Electrostatics Roles In Temperature Adaptation Of Serine Proteases And A Method For Predicting Antigenic Variation In Influenza A H3N2

This thesis was divivded into two main parts:1.Insights into the role of electrostatics in temperature adaptation of subtilisin-like serine proteasesElectrostatics plays an important role in the temperature adaptability of enzymes.To investigate the possible role of electrostatics in different temperature adaptations,in this section we selected the subtilisin-like serine proteases from psychrophilic Vibrio sp.PA-44(VPR),mesophilic Engyodontium album(Tritirachium album)(PRK),and thermophilic Thermus aquaticus YT-1(AQN)as research objects and performed a comparative study on the electrostatic interaction strengths(i.e.,those of salt bridges,salt-bridge networks,and calcium ions)and electrostatic surface potentials using methods of multiple-replica molecular dynamics(MD)simulations and continuum electrostatics calculations.Although the thermophilic AQN contains the least number of salt bridges among the crystal structures of these three proteases,more newly-formed salt bridges were formed in AQN MD simulations than in VPR simulations,and this results in a higher proportion of charged residues that form salt bridges/salt-bridge networks in AQN than in VPR.The electrostatic strength averaged over all salt bridges in AQN,PRK,and VPR is strongest,moderate,and weakest,respectively,at respective organism growth temperatures,indicating that salt bridges as a whole make the largest,moderate,and least contribution to the thermostability of VPR,PRK,and VPR,respectively.Actually,most salt bridges in AQN exhibit more negative electrostatic free energy values and hence aid in maintaining the protein thermostability at high temperatures,while nearly half of the salt bridges in VPR can interconvert between being stabilizing(negative electrostatic free energy values)and being destabilizing(positive electrostatic free energy values)and hence likely aid in enhancing the conformational flexibility at low temperatures.The individual salt bridges,salt-bridge networks,and calcium ions contribute differentially to local structural stability/conformational flexibility of these three proteases,depending on their spatial distributions and electrostatic strengths.The shared negatively charged surface potential around the active center of the three enzymes may ensure the active-center flexibility and hence benefit to nucleophilic attack and proton transfer.In addition,the differences in distributions of the electro-negative,electro-positive,and electro-neutral potentials,especially over the back surfaces of the three proteases,may affect not only the protein stability but also modulate the local conformational flexibility/rigidity.Our results indicate that electrostatics plays roles in both the heat and coldadaptations of subtilisin-like serine proteases through fine-tuning,either globally or locally,the structural stability and conformational flexibility/rigidity,thus laying the foundation for further mutagenesis and engineering studies.2.A method for predicting antigenic variation in influenza A H3N2Predicting antigenic variation in influenza A H3N2 based on the amino acid sequence comparison will be helpful in improving the capability of evaluating the immune efficacy of influenza vaccines and hence the efficiency of vaccine screening.With respect to the hemagglutinin 1(HA1)of flu virus A,its high mutation rate and mutation accumulation can result in continual anigenic drifts and jumps,making it challenging to predict the antigenic property of a novel strain based on amino acid sequences.Although considerable progress has been made in modeling antigenic variation,there is still a room for improvement in the accurate prediction.In this section,we establish an improved method which can enhance the prediction accuracy of antigenic variation in influenza A H3N2 and provide a better balance between the prediction sensitivity and specificity.In this novel method,first,the key candidate positions associated with the antigenic variation was identified by a two-step process:i)ranking the 329 amino acid sites of HA1 region according to their importance in antigenic variation and,ii)obtaining sets of key positions through multiple linear regression analysis of the ranked positions.Subsequently,the obtained key positions were used to infer optimal amino acid distances and the importance coefficients of key positions,which were used for constructing the final linear prediction model.Finally,the optimal prediction model for antigen feature prediction was selected according to the Bayesian information criterion(BIC).11 amino acid positions optimal for modeling the antigenic variation,which were fewer than the number of key positions(18-39)suggested by previous studies.Most importantly,our method exhibits highest prediction accuracy among the existing methods when applied to the validating data set consisting of 31 878 HI(haemagglutinin inhibition)assays.