基于蛋白相互作用指纹的抗原表位表征及抗体筛选模型研究

A number of diseases are difficult to be prevented and controlled effectively, which is mainly because of the frequent amino acids mutations and structural variations on the epitopes of disease-related antigens. Previous studies are not good at the epitope descriptions from respects such as antigen-antibody interface and interaction force types. In our previous research, interaction fingerprint, which combined space geometry, interaction energy decomposition, robustness and high computing efficiency, is better at the representation of protein active site than other methods. Based on the interaction fingerprint and the characteristics derived from antigen-antibody structures such as 3-amino acid propensities of antigens and aromatic islands on antibodies, this project will develop a new method to represent antigen epitope based on protein interaction fingerprint. Also, our previous work has shown that, the proteochemometric model constructed by interaction fingerprints not only obtains higher predictive ability, but also has more easily interpretation. With the newly developed protein interaction fingerprint, which will be further used as the cross-term, an antibody design model for antigens would be constructed by proteochemometric modeling methods. This model will achieve specific antibody screening and broad spectrum antibody screening at the same time. This project would propose a new method for antigen epitope representation and antibody screening. It is expected to be effective in the prevention of antigen-related diseases.

抗原表位上的氨基酸序列和结构变化是很多相关疾病难以有效防治的主要原因。以往的研究未能很好的从"抗原-抗体作用界面"和"相互作用力类型"等方面对抗原表位进行描述。我们前期研究表明：相互作用指纹因同时具有"空间几何"、"相互作用分解"、"鲁棒性"和"高运算效率"等特点，能更好地表征蛋白活性位点。本课题在此基础上，结合申请人所在课题组发现的抗原"3-氨基酸倾向性因子"和抗体"芳香岛"等特性，开发一种基于"蛋白相互作用指纹"的抗原表位表征新方法。前期工作还表明：以相互作用指纹为交叉项建立的蛋白质化学计量法模型不仅预测能力更好，也更容易进行解释，为此，本课题进一步以新开发的"蛋白相互作用指纹"为交叉项，运用蛋白质化学计量法构建病毒抗原的抗体筛选模型，同时实现特异性抗体筛选和广谱性抗体筛选两个目的。本课题的实现将为抗原表位表征和抗体筛选模型提供新方法，有望有效地解决抗原相关疾病防治难题。

抗原表位上的氨基酸序列和结构变化是很多相关疾病难以有效防治的主要原因。抗原抗体间存在特异性的结合，但同时，相似的抗原表位能够被相同抗体的对位区域交叉识别。因此，如何表征这个微小的变化并准确的衡量出这个变化是否对抗原抗体之间的识别产生影响是该领域的研究重点。以往的研究未能很好的从“抗原-抗体作用界面”和“相互作用力类型”等方面对抗原表位进行描述。前期大量系统性的研究表明：相互作用指纹因同时具有“空间几何”、“相互作用分解”、“鲁棒性”和“高运算效率”等特点，能更好地表征蛋白活性位点。.在此基础上，我们开发了一套高效的描述符，分别描述了抗原侧，抗体侧以及相互作用界面形状，理化特征等特点。同时，我们将这套蛋白描述符及蛋白相互作用指纹运用到蛋白质化学计量法模型中，建立了一个针对抗原抗体复合物的蛋白计量化学模型。研究表明，该模型不仅预测能力优于现有的蛋白描述符，同时，以蛋白相互作用指纹作为交叉项也更具有可解释性。该模型的建立将有助于实现特异性抗体筛选和广谱性抗体筛选，同时能为高通量的抗体筛选提供有效的指导和帮助。