Protein turnover estimation from fragment ions and precursor enrichment in heavy water labeled LC-MS experiments - SUMMARY Metabolic labeling with heavy water followed by LC-MS quantifies the turnover of thousands of proteins in vivo. As a cost-effective and easy-to-use stable isotope labeling tracer, heavy water has been used to study the turnover of proteins in the heart, liver, kidney, blood, and brain of various model animals such as mice, rats, dogs, and fish. Since the deuterium in heavy water is incorporated into all non-essential amino acids, virtually all peptides are labeled. However, despite the ubiquitous nature of heavy water labeling, only 30-40% of all quantified peptides are useful for protein turnover rate estimation. Most of the mass spectral data is filtered out due to the low goodness-of-fit characteristics (such as the coefficient of determination, R2). Our long-term goal is to develop a unified dynamics of biomolecules (proteins, lipids, and metabolites) using heavy water labeling and LC-MS-MS/MS to study changes in diseases. The objective of this project is to develop techniques to increase the throughput, statistical power, and accuracy of protein turnover estimation. To achieve our objectives, we will pursue two Aims. In the first Aim, we employ a new approach to estimate protein turnover from isotope labeling of fragment ions in tandem mass spectra. Unlike the intact peptide ions (used in current bioinformatics techniques), the fragment ions are less affected by space-charging, limited dynamic range, and overlapping profiles caused by co-eluting contaminants. However, new techniques need to be developed and applied to estimate label enrichment from truncated isotope distributions observed in tandem mass spectra. In addition, a statistical normalization modeling will be required to account for the limited isolation width of peptides chosen for fragmentation. In the second Aim, we develop a model of label incorporation which uses two pools of amino acids and accounts for the individual labeling kinetics of each amino acid. The traditional methods of estimation of label incorporation assume a single amino acid pool – newly synthesized, labeled amino acids. However, it has been observed that peptide rates are dependent on the amino acid composition. The labeling kinetics of each amino acid is dependent on its biogenesis pathway. The proposed computational tools will improve the accuracy of protein turnover estimations. They will increase the depth of proteome coverage and statistical power in the studies of in vivo protein turnover using heavy water metabolic labeling and LC-MS.
