Extending Plasma Proteome Coverage To Enable Biomarker Discovery

Two features make plasma a valued source of health information: First, plasma contains secreted proteins from all tissues in the body and can, therefore, provide information about the health status of all tissues. Second, plasma is available in large quantity using a minimally invasive procedure that is well tolerated. The status of plasma as a valued source of health information is matched by the difficulty in extracting proteome information from the samples. Using state of the art instrumentation a standard analysis of plasma yields quantitative information on approximately 300 proteins. Extreme fractionation strategies coupled with increased mass spectrometry time extends the depth of coverage to about 2,000 proteins. However, approaches requiring increased analysis time are not practical for most studies due to instrument availability and costs. The major roadblock to greater depth of coverage with short analysis times is the dynamic range of the plasma proteome and the very high abundance of a very few proteins in plasma. Essentially, serum albumin, antibodies and a handful of other proteins saturate the system and block the observation of lower abundance proteins. A successful strategy to increase depth of plasma proteome coverage is immunodepletion of the top 12 to top 22 most abundant proteins. This is effective in increasing depth of coverage to about 500 proteins but still leaves most proteins unobservable. We have adopted a strategy that enriches low abundance proteins that are part of the plasma proteome and are associated with particulates. We have developed this approach due to the high biological value of protein analytes in particulate fractions from plasma. These high value analytes are part of the cellular secretome. They include protein contained in exosomes, ribosomes, protein aggregates and other membrane vesicles. The total protein in all these particulate fractions is less than 2 parts in 10,000 of plasma protein. This low relative abundance makes the plasma particulate proteome, and all the high value analytes in that proteome, unobservable in unfractionated plasma. The hypothesis for this project is that analysis of particle enriched plasma will allow observation and quantitation of both soluble and particulate components of the plasma proteome. It is further hypothesized that using PE-Plasma instead of raw plasma will be a resource sparing strategy that will enable high throughput analysis of an extended plasma proteome. Extracellular vesicles from plasma, other body fluids and cell culture media hold great promise in the search for biomarkers. Exosomes in particular, the vesicle type that is secreted after being produced in the endocytic pathway and having a diameter of 30 nm to 150 nm, are considered to be a conveyance for signaling molecules and, therefore, to hold valuable information regarding the health and activity status of the cells from which they are released. The vesicular nature of exosomes is central to all methods used to separate them from the highly abundant proteins in plasma and other fluids. The enrichment of the vesicles is essential for mass spectrometry-based analysis as they represent only a very small component of all plasma proteins. The progression of isolation techniques for exosomes from ultracentrifugation through chromatographic separation using hydrophobic packing materials shows that effective enrichment is possible and that high throughput approaches to exosome enrichment are achievable. We developed informatics approaches to interrogate data from particle enriched plasma to assess the quality of preparation via mass spectrometry and to identify potential biomarker candidates of PTSD, fatigue, and chemical exposure in Middle Eastern refugees from Iraq and Syria. Localization of organelle proteins by isotope tagging (LOPIT) maps are a coordinate-directed representation of proteomics data that, when subjected to pattern analysis, aid in the biological interpretation of proteomic data. Analysis of organellar association of proteins as displayed using LOPIT plots is evaluated two types of proteomic data sets. Interpretating affinity selection experiments is complicated by high background leading to false positives. Particularly with streptavidin-based selection of biotinylated targets, even highly stringent washes with detergents leave high background proteomes. Using LOPIT maps helps in the interpretation of streptavidin affinity selection data through pattern recognition. A proximity labeling experiment with the Bir-A biotin ligase fused to the folate transporter was completed. Proteins of interest as potential interacting partners to the folate transporter were isolated using streptavidin beads with the final wash collected and evaluated along with the elute fractions. Mapping the subcellular localization of identified proteins and establishing a cut off for the elute to wash abundance ratio (E:W) simplifies biological interpretation. Interpreting data from prepared vesicle samples is complicated by high background of the soluble proteome leading to the false assignment of proteins to the vesicular proteome. Using LOPIT maps helps in the interpretation of data from the vesicular proteome through pattern recognition. We assess the quality of vesicle preparation by plotting the protein abundance and fold change data between fractions are to LOPIT maps to generate protein localization patterns for each preparation. Vesicular proteome localization patterns of validated vesicular proteins aid in evaluation of sample quality and allow us to discriminate between vesicular preparation methods. We show that the patterns that arise in LOPIT maps are easy to understand and compare which aids in the biological interpretation of proteomics data. Acquiring high depth of coverage data with a low number of missing values is important for the identification of biomarker candidates. Two choices most prevalent in current proteomics publications are using data dependent acquisition with isobaric labels for multiplexing or using data independent analysis with unlabeled samples. Tandem mass tags are isobaric peptide labels commonly used to increase throughput in mass spectrometry experiments. Tandem mass tags contain an NHS-Ester reactive group for labeling of primary amines. Tandem mass tags decrease the number of missing values for quantified proteins in a labeled sample pool. However, as more batches are needed, the number of missing values increases between batched. NHS-Esters are commonly used reagents for the labeling of primary amines with fluorescent or isobaric labels. We show that the number of missing values is decreased when using data independent acquisition of unlabeled samples in a large cohort study compared to using data dependent acquisition of samples labeled with tandem mass tags in a large cohort study. When unaccounted for, off-target labeling of hydroxyl-containing amino acids serine, threonine, and tyrosine complicate calculations for the degree of labeling and ultimately result in inaccurate quantitation. We corroborate recent reports demonstrating off-target labeling, investigate the addition of hydroxylamine after the reaction to reverse off-target labeling and investigate the addition of nickel (II) and imidazole in the reaction to block off-target labeling. We show that additional hydroxylamine does not decrease the labeling efficiency of the reaction but does decrease the number of off-target labeling events. We show that the addition of nickel (II) and imidazole decreases labeling efficiency and off-target labeling events. Study of mental health diseases, fatigue, and chemical exposure in individuals who have been in active war zones is critical for veteran and refugee quality of care improvements. We sought to identify candidate biomarkers of PTSD, fatigue, and chemical exposure in a small cohort study (n = 58) of refugees from Iraq and Syria. Extracellular vesicles and exosomes can interact with and cross the blood brain barrier, the enrichment of these particles from blood plasma provides an opportunity to identify biomarker candidates for mental health diseases. We used ultracentrifugation to enrich exosomes and extracellular vesicles and deplete the abundance of soluble proteins to assess the particulate proteome as a surrogate of the cell. We demonstrated the depth of coverage increased over 2X when screening particle enriched plasma to raw plasma on a one-hour gradient. We identified and combined information from differentially abundant proteins and gene ontology biological processes, cellular components, and molecular functions to narrow the dataset to specific biomarker candidate proteins for PTSD, fatigue, and chemical exposure