Team:SZU-China/Model/1
Molecular docking explores the binding modes of two interacting molecules and fits molecules together in favorable configurations according to the principles of geometric complementarity, energy complementation, and chemical environment complementarity. The technique is increasingly popular for studying protein-ligand interactions.
docking
To prepare for the molecular docking, we first need to obtain the 3D structure of the beta-glucosidase and geniposide. The 3D model of beta-glucosidase is offered by the Swiss Model after submitting corresponding nucleotides, and its high-quality 3D image is shown with Pymol (Fig. 1), the open-source molecular visualization system.
Fig 1. 3D structure of β-glucosidase shown by Pymol
The 3D structure of the ligand molecule, geniposide, was drawn via the molecular simulating software, Discovery Studio, and optimized with the more powerful and versatile simulating program, CHARMM (Chemistry at HARvard Molecular Mechanics), to obtain its 3D model with the lowest molecular energy state (Fig. 2).
Fig 2. 3D structure of geniposide drawn by Discovery Studio
The 3D structure of the ligand molecule, geniposide, was drawn via the molecular simulating software, Discovery Studio, and optimized with the more powerful and versatile simulating program, CHARMM (Chemistry at HARvard Molecular Mechanics), to obtain its 3D model with the lowest molecular energy state (Fig. 2).
Early methods on protein-ligand docking treat the target proteins as rigid molecules which may lead to large deviations in the result. And also, in our cases, geniposide can’t even get into the catalytic cavity of β-glucosidase using the simple docking method due to their rigid conformations and the relatively large size of the ligand.
With the development of computer performance, recent advances have also appeared in algorithms dealing with protein flexibility. Therefore, to fit our ligand into the catalytic cavity and gain more reliable results, we used flexible docking to allow a certain degree of conformational change of our protein side-chain and ligand. However, one thing deserves mention is that while flexible modeling can reveal a much more near-nature result, it demands longer running time due to its more complex program.
Firstly, we needed to locate the receptor's binding sites, among which we found potential key sites and converted them to be flexible for successful docking. The determination of the receptor binding site is very important for the accuracy of molecular docking and scoring results. As the model of beta-glucosidase, we downloaded from the swiss-model has already have accompanied co-crystal ligand implicating the potential active center, we could directly check and set corresponding amino acid residues to be flexible without struggling form the very beginning ( Fig. 3).
Fig.3 Lligand-protein interaction diagram
The binding mode between beta-glucosidase and its original ligand is shown in figure 3, revealing the key sites for interaction. The left picture is a 3D model, while the right one is a corresponding two-dimensional planar view. The dotted green line indicates the hydrogen bond between amino acid residues and ligand.
In order to scale down the calculation complexity, here we only set GLN20, HIS121, GLU405 and GLU406 as flexible amino acids. And after setting of flexibility on critical sites, we finally got a flexible model of beta-glucosidase that can be docked with our geniposide.
and geniposide
The flexible docking reveals 20 different conformations of geniposide. The bar charts below show how many kinds of geniposide with different conformations have interaction with the chosen key residues according to the four respective interaction type.
Fig 4. Distant The result of flexible docking: residue interaction histograms
It can be seen that the top five amino acid residues that generate hydrogen bonds are ASN246, GLU166, ASN223, HIS298, and SER296, and the top five amino acid residues that generate hydrophobic interactions are TRP324, VAL173, TRP406, HIS180, and HIS298.
Site-directed mutation of protein amino acids can be used in the design of enzymes and antibodies, but the efficiency is low due to the blindness in amino acid selection. Virtual amino acid mutations can determine the best combination of amino acid mutations through alanine scanning and saturation mutations, to guide amino acid site-directed mutations in experiments.
To obtain more detailed information about amino acid mutations to improve enzyme activity, we performed saturation mutations on the key amino acids (GLN20, HIS121, GLU166, GLU351, and GLU405), that is, these five amino acids are mutated to one of other 19 amino acids.
Fig 5. Interaction diagram of key amino acids and geniposide
It can be seen from the results that when GLN20 mutates into HSC, HIS121 into LYS, TRP or HSC, GLU166 into GLUH, HSC, TYR or GLN, and GLU351 into HSC, GLUH, TYR, HIS or PHE, the affinity between protein and ligand are improved.
Table 1. Saturation mutation results of key amino acids
In the future, we will synthesize new DNA according to the results of virtual amino acid mutation, and conduct wet experiments to verify whether the catalytic efficiency of beta-glucosidase is improved, and help orient experimental design.
