Hybrid Methods for Prediction and Design of Novel Human Influenza Antibodies

目前的提案“新型人类流感抗体预测和设计的混合方法” 将识别和开发新型流感抗体。已知流感病毒分为三种类型， 其中甲型流感根据血凝素（HA）进一步分为2组、17个亚型。 甲型流感 H1 和 H3 亚型病毒季节性感染人类。几种抗体的高分辨率结构 甲型流感 HA 已通过 X 射线晶体学测定。参与的抗体 三聚体 HA 分子的抗原多样性头部区域通常宽度较窄。最近，两 已经描述了表现出广泛中和多种流感亚型的抗体类型。这些 两种类型的抗体结合更保守的 HA 干区或保守的唾液酸受体结合 头部区域内的口袋。 H5N1 流感病毒传统上感染鸟类，但最近也造成了几次爆发 仅限于鸟类与人类之间的传播。最近的研究描述了 H5N1 流感病毒的适应性改变 雪貂之间的呼吸道飞沫传播能力，这可能模仿高度进化的未来发展 自然界中致病性大流行的人类 H5 病毒。人们对人类如何接种传统疫苗知之甚少 H5 免疫原可能会受到此类突变体的感染或疾病的保护，这些突变体具有有限的 HA 表面点突变的数量。我们提供有关人类中和隔离的初步数据 H5 头结构域的单克隆抗体，可识别野生型和呼吸道飞沫传播 来自接种传统 H5 HA 蛋白疫苗的人类 H5 HA。此次活动的第一个目标 提议是建立一个“RAPID”管道来快速识别和表征人类抗体 广泛消灭流感病毒——快速应对新出现威胁的重要战略 为了人类的健康。 我们最近证明了一种称为“多状态设计”的新计算方法可以概括 计算机上的抗体成熟，即预测增加抗体与其靶标亲和力的突变，并且 预测编码能够广泛识别多个靶标的抗体蛋白的抗体序列 蛋白质。该提案的第二个目标是设计和测试 (a) 计算机成熟的 RAPID 头结合抗体以增加对 HA 抗原的亲和力，以及 (b) 多状态设计以创建 茎结合抗体可识别多个不同进化枝、群体甚至类型的 HA。

The present proposal “Hybrid Methods for Prediction and Design of Novel Human Influenza Antibodies” will identify and develop novel influenza antibodies. There are three types of influenza viruses known, of which influenza type A is further differentiated into 2 groups, 17 subtypes based on the hemagglutinin (HA). Influenza A subtype H1 and H3 viruses seasonally infect humans. Several high-resolution structures of antibodies engaging influenza A HA have been determined by X-ray crystallography. Antibodies that engage the antigenically diverse head region of the trimeric HA molecules are most often narrow in breadth. Recently, two types of antibodies have been described that exhibit broad neutralization multiple subtypes of influenza. These two types of antibodies bind the more conserved stem region of HA or the conserved sialic acid receptor binding pocket within the head region. H5N1 influenza viruses traditionally infect birds, but have been responsible for several recent outbreaks limited to bird-to-human transmission. Recent research described adaptations of influenza H5N1 that confer respiratory droplet transmissibility from ferret to ferret, which may mimic the future development of a highly pathogenic pandemic human H5 virus in nature. It is poorly understood how humans vaccinated with conventional H5 immunogens might be protected from infection or disease with such mutants, which possess a limited number of surface point mutations in the HA. We present preliminary data on the isolation of human neutralizing monoclonal antibodies to the H5 head domain that recognize both wild-type and respiratory droplet transmissible H5 HAs from humans vaccinated with conventional H5 HA protein vaccine. The first objective of this proposal is to establish a pipeline ‘RAPID’ to rapidly identify and characterize human antibodies that broadly neutralize influenza viruses – an important strategy for the swift response to emerging threats to human health. We recently demonstrated that a new computational method termed “multi-state design” can recapitulate antibody maturation in silico, i.e., predicts mutations that increase antibody affinity to its target, and also predict antibody sequences encoding antibody proteins that are capable of broadly recognizing multiple target proteins. The second objective of this proposal is to design and test RAPID for (a) in silico maturation of head-binding antibodies to increase affinity for the HA antigen and (b) multi-state design to create stem-binding antibodies that recognize HAs of multiple different clades, groups, or even types.