Group Members: Yan Li, Marshall Pope, Noelle Griffin, Fred Long
Proteomics approaches theoretically provide a means to comprehensively define protein expression patterns and identify differences among samples in a relatively rapid manner. Though genomic analysis can be used to compare global changes in gene expression between tissues or states, large changes are needed to detect differences between tissues. Additionally, changes in gene expression don’t always correlate with changes in protein expression. Proteins may move between different locations within a cell, altering both their accessibility and function. Posttranslational changes can also alter protein location and function. To truly define the proteins present at the endothelial cell luminal surface and to use this identification as a means to assess function, protein expression itself must be characterized. Mass spectrometry based techniques can identify proteins based on the presence of digested peptides, but highly complex samples are difficult to separate. We have extensive experience in mass spectrometry-based approaches to protein identification. We have developed the most comprehensive classification of proteins at the luminal surface of endothelial cells to date, have pioneered methods to improve mass-spectrometry based identification of proteins, and have developed novel methods to normalize mass spectrometry data that allows data to be reliably quantified and compared. The proteomics facility’s main aims are to map the membranome of multiple organs in health and disease, to improve protein identification, to develop methods to quantify and compare mass spectrometry-based data, and to identify post-translational modifications of proteins.
The proteomics facility currently consists of a HETO vacuum centrifuge, a Sutter laser capillary tube puller, a Multiprobe II MassPREP Robotic Protein Handling System from Perkin Elmer, an Investigator 2-DE gel separation system, a Fisher Steroscope, two Agilent Technologies 1200 HPLC pumps coupled to two separate Thermo LTQ linear ion trap mass spectrometers equipped with nanospray ESI sources, multiple 2.4 Ghz PCs with database searching software (Sequest, Thermo-Finnigan), and a PowerEdge 2900 Server with two quad-core Xeon X5450 3 Ghz processors, 8 GB of RAM, and two 300 GB SAS disks to support the AVATAR bioinformatics platform. These instruments are used for large scale proteomics mapping. This facility also has full access to the complete PRISM lab space, which includes cold rooms, radioactive labeling preparation and analysis rooms, tissue culture suites, communal instrument rooms, scintillation counters, refrigerators, chemistry hoods ultracentrifuges and key rotors, tissue culture incubators ultra-low freezers, a MilliQ water purification system, and general-use computers.
The plasma membrane mediates a wide variety of basic biological functions including signal transduction, molecular transport, membrane trafficking, cell migration, cell-cell interactions, intercellular communication and even drug-resistance. Plasma membrane-associated proteins, especially integral membrane proteins (IMP) that traverse the lipid bilayer, are key elements mediating these vital biological processes. Developing a map of the proteins present at the luminal surface of vascular endothelial cells (a membranome) will shed new light on the function of these cells in vivo. Additionally, because these cells form a vital interface between the blood and tissue, defining the membranome will reveal new tissue- and disease-specific targets.
Intrinsic hydrophobicity, strong interaction with cholesterol, a wide concentration range of proteins, and other factors have hampered resolution and identification of integral membrane proteins using mass spectrometry-based strategies. In our initial survey of the rat lung endothelial cell membranome, we used 2-dimensional liquid chromatography coupled with an LCQ mass spectrometer to identify 450 proteins, of which ~15% were integral membrane proteins. This was a notable total number of proteins at the time and a significant step towards defining the complete membranome. However, many more integral membrane proteins were expected. Indeed, several well-known endothelial cell surface markers were not identified in these samples. Importantly, we found that only 66% of the identified proteins in any single mass spectrometry measurement of rat lung membrane extract were confirmed by any second measurement, and 7-10 measurements are needed to reach 95% confidence of analytical completeness.
Identifying new tissue-specific proteins requires comprehensive analysis of the proteins present in different tissues. Just because a protein is not detected in one or two mass spectrometry measurements does not mean it is not present in the sample. As single mass spectrometry measurements likely provide a small fraction of the total protein space, one cannot be truly confident in differential protein expression unless a sample is analyzed to completeness. This is necessary to reduce false positives and focus on proteins that are truly differentially expressed between samples.
To enhance mass spectrometry-based identification of lipid-embedded proteins, we have compared the ability of four different mass spectrometry-based methods to identify proteins in isolated endothelial cell plasma membranes. Traditional mass spectrometry methods that use both strong cationic exchange and reversed-phase columns to separate peptides (2DC) were compared with first separating proteins by size on SDS-PAGE gels followed by analysis by both reversed-phase mass spectrometry and 2DC (G2DC). Gel prefractionation preserved many membrane proteins that do not re-solubilize after the sample preparation steps required for 2DC. G2DC revealed more proteins than the other methods; however, each technique identified unique peptides. Combining multiple techniques dramatically increased sensitivity, especially for integral membrane proteins. Our previous analysis of plasma membranes identified 450 proteins using 2DC. Using the combination of methodologies described above, we have now identified 1834 fully non-redundant proteins. Additionally, fewer replicates were needed to provide comprehensive coverage of proteins in the sample. This membranome was recently published. Since then, we have identified another 1000 proteins at the endothelial cell surface using more sensitive mass spectrometers.
Though it requires heavy use of mass spectrometry instrumentation, using multiple analytical approaches and reaching 95% analytical completeness for each method yielded a significantly higher number of proteins and the quality of the data may well be worth the extra time invested. For our lab, we plan to adhere to a criterion of 95% analytical completion. The extra mass spectrometry measurement days may well be time well spent to avoid a long list of candidates with few actual “true positives”.
Developing methods to quantify and compare mass spectrometry-based data:
Replicate measurements and combining multiple methodologies is necessary to allow a comprehensive analysis of proteins present in the sample and to reliably detect differences types of tissue. Intrinsic variability between measurements and methodologies makes it difficult to compare the levels of peptides found with different techniques. We have developed a novel, label-free method to quantify and normalize mass spectrometry data that relies on intrinsic properties of the data to control for inherent bias and noise, thus facilitating direct quantitative comparison. By normalizing data around a Spectral Index that takes peptide number, spectral count, peak precursor ion intensities and protein length into account, we have successfully reduced variance between replicates and across a dynamic range of protein loads. We validated this method in the following manner: 1) unsupervised clustering of normalized mass spectrometry data was used to detect “sameness” and to separate multiple replicates into distinct samples; 2) intensities from Coomassie-stained SDS-PAGE gels were correlated with the spectral index for the same proteins, indicating that this is a good measure of protein abundance; 3) proteins were enriched in plasma membrane preparations to the same degree in normalized mass spectrometry data and in densitometry analysis of western blots, showing that this method can reliably quantify relative amounts of protein. Quantification and normalization of the data is essential for mapping the proteome of endothelium and its caveolae and to identifying targets for site-directed pharmacodelivery. This method of quantification allowed us to identify Aminopeptidase P (APP) as one of the most abundant proteins in caveolae, outnumbering even caveolin 1, a structural component of caveolae. This protein is expressed quite selectively in lung endothelium in vivo and has been used to target intravenously injected antibodies to the lung vasculature where they are rapidly and specifically pumped out of the blood and into lung tissue in a caveolae-dependent manner.
Analyzing post-translational modifications:
Post-translational modifications such as phosphorylation, ubiquitination, and nitrosylation regulate many cellular functions including proliferation, migration, plasticity, and differentiation. These modifications are often transient and many of the modifying enzymes are implicated in multiple signaling cascades. Untangling overlapping signaling pathways requires sensitive, simultaneous tracking of several targets, both known and unknown. We are currently expanding our mass spectrometry capabilities to include the LTQ Orbitrap Velos with electron transfer dissociation capability (ETD). The LTQ Orbitrap Velos is a next-generation mass spectrometer that is faster and has higher sensitivity and mass accuracy than current MS instruments on the market, and combines both ETD and CID to allow identification of post-translational modifications and to increase protein identification. The LTQ Orbitarp Velos will be commercially available in summer, 2009, and Thermo Scientific has promised delivery and set up by Spring, 2010.
Bioinformatics capabilities:
To analyze the massive amounts of data created in the mass spectrometry facility, we have created and maintained a relational database to store and analyze mass spectrometry results, called AVATAR (Accessible VAscular TARgets). We have recently updated the database to AVATAR 6.0 to reflect changes in published genome data, which include both new proteins and removal of redundant proteins. AVATAR enables us to perform extensive annotation, including addition of dat a concerning protein structure, localization, function, and any biological processes they may mediate, as well as enter scanned images of Western analyses, cell and tissue staining, and SPECT images generated during validation. Proteins identified through mass spectrometry-based methods have been entered into AVATAR for storing and relational analysis. We have eliminated much of the redundancy in the database and annotated it, both through linkage to various sites and by manual additions from literature.
