Improving the Identification Rate of Endogenous Peptides Using Electron Transfer Dissociation and Collision-Induced Dissociation

Tandem
mass spectrometry (MS/MS) combined with bioinformatics tools
have enabled fast and systematic protein identification based on peptide-to-spectrum
matches. However, it remains challenging to obtain accurate identification
of endogenous peptides, such as neuropeptides, peptide hormones, peptide
pheromones, venom peptides, and antimicrobial peptides. Since these
peptides are processed at sites that are difficult to predict reliably,
the search of their MS/MS spectra in sequence databases needs to be
done without any protease setting. In addition, many endogenous peptides
carry various post-translational modifications, making it essential
to take these into account in the database search. These characteristics
of endogenous peptides result in a huge search space, frequently leading
to poor confidence of the peptide characterizations in peptidomics
studies. We have developed a new MS/MS spectrum search tool for highly
accurate and confident identification of endogenous peptides by combining
two different fragmentation methods. Our approach takes advantage
of the combination of two independent fragmentation methods (collision-induced
dissociation and electron transfer dissociation). Their peptide spectral
matching is carried out separately in both methods, and the final
score is built as a combination of the two separate scores. We demonstrate
that this approach is very effective in discriminating correct peptide
identifications from false hits. We applied this approach to a spectral
data set of neuropeptides extracted from mouse pituitary tumor cells.
Compared to conventional MS-based identification, i.e., using a single
fragmentation method, our approach significantly increased the peptide
identification rate. It proved also highly effective for scanning
spectra against a very large search space, enabling more accurate
genome-wide searches and searches including multiple potential post-translational
modifications.

串联质谱（Tandem mass spectrometry，MS/MS）结合生物信息学工具，已实现基于肽段-谱匹配（peptide-to-spectrum matches）的快速、系统性蛋白质鉴定。然而，精准鉴定内源性肽类（如神经肽、肽类激素、肽类信息素、毒液肽及抗菌肽）仍颇具挑战。由于此类肽类的剪切位点难以可靠预测，因此在序列数据库中检索其MS/MS谱图时，无法设置任何蛋白酶酶切条件。此外，多数内源性肽类带有多种翻译后修饰，因此在数据库检索中必须将此类修饰纳入考量。内源性肽类的这些特性导致检索空间极其庞大，常导致肽组学（peptidomics）研究中肽段表征结果的置信度偏低。本研究开发了一款新型MS/MS谱图检索工具，通过结合两种不同的碎裂方法，实现内源性肽类的高精度、高置信度鉴定。本方法依托两种独立碎裂方法的结合：碰撞诱导解离（collision-induced dissociation）与电子转移解离（electron transfer dissociation）。两种方法分别独立完成肽段谱图匹配，最终得分由两个独立得分融合得到。实验证明，本方法可有效区分真实肽段鉴定结果与假阳性命中结果。我们将本方法应用于从小鼠垂体瘤细胞中提取的神经肽谱图数据集。与基于单一碎裂方法的传统质谱鉴定策略相比，本方法显著提升了肽段鉴定率。此外，本方法在超大检索空间下的谱图扫描中同样表现优异，可实现更精准的全基因组检索，以及纳入多种潜在翻译后修饰的数据库检索。