ENHANCED MEAN-FIELD SIMULATIONS OF ANTIBODY CDR LOOPS

重要生物分子（如蛋白质和核酸）精确的三维结构的可用性可以极大地帮助理解它们的功能，以及合理设计药物化合物。然而，结构数据库相对较小。结构通常使用比较建模来构建。但这些在环路区域通常不太准确。将开发一种新的方法来预测蛋白质中高精度的环构象，并具体应用于抗体CDR环。这种平均场方法将大大增加交替构象的采样，同时保持与最先进的模拟技术（如显式溶剂化中的分子动力学）的兼容性。单个水分子通常具有结构作用，用连续体模型代替可能会导致无法接受的精度损失。与目前使用的方法相比，这是该方法的一个关键优势。由于该方法是基于能量的，它比使用数据库的方法更通用，并且将改进结构建模的许多领域。该项目将包括发展一种增强的本地增强抽样方法。这将应用于预测抗体环中更困难的问题。最初的研究将集中于重现已知的重要H3环的诱导拟合案例，然后对H3构象未知的系统进行真正的预测。这些将为诱导契合的决定因素、溶剂的作用和几种具有生物医学重要性的抗体的关键结构细节提供新的见解。后续研究将预测几种催化抗体的抗体CDR区域的所有6个环。这些结构将有助于理解影响这些抗体催化活性的因素，以及效率如何受到环柔韧性、溶剂分子或抗体所针对的过渡态类似物的缺陷的影响。该方法不仅可以预测抗体的构象，还可以预测抗体-抗原复合物的结构。这些信息对于理解这些重要的生物分子相互作用至关重要。

The availability of accurate three-dimensional structtires for important biomolecules such as proteins and nucleic acids can greatly aid in the understanding of their fimction, as well as in the rational design of pharmaceutical compounds. However, the structural database is relatively smaii. Structures are often built using comparative modeling. but these are typically less accurate in loop regions. A new approach to predict highly accurate loop conformations in proteins will be developed, with specific application to antibody CDR loops. This mean-field method will dramatically increase sampling of alternate conformations while maintaining compatibility with state of the art simulation techniques such as molecular dynamics in explicit solvation. Individual water molecules often have structural roles, and replacement with a continuum model may result in an unacceptable loss of accuracy. This is a key advantage of the method over those currently in use. Since the method is enezgy-based it is more general than those using databases, and will improve many areas of structural modeling. The project will consist of development of an enhanced locally Enhanced Sampling approach. This will be applied to successively more difficult problems in prediction of antibody loops. Initial studies will focus on reproducing known cases of induced fit for the important H3 loop, followed by true prediction for systems in which the H3 conformation is unknown. These will provide new insights into the determinants of induced fit, the role of solvent and key structural details for several antibodies of biomedical importance. Subsequent studies will predict all 6 loops in antibody CDR regions for several catalytic antibodies. These structures will aid in understanding the factors that influence the catalytic activity of these antibodies, and how efficiency is aifected by loop flexibility, solvent molecules or deficiencies in transition state analogs against which the antibodies are raised. The method will be extended to the prediction not only of antibody conformations, but also the structures of antibody-antigen complexes. This information can be critical to the understanding of these important biomolecular interactions.