Prediction And Screening Of HA Epitopes In H5N1 Subtype Virus

Since the outbreak of H5N1 subtype of avian influenza, the poultry industry has suffered a huge loss. Highly pathogenic avian influenza virus infection causing the deaths of people continues to occur. Recombinant subunit vaccine has broad prospects in aspects of immunization control and can prevent immune failure. caused by genome variation.Thus, the problem of accuracy and quickness of epitope selection becomes more prominent.Based on the sequences of H5N1 from 1996 to 2007 ,mutation site of HA gene is identified by multiple sequence alignment,the phylogenetic tree is constructed.These bioinformatics methods can help us track the mutation of the virus, design timely antiviral drugs or vaccines and avoid blindness in triditional experiments.Multiple sequence alignment of 114 HA gene of H5N1 was conducted by ClustalW, It shows that 193 mutations have occurred in a total of 573 amino acid sites,with a mutation rate of 33.68%.Statistics for mutation in each site is obtained,and it can be seen that a large variation happened over the years,thus we deduce that epitopes which contain theses mutation sites are also constantly changing.Then PHYLIP software package is employed to construct the evolutionary tree,evolution relationship of HA gene over the years can be seen from the tree,and it illustrates obvious clustering according to age.There is a lot of algorithm of epitope prediction, each algorithm has its own advantages and disadvatages, and 100% accuracy can not be achieved. In our study, a liner epitope prediction based on sequences and a non-liner epitope prediction based on 3D structure are established.For non-liner epitope prediction, a variety of algorithm was combined to improve the accuracy of prediction. Goose/Guangdong/1/96 is the common ancester of isolates in mainland China, and can be selected as the virus model for prediction and screening of epitopes of avian flu HA protein. First of all, Kolaskar & Tongaonkar and the template Bepipred algorithm was employed seperately to predict epitopes of Goose/Guangdong/1/1996 (Gs / GD). The result is: Kolaskar & Tongaonkar algorithm for epitope prediction of 24, Bepipred algorithm for epitope prediction of 22 . Based on the prediction results of two algorithms above, epitope screening is conducted to improve the accuracy of prediction, which makes the epitopes predicted more suitable for epitope peptide antigen design and improve the success rate of the experiment.In general, the ideal antigenic recognition sites should have hydrophilic region, be located on protein surface and possess structural features such as being easy to deformation. in the secondary structure of protein,β-angle than theα-helix,β-sheet, are more likely to be epitopes.Based on these principles, the parameters of Hopp & Woods hydrophilicity prediction, Janin accessibility, Karplus flexibility and Chou-Fasman secondary structure are employed to analyze physical and chemical properties of the template of the virus strains.19 hydrophilic regions,20 surface accessbility regions,22 flexibility reagions are obtained. Secondary structure ofα-helix,β-sheet,β-angle is predicted in the template of protein. In order to show the priority of the epitope predicted, an evaluation criteria is created according to the results of prediction and screening,and points is made to the epitopes predicted. Prediction results of Kolaskar & Tongaonkar Antigenicity, Bepipred linear Epitope Prediction Algorithm and sencondary structure are measured respectively as 1 point.Hydrophilic, surface accessibility, flexibility are counted as a whole of 1 point,hence the total score is 4 point.Asa result,6 epitopes with a score above 4 points is obtained, the highest scoring epitopes are 95-109 and 203-218.Combined with previous study of mutaation sites of HA protein sequence of H5N1 subtype of avian influenza,it can be found that all epitopes predicted have experienced variations of at least three sites.For non-liner epitope prediction, the related references and experience of our lab are emphasized, CN3D and Disco Tope are employed to do the prediction.3 possible non-liner animo acid sequence are obtained, and their gene fragments are synthesized. Then, we constructed the encoding gene of vaccine by clone.Our essay constructs liner and non-liner prediction process of epitope using bioinformatics method. The liner prediction employs comprehensive analysis of antigenic region and multiple parameters evaluation, and is more accurate than single parameters evaluation. The non-liner prediction combines lots of methods and experience, and is therefore more convincible. Once the two sets of prediction process are confirmed by function experiment, they can provide the information of epitope immediately, thus avoid the huge loss by failure of the vaccine.