The goal of this model is to determine which amino acids in the structure of VIM2 MBL interact with the structure of meropenem when the molecule is docked in the enzyme.
To describe the nuances of this process, we have to give a simplified explanation on how protein synthesis in our bodies and in living cells in general works. The information needed for the construction of all cells is stored in special molecules called DNA inside of them. One can view DNA as the construction manual for tiny machines which do particular processes in the cell - the proteins. If in this particular copy of the manual there are small typos (mutations) in the DNA instructions, sometimes these typos lead to a catastrophic failure in the protein functionality. However, in rare cases they can improve the protein activity by making it easier to interact with other molecules in the body.
Biology has achieved tremendous success in understanding the way the cells function. Now we can introduce deliberate “typos” in the DNA structure of certain cells to enhance the activity of specific proteins. In our case, this is VIM-2 metallo beta-lactamase - a protein which can neutralise the antibiotic meropenem and other members of the carbapenem antibiotic class.
In order to avoid the construction of an enormous number of random ineffective proteins, we need to identify the places of interest in the tertiary protein structure. In a sense we have to work backwards from the VIM-2 MBL interaction with meropenem, then determine the sites in the protein structure which we want to change and then determine the necessary changes to the DNA structure to implement them.
The interaction of a protein and substrate is subject to an extensive amount of computational methods [1]. On the one hand there are the molecular dynamics techniques [2], where molecular movements are simulated in high detail. Their main disadvantage there is having a tremendous computational cost which often requires the use of a supercomputer, and also the need for a lengthy time period to produce results. On the other hand lie the statistical approaches based on the generation of system description that is afterwards used in a statistical method [3]. Their main drawback is the necessity for a great data volume which is more often than not, hard to come by. In contrast, the molecular docking approaches, one of which is used in this project, offer a balance between resource efficiency and computational cost [4].
Molecular docking is often compared to a matchmaking process. The main goal of the technique is to find a configuration between a large and immovable molecule called a receptor (usually a protein) and a small and flexible one called a ligand (the substrate). Each configuration is compared using its energy calculated by a specific function, called cost function. This function takes into consideration a variety of interactions between atoms in the receptor and ligand structures.
For the docking procedure were used the packages AutoDockTools [5] and AutoDockVina [6]. The specific Force Field used in the process is AutoDock4 Zn [7]. Due to the accuracy increase associated with hydrated docking and the ease to implement it in the model, the hydrated rigid docking method [8] implemented in AutoDock was also employed. All of the protein visualization shown are done with UCSF ChimeraX [9]
The preparation of the ligand was done by the traditional method, described in the manual of AutoDock Vina. It was important to check whether the automatic tools have assigned the correct atom types to the atoms of the logan structure.
The docking was done by generating the 45 best poses with exhaustiveness of 1000. The process was carried out on a personal computer with 16 cores all of which were utilised.
In the image provided, you can observe how the meropenem molecule (displayed in orange) interacts with both the two zinc (Zn) atoms within the active site and the surrounding amino acids. These interactions have also been extensively documented in scientific literature, and some of them have been well-detailed and reported [11,12,13]. It's important to mention that in scientific literature, the numbering of amino acids may vary because different protein structures are used. However, these structures are very similar (differing in their RMSD by less than 1 Angstrom), and thus can be treated as essentially the same for our purposes.
The hydrated docking did not show significant difference in the structure of the docked meropenem. Hence, it may be correct to assume that the mechanism of hydration of the meropenem by VIM2 is not rate-bound by the displacement of water from the active cite.
For our project, we aimed to modify specific amino acids in the active site of the enzyme. Our strategy was to focus on amino acids that interact with the meropenem molecule but are not tightly bound to the zinc atoms. This approach ensures that we don't significantly alter the enzyme's overall structure, which could negatively impact its performance.
As shown in the magenta markings on the picture, we selected the amino acids E149, N165, F61, and the neighbouring F61 D62 (not shown in the picture due to the cross-section take) for modification. Their close proximity to each other simplified the subsequent modification process, therefore increasing the over efficiency.
Sliwoski G, Kothiwale S, Meiler J, Lowe EW. Computational methods in drug discovery. Pharmacological reviews. 2014 Jan 1;66(1):334-95.
Filipe HA, Loura LM. Molecular dynamics simulations: Advances and applications. Molecules. 2022 Mar 24;27(7):2105.
Lo YC, Rensi SE, Torng W, Altman RB. Machine learning in chemoinformatics and drug discovery. Drug discovery today. 2018 Aug 1;23(8):1538-46.
Pinzi L, Rastelli G. Molecular docking: shifting paradigms in drug discovery. International journal of molecular sciences. 2019 Sep 4;20(18):4331.
Huey R, Morris GM. Using AutoDock 4 with AutoDocktools: a tutorial. The Scripps Research Institute, USA. 2008 Jan 8;8:54-6.
Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. Journal of computational chemistry. 2010 Jan 30;31(2):455-61.
Santos-Martins D, Forli S, Ramos MJ, Olson AJ. AutoDock4Zn: an improved AutoDock force field for small-molecule docking to zinc metalloproteins. Journal of chemical information and modeling. 2014 Aug 25;54(8):2371-9.
Forli S, Olson AJ. A force field with discrete displaceable waters and desolvation entropy for hydrated ligand docking. Journal of medicinal chemistry. 2012 Jan 26;55(2):623-38.
Pettersen EF, Goddard TD, Huang CC, Meng EC, Couch GS, Croll TI, Morris JH, Ferrin TE. UCSF ChimeraX: Structure visualization for researchers, educators, and developers. Protein Science. 2021 Jan;30(1):70-82.
Garcia-Saez I, Docquier JD, Rossolini GM, Dideberg O. The three-dimensional structure of VIM-2, a Zn-β-lactamase from Pseudomonas aeruginosa in its reduced and oxidised form. Journal of molecular biology. 2008 Jan 18;375(3):604-11.
Yuan C, Yan J, Song C, Yang F, Li C, Wang C, Su H, Chen W, Wang L, Wang Z, Qian S. Discovery of [1, 2, 4] triazole derivatives as new metallo-β-lactamase inhibitors. Molecules. 2019 Dec 23;25(1):56.
Ramachandran B, Muthupandian S, Jeyaraman J, Lopes BS. Computational exploration of molecular flexibility and interaction of meropenem analogs with the active site of oxacillinase-23 in Acinetobacter baumannii. Frontiers in Chemistry. 2023 Feb 23;11:1090630.
Moosavi SS, Madani M, Mirzaie S, Jazani NH. Discovery of a Potent Inhibitor to Overcome Carbapenem Resistance in Pseudomonas aeruginosa Strains via Inhibition of VIM-2 Metallo-β-lactamases. Journal of Health Reports and Technology. 2022 Apr 30;8(2).