穿膜蛋白質之結構預測
No Thumbnail Available
Date
2009
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
穿膜蛋白對於維持細胞間的正常運作具有極大的影響性，但因其難以利用X-ray晶體繞射來獲得高解析度的蛋白質結構，使進行原子等級的動力學研究相當不易。而因為穿膜蛋白位於脂質膜中的結構特性，與近代電腦演算能力的大幅躍進，使得電腦模擬成為研究蛋白質摺疊動力學的一大利器。
我們自PDB資料庫中匯出兩種retinal proteins－bacteriorhodopsin (BRD)和halorhodopsin (HRD) 結構中的七根穿膜螺旋之胺基酸序列，並配合AMBER套裝軟體演算出適合的二級結構，並取出其中的Cα原子座標來做為我們模型中的α螺旋結構。隨後將得到的七根螺旋結構隨機地插入雙層膜的模擬環境中，再計算模型中螺旋與環境及螺旋與螺旋間的交互作用，並配合蒙地卡羅演算法來獲得一具有最低能量且和PDB原始結構相似的穿膜蛋白質結構。在我們所得的預測結果和PDB原始結構間的均方根偏差值 (RMSD) 部分， BRD 和HRD的模擬結構分別為5.03 Å和6.70 Å。
爾後為了改進模型在預測傾斜角與方位角準度較差的部份，我們在補足結構中所有其他原子後，配合AMBER套裝軟體來進行10 ns的分子動力學模擬。在BRD與HRD的部份皆有效地降低傾斜角、方位角與PDB結構間的差異度，並同時將RMSD值分別降為4.38 Å和5.70 Å。
Although transmembrane proteins have crucial influence for maintaining the normal functions of cells, it’s still lack of their high resolution 3D structures due to the technology requirements of X-ray diffraction crystallography. Therefore the progress of studying their dynamic systems in atomic level is slow. But because of the structural property in lipid membrane and considerable progress in computational technique, it’s comparatively convenient to use computer simulations as a tool of protein folding dynamics research. First of all, we took amino acid sequences of 7 transmembrane helices in bacteriorhodopsin (BRD) and halorhodopsin (HRD) from Protein Data Bank (PDB). Then use AMBER suit software to get the proper secondary structures with desired sequence. After this, we selected all the Cα atoms from them and constructed a Cα-backbone off-lattice model, which means that 7 α helices were randomly inserted into simulated lipid bilayer and they can move free conditionally in the environment. Then we calculated the interaction energy between helices and environment in our model and cooperate with Monte Carlo algorithm to obtain predicted structure with the lowest energy which is also PDB native structure-like. The root mean square deviation (RMSD) between PDB structure and our predicted structure is 5.03 Å for BRD and 6.70 Å for HRD. To make up the deficiency in predicting tilting angle and orientation angle, we used AMBER suit software again to construct an all-atom model and ran Molecular Dynamics Simulations for 10 ns. The difference in angles between PDB and predicted structure is decreased effectively for both BRD and HRD. And at the mean time RMSD is also reduced to 4.38 Å and 5.70 Å respectively。
Although transmembrane proteins have crucial influence for maintaining the normal functions of cells, it’s still lack of their high resolution 3D structures due to the technology requirements of X-ray diffraction crystallography. Therefore the progress of studying their dynamic systems in atomic level is slow. But because of the structural property in lipid membrane and considerable progress in computational technique, it’s comparatively convenient to use computer simulations as a tool of protein folding dynamics research. First of all, we took amino acid sequences of 7 transmembrane helices in bacteriorhodopsin (BRD) and halorhodopsin (HRD) from Protein Data Bank (PDB). Then use AMBER suit software to get the proper secondary structures with desired sequence. After this, we selected all the Cα atoms from them and constructed a Cα-backbone off-lattice model, which means that 7 α helices were randomly inserted into simulated lipid bilayer and they can move free conditionally in the environment. Then we calculated the interaction energy between helices and environment in our model and cooperate with Monte Carlo algorithm to obtain predicted structure with the lowest energy which is also PDB native structure-like. The root mean square deviation (RMSD) between PDB structure and our predicted structure is 5.03 Å for BRD and 6.70 Å for HRD. To make up the deficiency in predicting tilting angle and orientation angle, we used AMBER suit software again to construct an all-atom model and ran Molecular Dynamics Simulations for 10 ns. The difference in angles between PDB and predicted structure is decreased effectively for both BRD and HRD. And at the mean time RMSD is also reduced to 4.38 Å and 5.70 Å respectively。
Description
Keywords
視網膜蛋白, 穿膜蛋白質, 蒙地卡羅演算法, 分子動力學模擬, 蛋白質摺疊, retinal protein, transmembrane protein, Monte Carlo Algorithm, Molecular Dynamics Simulation, protein folding