靶向PD1/PD-L1通路的虚拟噬菌体展示模型构建与实验验证

Phage display is an effective technique to obtain peptides and proteins with specific affinities to any given target. It has been widely used in basic and applied studies on biomedicine. Tumor has become the most serious threat to human health; screening for biodrugs targeting the PD1/PD-L1 pathway has been an increasingly hot topic in the field of the cancer research. Based on our previous experience of bioinformatical and experimental study on phage display, in this project, we want to: (1) develop software for simulating peptide libraries; (2) obtain a lot of antagonistic peptides for PD1 and PD-L1 through 2nd generation sequencing phage display and peptide chip technology; (3) construct optimized models through multiple machine learning methods with various features extracted from the data obtained; (4) screen several virtual peptide libraries computationally against PD1 and PD-L1 model; (5) verify the virtual panning results with molecular dynamics simulation and detection using enzyme-linked immunosorbent assay, quartz crystal microbalance, and surface plasmon resonance; (6)implement two online in silico phage display systems resembling PD1 and PD-L1 respectively. The project described above is trying to build a new paradigm for computer-aided biodrugs screening. The panning results are promising candidates for the development of anti-cancer drugs. Thus, this proposal is scientifically significant.

噬菌体展示是一种淘选具有特定亲和力的多肽或蛋白质的现代分子生物学技术，已广泛用于生物医药相关基础与应用转化研究。肿瘤是人类健康的最大威胁；靶向PD1/PD-L1通路的生物药物淘选是癌症研究的前沿热点。本项目将在我们长期从事噬菌体展示生物信息学与实验研究的基础上，开发肽库模拟程序；通过二代测序噬菌体展示与多肽芯片实验，大量获得PD1、PD-L1拮抗肽；进而采用多种特征与机器学习方法，构建、优化靶向PD1/PD-L1通路的计算淘选模型并对多种虚拟肽库进行筛选；接着通过分子动力学计算及酶联免疫、石英晶体微天平、表面等离子体共振等实验对虚拟淘选结果进行多重校验；最终根据计算与实验反馈，优化整合出在线的虚拟噬菌体展示模式系统，实现PD1、PD-L1拮抗肽的低成本、高效率获取。该研究可望建立一种新的计算机辅助生物药物淘选研究范式，具有重要的科学意义；淘到的多肽可望开发为抗肿瘤新药，具有广泛的应用前景。

噬菌体展示是一种淘选具有特定亲和力的多肽或蛋白质的现代分子生物学技术，已广泛用于生物医药相关基础与应用转化研究。肿瘤是人类健康的最大威胁；靶向PD1/PD-L1通路的生物药物淘选是癌症研究的前沿热点。本项目进一步升级完善了BDB生物淘选数据库，开发了SABinder、PSBinder、PhD7Faster2.0等一系列生物淘选噪声检测方法，集成整合为生物淘选实验结果假阳性多肽检测软件包SAROTUP3.0。我们提供的生物淘选数据库及生物淘选数据检测服务有助于使用噬菌体展示技术进行基础研究与应用开发，具有重要意义。作用实现了虚拟肽库定制程序，搭建了基于链亲和素的虚拟噬菌体展示模式系统框架。以PD1、PD-L1为靶标，通过二代测序噬菌体展示技术获得了大量数据，构建了PD1、PD-L1淘选富集后的多肽数据集。采用二进制编码、单氨基酸组成、二肽组成、CTD、方差分析、主成分分析等多种特征提取与选择方法，支持向量机、朴素贝叶斯网络、K近邻、随机森林、卷积网络等进行机器学习，然而模型性能均欠佳，难以预测PD1富集肽、PD-L1富集肽，未能实现PD1及PD-L1富集肽的低成本、高效率获取。本项目努力推动噬菌体展示技术从体外、体内到计算机的发展，但离建立一种计算机辅助生物药物淘选研究的新范式还有很大差距，需要进一步总结经验教训，深入研究。