Protein Sequencing Tools for Biological Therapeutics

描述(申请人提供)：质谱学已经成为识别和表征复杂混合物中少量蛋白质的一种选择方法。然而，以高通量方式进行鉴定的能力取决于蛋白质序列数据库的可用性。这意味着，来自具有未测序基因组的生物的蛋白质(例如多肽毒素)和对环境做出反应而快速修改其初级序列的蛋白质(例如抗体)已被排除在高通量分析之外。我们建议开发算法和软件，并改进实验室方法，使抗体和多肽毒素的测序成为一项快速和低成本的工作。这将使我们能够获得个人循环抗体库，用于临床应用，包括疫苗开发，并获得用于基础研究和离子通道药物开发的大量生物活性毒液成分。为了改善实验室，抗体多肽和毒素将被化学标记以提高光谱质量，我们将使用不同类型的质谱碎片。数据采集将得到优化，以便于识别诊断相关的多肽，并将使用气相消化策略来增加较大多肽的序列覆盖率。我们建议开发用于抗体和多肽毒素测序的改进算法。这些将整合从头开始和数据库测序，并将包括结合多个信息通道的候选生成算法：来自不同电荷状态和碎片方法的光谱、同源性约束、成分约束以及数据库的电子突变。还将利用微妙的光谱线索开发改进的评分算法，目前仅用于手动从头测序。我们将生产原型软件，并以手动注释的质谱图为基准进行基准测试。然后，该软件将被应用于自动对来自HIV长期非进展者的大量抗体数据以及蜘蛛和锥形蜗牛毒素数据进行排序。

DESCRIPTION (provided by applicant): Mass spectrometry has become a method of choice for identifying and characterizing small quantities of proteins in complex mixtures. However, the ability to perform the identification in a high- throughput fashion has depended on the availabilit of protein sequence databases. This means that proteins from organisms with unsequenced genomes (e.g. peptide toxins) and proteins that modify their primary sequence rapidly in response to the environment (e.g. antibodies) have been excluded from high-throughput analysis. We propose to develop algorithms and software along with improving laboratory methods that make sequencing of antibodies and peptide toxins a fast and low-cost effort. This will allow us to access the circulating antibody repertoire of individuals for clinical application including vaccine development, and to access the vast number of bioactive venom components for basic research and ion-channel drug development. For the laboratory improvements, antibody peptides and toxins will be chemically labeled to improve spectral quality and we will use different types of mass spectrometric fragmentation. Data acquisition will be optimized to facilitate identification of diagnostically relevant peptides and a gas- phase digestion strategy will be used to increase the sequence coverage for larger peptides. We propose to develop improved algorithms for sequencing of antibodies and peptide toxins. These will integrate de novo and database sequencing and will include candidate generation algorithms incorporating multiple channels of information: spectra from different charge states and fragmentation methods, homology constraints, composition constraints, and in silico mutation of databases. Improved scoring algorithms will also be developed using subtle spectrum clues, currently used only in manual de novo sequencing. We will produce prototype software, and benchmark it against manually annotated mass spectra. The software will then be applied to automatically sequence a large set of antibody data from long- term non-progressors of HIV, and spider and cone snail toxin data.