Protein Molecules Conformation Recognition And Their Interactions Characterization Based On Terahertz Spectroscopy

Due to the characteristics of low energy and coherence,terahertz radiation can record the changes of electric field amplitude and phase without destroying protein molecular structure.Moreover,it has been confirmed that the skeletal vibration,dipole rotation and other collective vibration modes play a leading role in the protein structure and function,and the corresponding dynamic energy levels just fall in the terahertz frequency range.Therefore,terahertz spectroscopy is a label-free and non-destructive detection method for protein structure and function expression.In this dissertation,protein molecules conformation and their interactions are characterized and analyzed by transmission terahertz time-domain spectroscopy(THz-TDS).Protein samples characterized in this dissertation refer to solid phase and liquid phase samples,and the spectral information studied here involves the absorption spectra and dielectric spectra.The main contents of this dissertation are as follows:1)The terahertz absorption spectra of bovine serum albumin(BSA)solid samples heated to different temperatures and pepsin solutions in various pH conditions are characterized by THz-TDS.The results show that,for the absorption spectra of protein in various conformational states,the differences in the height can sensitively reflect the secondary structure changes.This study proves that terahertz spectroscopy can be used to characterize the folding and unfolding of protein structure caused by changes in environmental conditions,which provides a theoretical basis for non-destructive and label-free detection of biomolecular conformation changes in the application fields of protein drug quality control.2)Most proteins have no characteristic absorption peak within terahertz band.Although THz-TDS can be utilized to detect the conformational changes by investigating the absorption curve height difference between the sample and the protein in the natural conformation state,the specific identification rule between different conformations and their corresponding terahertz absorption spectra cannot be obtained.Therefore,taking BSA solid samples with various conformations induced by different temperatures as an example,its sample size is expanded.Then,the original terahertz spectral information of all samples is trained by machine learning model combined with dimension reduction algorithm.Finally,the exact mapping relation between different conformational proteins and their absorption spectra is established.At the same time,it's discovered that the extreme gradient boosting model(XGBoost)combined with t-distributed stochastic neighbor embedding algorithm(t-SNE)has better prediction result than the general model.In this part,THz-TDS and machine learning model are combined to realize the recognition and prediction for the conformation of the single given protein sample.3)Firstly,in the characterization of intermolecular interactions,the terahertz absorption characteristics of BSA solutions at various concentrations were analyzed.It is observed that the relationship between absorption coefficient and concentration doesn't conform to Lambert-Beer law.This phenomenon verifies the existence of hydration layer between protein and water molecules.Secondly,the dynamic process of BSA hydrolyzed by pepsin is detected by THz-TDS over a time span.On the one hand,based on the absorption characteristic analysis,it is found that the change in the absorption coefficient during hydrolysis is time-dependent.On the other hand,based on the dielectric characteristic analysis combined with Debye model,it has shown that the hydrolysis process would affect the types of water molecules and the hydrogen bond grid layout in solution.In this part,a method combining absorption characteristic analysis with dielectric characteristic analysis based on Debye model is proposed.Besides,the monitoring of protein hydrolysis process has been achieved both from macroscopic perspective and at molecular level,which provides a novel characterization method for the research on biomolecular interaction.4)For the dynamic process of protein-protein interaction at millisecond level,this dissertation has explored a high-speed detection method of terahertz spectrum based on Michelson interferometer.From the analysis of structural device,spectrum calculation,parameter selection and simulation,this study theoretically demonstrates that this high-speed method could realize terahertz spectrum detection based on interference phenomenon.Besides,it could significantly accelerate the terahertz spectrum acquisition rate by improving the mobile speed of moving mirror in the theoretical structure,and satisfy the terahertz spectroscopy characterization requirements of dynamic process at millisecond level.