PSMeeting_SSF_11-03_abstracts.htm

Title	Abstract
I. Trends in Proteomics: "Accommodating Differential Display Data in Systems Biology" Luke Schneider, Target Discovery, luke_Schneider@targetdiscovery.com	Unlike the measurements conducted in other scientific disciplines, those common to biology are often ratiometric (differential display) rather than absolute methods. This is particularly true of mass spectrometric methods (e.g, ICAT, IDBEST), but is also true of 1-D and 2-D gels. The result is the proliferation of ratiometric data for protein, DNA and metabolite expression, rather than absolute concentrations and rates. Statistical data mining methods are often invalid in biology because the number of features measured typically far exceeds the number of samples, restricting researchers to model-based analyses. However, it is very difficult to apply traditional mathematical modeling strategies to such differential display data. We introduce a new mathematical paradigm for handling the differential display data type directly in systems biology modeling.
"A New Trend in Proteomics – Getting Useful Information" Kelvin Lee, Cornell University, KHL9@cornell.edu	This presentation will cover some of the state of the art methods that are used for proteomics with an emphasis on approaches that are accessible to many laboratories. Although there is a wealth of data being collected on various problems, a key issue is the use of this data to understand the biological system and ultimately to control that system. We will cover one way in which proteomic information can be used by integrating this data with DNA sequence information and mRNA expression profiles into a deeper understanding of protein synthesis and discuss the successful application of proteomics to achieving enhanced protein secretion in Escherichia coli.
II. Bioinformatics "Discovering Known and Unknown Protein Modifications Using MS/MS Database Searching" Wilfred Tang, Applied Biosystems, tangwh@appliedbiosystems.com	We present an MS/MS database search algorithm with the following novel features: (1) zone modification searching, which enables the discovery of protein modifications of known (i.e., user-specified) and UNKNOWN delta masses, and (2) pre-indexing of protein databases both by the masses of the peptides obtained from digestion of the proteins using trypsin (or some other digest agent) as well as by the theoretically expected MS/MS fragment masses of each of the peptides. In zone modification searching, the mass range covered by each MS/MS spectrum is divided into six equally-sized zones. For each of the six zones, it is hypothesized that a modification occurs in that zone, and the search against the protein database isperformed accordingly. Pre-indexing of the protein database results in significant efficiency gains and is particularly advantageous in a large zone modification search. All of these features are implemented in Interrogator, the search engine which run behinds Pro ID and Pro ICAT. The ability to search for a large variety of known and unknown modifications allows a significantly greater percentage of MS/MS scans to be identified. Moreover, we provide several examples where the ability to search for UNKNOWN modifications allows the scientist to discover: (1) unexpected modifications which have biological meaning, (2) non-tryptic peptides in a sample which has nominally been digested using trypsin, and (3) other unintended consequences of sample handling procedures.
III. Sample Preparation "Immunoaffinity Depletion Of High-Abundant Proteins From Human Serum For Proteomic Sample Preparation" Chris Miller, Agilent Technologies, Christine_miller@agilent.com	Human serum, the most complex sample of the human proteome, is valuable in the discovery of new biomarkers. Difficulties in the analysis of serum arise from the fact that proteins range in concentration over 12 orders of magnitude. A few well-characterized high-abundant proteins represent up to 90% of the total protein mass and mask identification and characterization of the lower abundant proteins by limiting the dynamic range of mass spectral and gel electrophoretic analysis. We have developed a single-device multiple affinity removal system for simultaneously selecting and removing six high-abundant proteins from human serum. This uses an immunoaffinity-based column and optimized buffers for efficient reduction of non-specifically bound proteins and effective column regeneration. To compare the performance of the immunoaffinity column to dye-affinity chromatography (Cibacron Blue), bound fractions were resolved electrophoretically, then protein bands were analyzed by LC/MS/MS. We found that the immunoaffinity system exhibits superior specificity of depletion in comparison to Cibacron Blue. Optimized use conditions permit the column to be used multiple times and column reproducibility was assessed by analysis of the flow-through fractions from more than 200 consecutive runs. The column provides consistent, robust depletion of high-abundant proteins as determined by 1D gel patterns, MALDI-TOF and LC/MS analysis. ELISA results indicate that the multiple affinity removal column depletes 98-99% of targeted high-abundant proteins in a single pass through the column. The column expands the dynamic range of current LC/MS, 1DGE and 2DGE analytical methods, enabling increased protein identification in human serum.
"THE PERFECT 2D GEL – Fact or Fiction?" Barbara Cottrell, UCSD, bcottrell@ucsd.edu	Meticulous sample preparation is the key to clean, reproducible comparative 2D gels and there is no universal protocol. Protein extraction procedures will depend on the source of the protein: cell culture, tissue, organelles or intact organisms (C. elegans). The elimination of salts is critical and precipitation or dialysis may be required, especially if protein concentrations are low. While there are many kits offering a variety of extraction procedures and detergents, finding the correct combination for a given system is a matter of trial and error. 2D gel patterns can be detergent dependant. Immunoprecipitation with Sepharose requires careful controls to account for non-specific binding and the use of antibodies bound to magnetic beads can be beneficial. Additional concerns are the relevant abundance of the protein of interest and pre-fractionation of organelles may be necessary to enrich for a specific group of proteins. Sequential extraction can be helpful in reducing the complexity of a mixture but there may be carryover in subsequent steps, an important issue when quantifying proteins. Serum proteins can affect the protein profiles in cell culture extracts as well as tissue. The reduction and alkylation of protein disulfides prior to isoelectric focusing can prevent vertical streaking Human keratin may be a problem and can be eliminated with a few simple precautions. While the number of variables may seem over-whelming, a perfect 2D gel will be the reward for careful planning and attention to detail.
IV. Biological Applications "Host-Pathogen Interactions: Proteomic and Genomic Analysis of Host Response" Sandra Maloney, Lawrence Livermore National Labs, smaloney@llnl.gov	Yersinia pestis, the causative agent of plague, is a gram negative, highly communicable bacterium, known both as a natural pathogen and as a biothreat agent. Y. pestis functions via the activation of the Type III secretion mechanism that enables the bacterium to adhere to and deliver virulence factors inside the host cell. This secretion mechanism is common to several human pathogens including Y. psuedotuberculosis and Y. enterocolitica, near neighbors of Y. pestis with very similar genomic content yet dramatically different clinical manifestations – Y. pestis has a high mortality rate while Y. psuedotuberculosis and Y. enterocolitica cause intestinal distress. We have applied genomic and proteomic technologies to characterize host response to these three closely related pathogens as a means to define presymptomatic biomarkers for plague. Specifically we have used targeted gene arrays and 2D DIGE to characterize the genomic immune response and the global proteomic response. Total RNA and protein fractions were isolated from macrophages after pathogen exposures. Our results reveal shared host expression changes indicative of a common immune response; however, distinct pathogen-specific expression differences were also observed. Of 268 immune response genes analyzed, 180 were differentially expressed after pathogen exposure and 18 were specific for Y. pestis exposure. By 2D DIGE, we detected 1000’s of proteins of which ~250 were differentially expressed in response to the pathogen exposures and ~25 were specific for Y. pestis. These pathogen-specific genomic and proteomic changes clearly distinguish the three different pathogen exposures and represent novel biomarkers for plague. This combined proteomic and genomic approach promises to identify diverse biomarkers for a variety of diseases. This work was performed under the auspices of the U.S. Department of Energy by University of California Lawrence Livermore National Laboratory under contract No. W-7405-ENG-48 with support from the Department of Homeland Security (Biological Warning and Incident Characterization Program).
"Gene and Protein Expression Profiling Reveal Unique Cellular Targets and Signaling Pathway of Tumor Suppressor Gene FHIT in Lung Cancer" Jing Zhu, Ciphergen Biosystems, jzhu@ciphergen.com	Tumor suppressor gene (TSG) plays an important role in human cancer development. Its activity modulates a biological network involving many genes and signaling pathways that control various cellular processes. High-throughput systems that can provide a global view of cellular function are necessary to gain a comprehensive understanding of such a complex biological system. In this study, we used a complementary approach with DNA microarray (Affymetrix GeneChips) and ProteinChip (Ciphergen) technologies to quantitatively monitor the FHIT TSG-induced changes at both gene and protein levels in non-small cell lung carcinoma (NSCLC) cells. We performed gene expression analysis on Ad-FHIT-transduced cells, compared with the PBS-treated mock and empty vector or Ad-LacZ-treated negative controls. The data were analyzed by a perfect-match (PM) model-based analysis. About 200 genes differentially and specifically expressed in Ad-FHIT-transduced cells were identified and these genomic data were further compared and validated with the public databases. A protein expression analysis using ProteinChip technology was also simultaneously performed using the same samples. About 40 cellular targets that are either differentially up-or down-regulated by FHIT were identified by a serial analysis of fractionation, profiling, and identification. More than 70% of these differentially expressed proteins identified are also found their corresponding gene targets pulled out on DNA microarray. A comparative analysis of the gene and protein expression profiling revealed several unique cellular targets and signaling pathways involved in FHIT tumor suppressing activity. Some of these targets were further confirmed by immunoblotting and biologically validated by gene-specifc RNAi. These findings clearly linked FHIT tumor suppressing function to the regulation of cell proliferation and apoptosis. Our data demonstrated that the complementary gene and protein expression technology is a powerful tool for generation of hypothesis and for identification of specific cellular targets and signaling pathways mediated by a specific gene product in a complex biological network.
TUTORIAL: "Analyzing Peptides and Proteins with Mass Spectrometry and Database Searching" by *Jimmy Eng,* Institute for Systems Biology jeng@systemsbiology.org	Methods of searching public sequence databases with mass spectrometry data for peptide and protein identification will be presented. Specifically, the mechanisms of mass spectrometry database searching (aka peptide mass fingerprinting), and tandem mass spectrometry database searching will be discussed, along with pointers to access public search engines and publicly available sequence databases.
"Advanced Mass Spectrometric Approaches for Rapid and Quantitative Proteomics" by Ljiljana Pasa-Tolic, Senior Research Scientist, Pacific Northwest National Laboratory ljiljana.pasatolic@pnl.gov	This talk describes and demonstrates a global strategy that extends the sensitivity, dynamic range, comprehensiveness, and throughput of proteomic measurements. The two stage strategy exploits Fourier transform ion cyclotron resonance mass spectrometry (FTICR) to first validate accurate mass tags (AMTs)1 produced by global protein enzymatic digestion for a specific organism, tissue or cell type from potential mass tags identified using conventional tandem mass spectrometry (MS/MS) methods, providing the basis for subsequent measurements without the need for MS/MS.2,3 This global tryptic digest approach bypasses the 2-D PAGE separations and when combined with the enhanced sensitivity and dynamic range of the FTICR instrumentation (i.e., by the application of data-dependent active FTICR dynamic range enhancement methods using external ion m/z selection)4,5 expands the identification of proteins while eliminating the inherent complications and more limited coverage associated with 2-D PAGE. A key advantage of this approach is, however, that it enables high throughput and high precision quantitative measurements of changes in gene expression based upon stable isotope labeling (i.e., during cell culture using labeled media or Cys-peptide specific labeling methods), predetermined AMTs and the high MMA achievable by FTICR MS.
“Implementation and Uses of Automated de novo Peptide Sequencing by Tandem Mass Spectrometry” *Richard S. Johnson,* Senior Staff Scientist, Immunex Corporation, JohnsonR@immunex.com	There are several computer programs that can match peptide tandem mass spectrometry data to their exactly corresponding database sequences, and in most protein identification projects these programs are utilized in the early stages of data interpretation. However, situations frequently arise where tandem mass spectral data cannot be correlated with any database sequences. In these cases, the unmatched data could be due to peptides derived from novel proteins, allelic or species-derived variants of known proteins, or post-translational or chemical modifications. Two additional problems are frequently encountered in high throughput protein identification. First, it is difficult to quickly sift through large amounts of data to identify those spectra that, due to poor signal or contaminants, can be ignored. Second, it is important to find incorrect database matches (false positives). We have chosen to address these difficulties by performing automatic de novo sequencing using a computer program called Lutefisk. Sequence candidates obtained are used as input in a homology-based database search program called CIDentify to identify variants of known proteins. Comparison of database-derived sequences with de novo sequences allows for electronic validation of database matches even if the latter are not correct. Modifications to the original Lutefisk program have been implemented to handle data obtained from triple quadrupole, ion trap and quadrupole / time-of-flight hybrid (Qtof) mass spectrometers. For example, the linearity of mass errors due to temperature-dependent expansion of the flight tube in a Qtof was exploited such that isobaric amino acids (glutamine / lysine and oxidized methionine / phenylalanine) can be differentiated without careful attention to masscalibration.

"Host-Pathogen Interactions: Proteomic and Genomic Analysis of Host Response"