavily rely on crystallographic experiments for information of accurate binding structures. However, this technique is too expensive and time consuming. The only alternative to this approach is to utilize the covalent docking tools to predict the structures and have information of binding sites. However, the barrier posed by the covalent bond between proteins and ligands is shared by all of the available docking tools and their success rate is not encouraging. To bridge this gap between the crystallographic approach and covalent docking tools, a significant advancement in the field has been published in the Journal of Medicinal Chemistry. A group jointly led by Dr. Li Rao, Prof. Jian Wan from Central China Normal University and Prof. Xin Xu from Fudan University developed and validated a hybrid approach known as Cov_DOX that will serve as an alternative and efficient approach in Structure Based Drug Design (SBDD) to identify the structure of the protein-ligand complex.
For comparison, the research team compared their data of 405 ligand complexes with two published reports which utilized 207 ligand sets and 330 ligand sets. Earlier published data on covalent docking tools showed that Cov_Dock’s (58%) and ICM-(53%) pro’s performances are marginally superior to those of GOLD’s (46%) and MOE’s (46%). However, Cov_DOX the tool developed by authors significantly outperforms these covalent docking front runners with a success rate of up to 81 percent for the Top 1 pose prediction. Testing against the combined set of 405 protein-ligand complexes found that Cov_DOX- (GSA Top300) has a sampling success rate of 86%, which is much higher than that of AutoDock4 (58%) GOLD (62%) MOE (67%) CovDock (72%) and ICM-Pro (75%). Investigation team showed that a major fraction of the energetics that defines the binding configuration is accounted for through a covalent bond. However, sometimes X-ray structures may also be inaccurate. According to researchers, Cov_DOX may assist in the crystal structure refinement or revision for the covalent protein-ligand complexes.
The researchers have created a completely revised version of the DOX procedure that is applicable to covalent protein-ligand complexes. A sophisticated sequential combination of three theoretical levels—a coarse level GSA binding pose sampling, a medium level PM7 binding pose refining, and a fine level XO final binding pose ranking—make up the novel Cov_DOX procedure. The novel Cov_DOX protocol, which is based on the funnel-like CSAMP strategy, could fully utilize the benefits of the implemented MM (molecular mechanics) and QM (quantum mechanics) methods, enabling it to provide accurate predictions on covalent protein-ligand binding structures. Researchers discovered that the Cov_DOX prediction ability is accurate and independent of the receptor classes, the warhead types, and the reaction mechanisms by using a bigger than ever covalent protein-ligand complex test set. Authors draw attention to the fact that many structure-based drug design scenarios, such as lead structure optimization, still strongly rely on the accuracy of the X-ray structures.
In summary, the new Cov_DOX is immensely useful for covalent protein-ligand complex experiments where the accuracy of the X-ray structure is desperately needed but not available or/and not affordable., A web server is provided for the free tryout of Cov_DOX protocol, which can be assessed at http://doxwebserver.ccnu.edu.cn/. Previous version of non-covalent DOX could also be assessed at this URL.
Wei L, Chen Y, Liu J, Rao L, Ren Y, Xu X, Wan J. Cov_DOX: A Method for Structure Prediction of Covalent Protein–Ligand Bindings. Journal of Medicinal Chemistry. 2022 Mar 30;65(7):5528-38.