| Antibodies are immunoglobulins produced by plasma cells transformed from B lymphocytes under the interaction of antigen and immune system,which can specifically bind to the corresponding antigen.When antibodies bind to antigens,the CDR regions of the antibody bind to a region of the antigen,the antigen determinant,which is also called as epitope of antibodies.The region of antibodies which binds to the epitope is called as paratope.Epitopes are divided into linear epitopes and conformational epitopes.The linear epitopes consist of a series of consecutive amino acids,and the conformation epitopes are usually composed of a number of amino acids which are discrete on the primary sequence but close to each other in the spatial structure.Therefore,changes in antigenic structure may significantly affect the conformational epitope,but the effect on the linear epitope is not significant.Epitope of antibodies is one of the most important properties of antibodies,but also the most desired information to researchers.By studying the epitopes of antibodies,we can get the protective mechanism of antibodies,and study the pathogenic mechanism of pathogens,and can reverse research vaccines which could induce the protective antibodies.The current method of studying antibody epitopes is identified by experiment.Some of these methods require special experimental conditions and equipment,some methods are of a large workload,and some methods are difficult to succeed.With the improvement of computer performance and simulation method,there have been some molecular simulation methods in biology,which could be used in simulation of the biological macromolecules.The main features of these methods are of the low requirements of the experimental conditions and equipment,and most calculations are completed by the computer.It could provide a clear guide to the experimental design and a reasonable explanation to the experimental phenomenon.Researchers are paying more and more attention to this approach.The main purpose of this study is to establish a method of identification antibody epitopes by computer modeling,molecular docking and other methods,to predict antibody epitopes,guide the experimental design,and identify the antibody epitopes quickly and easily.This method is characterized by low threshold,that predicting epitopes requires only antibody sequences and antigen crystal structure with no experimental requirements.Of course,the predict results need to be verified by the experimental approaches.However,the computer’s prediction results provide us a goal to design the experiment.With this goal,the prediction could be verified through a simple mutation experiment.It greatly predigests the difficulty of epitope identification.In this paper,we identify the epitope as the following steps.(1)The molecular structures of antibodies are established with the antibody primary amino acid sequence by the Discovery Studio software.(2)Most of the crystal structures of the antigens have been obtained by X-ray crystal diffraction,so the corresponding antigen crystal structure is downloaded from the PDB database;(3)ZDock is performed by using the molecular structure of the antibody structures obtained from the simulation and the antigen structures downloaded from the PDB database.(4)Predict the critical amino acids in the interaction of antibodies and antigens by the Residues Contact Frequency(RCF)algorithm.(5)Verify the predict results by experiments.We selected 22 pairs of antigen-antibody pairs from Docking BenchMark 5 as test sets to validate the effectiveness of these predictive methods.Firstly,we validated the modeling of the antibody structure by Discovery Studio software.The 22 antibodies were modeled,and the models were compared with the real molecular structure.We found that the accuracy of antibody modeling is high enough.Secondly,we validated the prediction of RCF algorithm to analyze the ZDock results.RCF is a method to predict critical amino acids in protein-protein interactions by statistical analysis of ZDock results.We implemented the RCF algorithm in the WorkScript window of DS software by the Perl language.We designed three optimizations for the RCF algorithm based on the particularity of the antibody structure: 1.RCF analysis was only considered with the atoms in CDR regions of the antibody,instead of all atoms of antibody;2.The pose of ZDock prediction was filtered by the angle of the antigen-antibody;3.During the calculation of RCF,the score of each pose multiplied a weight function-cos θ,where θ the angle of the antigen-antibody molecule.We analyzed the performance of RCF algorithm and the three optimizations in the 22 pairs of antigen-antibody.The results showed that the RCF algorithm and the three optimizations were able to predict the critical amino acid sites of the interaction,and the three optimizations performed better than the RCF algorithm.The performance of the three optimizations were not distinguished.So,we chose the first optimized RCF algorithm for subsequent prediction.After verifying the predictive validity of antibody molecular modeling and RCF optimization algorithms,we apply this method to specific antibody epitope analysis.F2H5 is a murine antibody of Yersinia pestis F1 protein provided complete protection,which prepared by hybridoma technology.The laboratory has humanized F2H5.We first experimentally confirmed that F2H5 bound to F1 in Western Blot and ELISA.We synthesized the F1 protein overlapping peptide library to identify the epitope of F2H5.However,it was surprising that all polypeptides did not bind to F2H5 antibodies.Therefore,we identified the epitope by the method of prediction above,and validated experimentally.First,we used the DS software model the F2H5 antibody.The crystal structures of the F1 protein were downloaded from the PDB database.Five structures with high resolution were selected.The structure of the F2H5 was docked with F1 structures.Five result files,ZDockResults.dsv,were generated.RCF optimization algorithm were used to analyze these five results.The predicted results indicate that F96 and E105 may be epitopes of F2H5.Based on this result,we performed alanine-scanning mutagenesis of F1 protein at position 95-111.The results shown that F1-G104 A,F1-E105 A and F1-N106 A could not bind to F2H5 in both ELISA and Western Blot,and F1-K101 A and F1-N103 A bound F2H5 weakly.We identified the epitopes of F2H5 antibody by RCF optimization algorithm.After the identification of the epitope of F2H5 antibody,we used the epitope,G104E105N106 to filtered the pose produced by ZDock,and selected the pose with the highest ZRank score from the remaining poses as the structure of F2H5-F1 complex.We further optimized this pose in the DS software.According to the optimized pose,we calculated the mutation energy of alanine mutations at the 95-111 position of F1.This results were in agreement with alanine-scanning mutagenesis experiment.We further analyzed the critical amino acid sites on the antibody.In the RCF optimization algorithm,Y170 and Y214 are predicted to be critical to the interaction.At the same time,we used another algorithm which was used to capture antibody affinity enhance mutations,Amino Acid Interface Fitness(AIF),to analyze the interaction between F2H5 and F1.AIF also indicated that these two residues were the most critical site,and tyrosine is the most suitable amino acid at both sites.Based on the structure of the complex,we calculated the mutant energy of the saturation mutations at all sites of CDR2 and CDR3,and selected 20 mutants with a significant energy changes to be verified experimentally.The affinities of 11 mutants were predicted to weaken,which was consistent with the experimental result.Of the remaining nine mutants,whose affinities were expected to enhance,five mutations bound to F1,and two of them,F2H5-D218 R and F2H5-D218 Y,actually had a lower EC50 than native F2H5.We established a computer-based method for predicting antibody epitopes based on ZDock molecular docking and RCF optimization algorithms.We used this method to predict epitopes of a Yersinia pestis F1 protein antibody and to validate it experimentally.The antigen-antibody complex structure was obtained by computer analysis and the affinity enhance mutations of the antibody were captured.Our results suggest that it is effective and meaningful to apply the computer-aided methods in identifying antibody epitopes. |
