8th International Symposium on Mass Spectrometry in the Health and Life Sciences: Molecular and Cellular Proteomics

Speaker Abstracts

MS.1

Quantitative Proteomics: An Overview

Ruedi Aebersold
Institute for Molecular Systems Biology, ETH Zurich; Faculty of Science, University of Zurich, Switzerland; Institute for Systems Biology, Seattle, WA.

Accurate quantification of the proteins that are detected or identified in proteomics experiments has become a primary goal of proteomics research. Quantitative information is essential in many areas of biological and clinical research, including for the detection of dynamic change in biological systems, for comparative analysis between samples and for generating boundary conditions in mathematical models of biological processes. Over the last few years a multitude of methods for quantitative proteomics have been described. In this presentation we will provide an overview of the currently available methods and their performance. Specifically, we will discuss the following methods: Quantitative proteomics based on stable isotope labeling or tagging and tandem mass spectrometry; quantification based on LC-MS pattern matching with or without isotope tagging; absolute quantification via isotope labeled external standards and targeted (multiple reaction monitoring) mass spectrometry. We will also discuss issues common to all methods, including error analysis and quantitative accuracy and current methods for the generation of isotope labeled reference peptides.

MS.2

Profiling protein expression by label-free quantitative mass spectrometry

Pedro R. Cutillas and B. Vanhaesebroeck
Bart's Institute of Cancer, Queen Mary University of London, United Kingdom

The role of mass spectrometry (MS) in proteomics has evolved from being a technique used to identify proteins present in 2D gel spots to be the preferred method for quantitative analyses. Indeed, several strategies now exist that use MS to quantify proteins; it may be argued that the most influential of all them was based on isotope labelling using ICAT reagents, an approach that made the analytical biochemist appreciate the potential of MS for large-scale protein quantification. This early work inspired the development of other, perhaps more robust, methods for global protein quantification, most of which were also based on isotope labelling (prominent and popular examples include SILAC, iTRAQ and 18O labelling). More recently, MS-based quantitative methods that do not require protein labelling are also becoming popular. These label-free methods are based on spectral counts or on using peptide ion intensities as a read-out of protein abundance. The increasing popularity of the spectral count approach may be due to its simplicity, but the accuracy of such approach is low. In contrast, label-free approaches based on measuring peptide ion intensities show levels of accuracy close to those afforded with labelling techniques, and allow researches to compare an unlimited number of samples with simple workflows that do not require chemical reaction steps, or custom reagents or media. In addition, label-free LC-MS may also allow to obtain estimates of protein abundance in absolute units. However, although in principle very powerful, implementation of this approach to quantify proteins on a genomic scale (the ultimate aim of expression proteomics) requires the use of (i) appropriate strategies in order to compare the same peptides across the samples to be evaluated, (ii) of normalization procedures to correct for experimental sources of variability, and (iii) of informatics tools to automate data analysis. In this presentation we will discuss how we have approached these problems by creating a computer program, termed PESCAL, which extracts and compares ion intensity values of peptides that are selected for MS/MS at least once across the samples to be compared. We also implemented normalization procedures that allowed us to quantify proteins on a genomic scale with a precision close to those obtained using isotopic labelling strategies. In order to investigate the limits of this approach, we tested it in a challenging sample set that consisted of 5 murine proteomes. Using a first generation Q-Tof mass spectrometer we were able to derive ~ 44,000 independent data points to quantify the expression of ~1000 proteins in 5 organs. Thus, with appropriate normalization procedures and informatics tools, label-free LC-MS is a powerful alternative to isotope labelling for quantitative proteomics.

MS.3

Profiling Unlabeled Peptide Ions: A Versatile Approach to Quantitative Proteomics and to Mapping of Post-Translational Modifications

G. Jaitly1, E. Bonneil1, N. Jaitly2, C. Pomies1, M. Marcantonio4, P. Drogaris1, Pierre Thibault1 3
Institute for Research in Immunology and Cancer1, Department of Chemistry3, and Department of Biochemistry, Université de Montréal, Montréal, Canada 4; Biological Science Division, Pacific Northwest National Laboratory, Richland, WA, USA2

The ability to monitor the subtle changes of protein abundances in response to specific perturbations of a biological system (e.g. cell signalling and differentiation, chemical stimulation, etc…) plays an important in the identification of potential lead candidates as part of biomarker discovery programs. However, this task presents sizable difficulties in view of the overwhelming sample complexity and variability associated with cell extracts. In order to profile low abundance expression changes across cell extracts, we have developed MassSense, a software that provide comprehensive peptide detection and segmentation analyses from data files of different MS platforms. The detection efficiency was determined by manual examination of a dense region of a representative peptide map with ion intensities distributed over 3 orders of magnitude. We evaluated the performance, reproducibility, statistical significance and dynamic range of peptide detection on a nanoLC-MS LTQ-Orbitrap mass spectrometer using scaled mixes of protein standard digests spiked in complex cell extracts. A linear response for all spiked proteins was observed over more than 2 orders of magnitude with detection limits of 1 fmole. We also evaluated the application of quantitative proteomics in two-dimensional nanoLC-MS/MS experiments for combined expression and identification analyses of human monoblastic U937 cells stimulated with phorbol ester. Several proteins including stathmin, hnRNPQ1-3, and ribophorin that showed differential expression and phosphorylation were correlated by western blot experiments. The capability to identify subtle abundance changes across different sample sets also provided unique advantages to monitor sites of modifications in complex cell extracts. This latter aspect will be exemplified for differential phosphoproteome analyses of J774 macrophages cell exposed to interferon-γ and for functional assays on histone acetyl transferases to locate precise sites of acetylation.

MS.4

Quantitative approaches for analysis of regulatory post-translational modifications.

Jeffrey J. Gorman1, T.P. Wallis1, K.A. Dave1, B.R. Hamilton1, M.J. Headlam1, S. Linke2 and D.J. Peet2
Protein Discovery Centre, Queensland Institute of Medical Research. Brisbane, Queensland, Australia1;, and School of Molecular and Biomedical Science, The University of Adelaide, Adelaide, Australia2.

Post-translational modifications provide functional switches and docking points within cellular protein networks. Functionally important post-translational modifications may be dynamic or transient in nature to respond to signals requiring pathways to be up- or down-regulated. It is essential to be able to detect such modifications and to monitor them quantitatively in a dynamic fashion in order to be able to assess their regulatory and functional significance to protein networks.
Accordingly, we have assessed and/or developed quantitative approaches for analysing post-translational hydroxylation of transcription factors involved in response to hypoxic stimuli. A comparison has been made between label-free and stable isotope labeling methods in conjunction with MALDI-TOF/TOF-MS/MS and ESI-LTQ-Orbitrap analysis of modification of Notch by the asparagine hydroxylase (FIH) that is involved in hypoxia inducible factor regulation. Quantitative analysis was an essential component of the study of this system in order to differentiate between asparagine hydroxylation and methionine oxidation of specific regions of Notch. As a consequence we have defined two specific sites of asparagine hydroxylation in Notch catalysed by FIH.
We have also investigated a stable isotope labelling approach for quantitative analysis of the post-translational phosphorylation in relation to regulation of the transcriptional activity of the Dioxin Receptor (DR).
This presentation will describe these approaches as well as the discovery of other post-translational modifications of DR and phosphorylation sites on Newcastle disease virus proteins.

MS.5

Quantification by Spectral Counting in Large Datasets

Eric Deutsch
Institute for Systems Biology, Seattle, WA

Several recent works have used spectral counting as a method of label-free protein quantification in complex samples. The method is effective in shotgun proteomics experiments because higher abundance proteins will bring more distinct peptides into the detectable range and more abundant peptides will be sequenced multiple times despite efforts to minimize this. Most examples of this technique compare protein abundance within a small set of samples within one experiment. We apply spectral counting methods to large groups of datasets within the PeptideAtlas to set an approximate scale of protein abundance within specific sample types such as human plasma. We present statistical issues related to quantification by spectral counting in large groups of datasets and its utility for planning targeted proteomics experiments.

MS.6

Protein Abundance Ratios for Microbial Proteomes

Murray Hackett, Q. Xia, T. Wang, G. Bosch, and F. Taub
Department of Chemical Engineering, University of Washington, Seattle, WA

The use of multidimensional capillary HPLC combined with tandem mass spectrometry has allowed high qualitative and quantitative proteome coverage of prokaryotic organisms. The determination of protein abundance change between two or more conditions has matured to the point that false discovery rates (FDRs) can be very low and for smaller proteomes coverage is sufficiently high to explicitly consider false negative error. Selected aspects of using these methods for global protein abundance assessments will be discussed. These include instrumental issues that influence the reliability of abundance ratios; a comparison of sources of nonlinearity, errors, and data compression in proteomics and spotted cDNA arrays; strengths and weaknesses of spectral counting and other nonlabel approaches versus stable isotope metabolic labeling; and a discussion of four microbiological applications of global abundance analysis at the protein level. The applications will include examples from ongoing studies of Porphyromonas gingivalis, Methanococcus maripaludis, Agrobacterium tumefaciens and Methylobacter extorquens AM1. The oral pathogen Porphyromonas gingivalis will be discussed as an example of an organism where a large percentage of the proteome differs in relative abundance between the intracellular and extracellular phenotype. Such a global analysis presents special challenges for existing approaches to normalization and multiple hypothesis testing, that typically assume only a small percentage of the proteome is changing with respect to relative abundance. Wholecell quantitative proteomic analyses were conducted to investigate the change from an extracellular to intracellular lifestyle for P. gingivalis, a Gramnegative intracellular pathogen associated with periodontal disease. Global protein abundance data for P. gingivalis strain ATCC 33277 internalized for 18 hours within human gingival epithelial cells and controls exposed to gingival cell culture medium were obtained at sufficient coverage to provide strong evidence that these changes are profound. A total of 385 proteins were overexpressed in internalized P. gingivalis relative to controls; 240 proteins were shown to be under expressed. Production of several proteases, including the classical virulence factors RgpA, RgpB, and Kgp, was decreased. A separate validation study was carried out in which a 16fold dilution of the P. gingivalis proteome was compared to the undiluted sample in order to assess the quantitative false negative rate (all ratios truly alternative), or FNR. Truly null (no change) abundance ratios from technical replicates were used to assess the rate of quantitative false positives (FPR) over the entire proteome. Similar studies of the methanogenic Archaeon M. maripaludis were carried out in which known dilutions of the entire proteome were treated as unknowns, and calculated abundance ratios were compared with the known true values. Spectral counting has poor statistical power characteristics (1FNR) but favorable FPRs and FDRs. A nonlabel approach based on summed signal intensity for each protein has better power but higher FPRs and FDRs. The best overall balance between FNR and FPR was achieved using traditional stable isotope metabolic labeling. While FNRs may not be as important for other applications, for studies of microbial gene expression complete quantitative proteome coverage is desired and FNRs become a convenient metric for both coverage and the degree of confidence that we can detect abundance change at a given level of power for any particular proteinencoding ORF.

MS.7

Quantitative Proteomics in the Context of Systems Biology

Simon J Gaskell
Michael Barber Centre for Mass Spectrometry, and Manchester Centre for Integrative Systems Biology, University of Manchester, Manchester, UK

The primary focus on relative quantification that has characterised the vast majority of the literature on quantitative proteomics now needs to be extended to allow absolute quantification to meet the emerging needs of systems biology approaches. No new analytical concepts are required "absolute quantification merely represents relative quantification when one of the comparators is known", but attention is directed to the need for multiple defined internal standards. Extending the principle of surrogacy that is successfully applied in most qualitative proteomics experiments, standards for quantification are typically proteolytic peptides that serve as signatures for the protein sequences from which they are derived. In conjunction with the Beynon laboratory in Liverpool, we have developed and exploited the simultaneous production of multiple isotopically labelled peptide internal standards by expression of artificial genes that correspond to concatenations of tryptic peptide sequences [1-3]. Selection of appropriate signature peptides may be facilitated by the application machine learning approaches to predict mass spectrometric detectability —work conducted by the Hubbard group in Manchester [4].
In order to achieve accurate and precise quantification of proteins, analytical lessons familiar from the determination of small molecules need to be applied. Thus there are clear advantages to the implementation of high-duty cycle mass spectrometric modes such as selected ion monitoring and selected reaction monitoring during LC-MS and LC-MS/MS analyses, respectively. Moreover, optimisation of the properties of peptides to match the analytical mode can lead to substantial enhancement of the sensitivity (and consequently quantitative precision) of analyses. We have explored the use of N-terminal thiocarbamoyl derivatives (Edman derivatives and analogues) in order to promote N-terminal peptide bond cleavage and thereby achieve a concentration of product ion current in one or two fragment ions following collisional activation of peptide ions. Substantial enhancements in the sensitivity of detection have been achieved [5]. Moreover, the implicit characterisation of tryptic peptide analytes in terms of molecular mass, N-terminal residue and chromatographic retention time has been shown to be adequate for highly selective detection.
These analytical approaches have been applied in several areas of absolute quantification, including the determination of enzymes involved in the yeast glycolysis pathway [6].

1. Beynon RJ, Doherty MK, Pratt JM, Gaskell SJ. Nature Methods 2005; 2: 587-589.
2. Pratt JM, Simpson DM, Doherty MK, Rivers J, Gaskell SJ, Beynon RJ. Nature Protocols 2006; 1: 1029-1043.
3. Rivers J, Simpson DM, Robertson DH, Gaskell SJ, Beynon RJ, Mol Cell Proteomics 2007, available on-line.
4. Kin Wai Lau, Siepen JA, Wedge D, Eyers C, Gaskell SJ, Hubbard SJ, Annual Conference of the American Society for Mass Spectrometry, 2007, Indianapolis.
5. Riba-Garcia I, Hart S, Beynon RJ, Gaskell SJ, in preparation.
6. Carroll K, Kell DB, Simpson DM, Beynon RJ, Gaskell SJ, unpublished.

MS.8

Analysis of Protein Levels and Phosphorylation Stoichiometry from Complex Samples using the iTRAQ Reagent

Jonathan C. Trinidad1, A.Thalhammer2, C.G. Specht2, P.R. Baker1, A.J. Lynn1, R. Schoepfer2, and A.L. Burlingame1
Mass Spectrometry Facility, Department of Pharmaceutical Chemistry, University of California, San Francisco, CA1 ; Department of Pharmacology, University College London, United Kingdom2

This talk will discuss practical issues involved in the quantitative analysis of protein levels and phosphorylation stoichiometry, focusing on: the use of the iTRAQ reagent for isotopic quantitation; and on the development of software tools to aid in making sense of the results.
Acquiring knowledge of protein and post-translational dynamics is a crucial step in understanding the functioning of complex cellular environments. Towards this end, we have been focused on developing methods to acquire, process, and derive insight from large-scale studies of protein levels and phosphorylation from biological samples under a range of physiological conditions. The experiments that will be discussed involve the application of the iTRAQ (Isobaric Tags for Relative and Absolute Quantitation) reagent.
There are two main advantages of the iTRAQ reagent for our purposes. Firstly, the reagent reacts with free alpha and epsilon amines, leading to labeling of essentially all potential peptides resulting from a tryptic digest (a key consideration when doing quantitation of phosphorylation using isotopic labeling). Secondly, the multiplexed nature of the reagent allows up to four samples to be processed simultaneously. To obtain statistically reliable quantitative information regarding cellular states, larger numbers of replicates need to be analyzed than is currently prevalent in the literature. Multiplexed sample analysis will help to alleviate this bottleneck in the sample processing pipeline.
In terms of developing tools to aid in the analysis of large-scale quantitative proteomic data, much remains to be done. We have invested considerable efforts into quantitative aspects of data analysis using the Protein Prospector software package. The software now supports a range of isotopic quantitation techniques. We have begun to integrate Protein Prospector with downstream processing software that calculates such things as protein expression ratios from multiple peptides as well as relative phosphorylation stoichiometry (given protein and phosphorylation quantitation information).

1.1

Expression proteomics at last: The determination of proteome wide protein abundance changes by SILAC and high resolution mass spectrometry

L. de Godoy, J. Cox, J. Olsen, T. Bonaldi, C. Kumar, N. Hubner, B. Macek and Matthias Mann
Department of Proteomics and Signal Transduction, Max-Planck Institute for Biochemistry, Martinsried, Germany

Mass spectrometry based proteomics has become increasingly successful at the identification of a large proportion of the proteins in complex mixtures. However, the goal of quantifying complete proteomes has remained elusive for the last three decades. Here we use Stable Isotope Labeling by Amino acids in Cell culture (SILAC), together with LTQ-Orbitrap based mass spectrometry and sophisticated bioinformatics to quantitate a large proportion of the proteome. We will cover the work flow used in these experiments including automated statistical analysis of quantitation results. Results from yeast, drosophila and human proteomes will be introduced. In drosophila we demonstrate quantitation of more than 4400 proteins after dsRNA knock down of a chromatin remodeling factor. Interestingly, the second most down-regulated protein - after the RNAi target itself - is a protein in a direct complex with the target. The message level was not affected, demonstrating a key advantage of measuring proteins in addition to mRNA levels in 'systems biology'. In the human system we have compared the comprehensiveness of modern proteomic methods with standard microarray methods. We conclude that proteomics is still more time consuming but that it is now competitive with microarrays in terms of coverage of gene expression.

1.2

Statistical Analysis of Quantitative Proteomics Datasets Via Normalized Spectral Abundance Factors

Michael P. Washburn
Stowers Institute for Medical Research, Kansas City, MO, USA

Quantitative proteomic experimentation can now yield expression data for hundreds to thousands of proteins. Often times these datasets will be generated with the intention to discover new biology that will be followed up with focused biochemical, cell biological, or molecular biological experimentation. Our laboratory has been focused on the use of a modified form of spectral counting, named the normalized spectral abundance factor (NSAF) as the basis for quantitative proteomic analysis in combination with multidimensional protein identification technology. Since proteases like trypsin largely generate more peptides per protein as any given protein increases in length. The key features of the NSAF approach is that it takes into account the length of the protein being analyzed and the total intensity of the run. NSAF values alone do not follow a normal distribution and must be natural log transformed in order to use statistical tests like the t-test. When we investigated the statistical parameters of NSAF values compared to Affymetrix GeneChip datasets, we found that data from both approaches have similar dynamic range and distribution properties of numeric values. In addition, we observed that the standard deviation (SD) of a protein's NSAF values is dependent on the average NSAF value of the protein itself, following a power law. We have begun applying these approaches to elucidate the molecular mechanisms of rapamycin, an immunosupressive and anticancer drug. Rapamycin inhibits the protein kinase TOR (Target of Rapamycin), a central controller of cell growth in eukaryotic organisms. The targets of TOR have not yet been identified and gaps in the signalling pathway remain to be filled. In order to gain insights into TOR function and a global understanding of rapamycin effects, we performed a time course analysis of protein and mRNA changes in Saccharomyces cerevisiae in response to rapamycin. Proteomic and transcriptomic profiles have been compared between Saccharomyces cerevisiae grown in N15 media (none treated) versus N14 media (treated) during 6 hours of rapamycin treatment. Interestingly, the correlation between mRNA and protein abundance changes was low at any given time point but increased in a delayed fashion when changes at the mRNA level at early time point were correlated with changes at the protein level at later time point. This seminar will focus on the application of the NSAF approach to protein expression analysis and the statistical analysis of these datasets. For more information please see the posters by Pavelka et al and Fournier et al.

1.3

Insight into cell biology from physical and genetic protein interactions

Gerard Cagney1, N. Krogan2, J. Weissman2, A. Emili3, J. Greenblatt3
Conway Institute, University College Dublin, Ireland1 ; Department of Cellular & Molecular Pharmacology, University of California, San Francisco, CA, USA2 ; Department of Medical Research, University of Toronto, Ontario, Canada3

Pathways and complexes can be considered fundamental units of cell biology, but their relationship to each other is difficult to define. Comprehensive tagging and purification experiments have generated a network of interactions that probably represents most of the stable protein complexes in mitotically-growing yeast cells. We describe this work, and show how the analysis of pairwise epistatic relationships between genes complements the physical interaction data, and furthermore can used to classify gene products into parallel and interacting pathways.

1.4

Exploiting peptide identification databases for improved bioinformatics for proteomics

Simon Hubbard
Faculty of Life Sciences, University of Manchester, United Kingdom

Modern proteome science and bioinformatics are closely coupled. High-throughput LC-MS/MS experiments are generating large numbers of peptide identifications which are of course dependent on bioinformatics tools to assign the most likely amino acid sequences from protein sequence databases. This growing number of peptide identifications provides an excellent resource from which to mine useful principles, learn rules and feedback into the peptide identification process. We have developed PepSeeker (www.ispider.manchester.ac.uk/pepseeker), once such peptide identifications database which captures explicit fragment ion information and supports queries over this data coupled to amino acid sequence patterns. The current version contains over 2 million spectra and 250000 peptide identifications and can be accessed via a BioMart interface. Peptide identifications contained in PepSeeker and other public repositories have been used to examine several features of proteome peptides, including distributions of fragment ion types in different instruments, sequence signals associated with missed cleavage by trypsin, and peptide "flyability" prediction for the design of quantitative QconCAT proteins. These applications will be discussed in the context of the peptide identification process, such as a simple database processing step which can improve identifications via peptide mass fingerprinting. Finally, peptide identifications themselves can be used to improve the genome annotation process and an example will presented showing how proteomics can help rationalise different gene model predictions for the Aspergillus niger genome.

1.5

Refining Peptide LC/MS Analysis for Qualitative and Quantitative Proteomics

A.A. High and Clive A. Slaughter
St. Jude Children's Research Hospital, Memphis, TN, USA

Recent improvements in mass spectrometer sensitivity and scan rate, along with enhanced peak capacities available with a new generation of chromatographic stationary phases, offer the potential for greater sensitivity in protein identification and greater penetration of proteomes. We have coupled one- or two-dimensional liquid chromatography at high resolution using 1.7 μm stationary phases with mass analysis using a rapid scanning, linear ion-trap mass spectrometer, and have developed procedures that take advantage of the capabilities these tools provide. Specifically, we have evaluated the gains realized through attention to instrument configuration, sample loading, off-line fractionation parameters and mass spectral acquisition settings. We illustrate the effects of these gains in a qualitative study of intrinsically unstructured proteins, and in a quantitative study of the MHC-linked presentation of peptides derived from influenza proteins in virus-infected cells. We approach the task of comparing the capabilities of the new methodology with those required for global proteome analysis, and consider the factors presently limiting performance. Achieving an accurate understanding of these issues, however, is complicated by the frequency with which commonly used search engines either misassign spectra or fail to assign them. These problems highlight the need for continuing improvements in search methods and for making raw spectral data available when publishing proteomic results.

2.1

Deciphering the Dynamics of Proteasome Interacting Proteins Using Quantitative Mass Spectrometry

X. Wang and Lan Huang
Departments of Physiology & Biophysics and Developmental & Cell Biology, University of California, Irvine, CA

Protein-protein interactions are one of the major mechanisms for controlling protein functions in various cellular processes. In order to fully understand these functions, global mapping of protein-protein interactions has become a major goal in current proteomics research. In combination with affinity purification, mass spectrometry-based interactive proteomics has become the method of choice for analyzing functional protein complexes. In addition to the interactions of varied affinity, the dynamics of protein interactions is another important aspect for understanding the functions of protein interactions. The dynamics can be classified into two layers: dynamic protein interactions that change temporally with cellular states at different time of extracellular signaling, and those that have high association and dissociation (i.e. on/off) rates with their interacting partners. Although much effort has been made to understand the first layer of interaction dynamics, lack of an efficient strategy to distinguish stable and dynamic interactors has hampered the efforts towards the understanding of the second type. In this work, we have developed a new method, MAP (mixing after purification)-SILAC to quantitatively investigate the interactions of protein complexes by mass spectrometry. In combination with the original SILAC approach, stable and dynamic components are unambiguously distinguished based on their relative abundance ratio changes. We applied the newly developed strategies to decipher the dynamics of the 26S proteasome interacting proteins (PIPs), and have identified new putative PIPs while the nature of the identified interactors was fully characterized. The methods reported here provide a valuable expansion of proteomics technologies for identification of important but previously unidentifiable interacting proteins.

2.2

The Proteome Biology of Proteasome Complexes: Molecular Organization, Function, and Regulation.

Peipei Ping
Department of Physiology and Medicine/Cardiology, NHLBI PPG on Myocardial Ischemia Injury, Proteomic Core Laboratory at CVRL, University of California, School of Medicine, Los Angeles, CA

The proteasome system plays a critical role in governing the intracellular protein degradation processes in mammalian cells. The proteome biology of these proteasome complexes, i.e., the assembly of proteasome subunits, the post-translational modifications of the subunits, and the associating partners of the complexes, dictates the intracellular proteolytic functions that these multi-protein complexes perform. Using an organelle proteomic approach, our recent investigations have defined the molecular organization, function, and regulation of this organelle in mammalian cell types.

2.3

Identification of newly synthesized proteins using bioorthogonal noncanonical amino acid tagging (BONCAT).

D.C. Dieterich1, A.J. Link2, D.A. Tirrell2, J. Graumann1, and Erin M. Schuman¹
Division of Biology/HHMI1 & Chemistry2, Caltech, Pasadena, CA

Alterations in protein synthesis and degradation enable cells, including neurons, to adapt to changing external conditions. In neurons, there is increasing evidence that local dendritic protein synthesis is used to allow individual synapses to respond dynamically to the environmental changes that accompany the establishment, maintenance and plasticity of synaptic connections. The identification of the activity-modulated dendritic proteome promises to offer a more thorough understanding of synaptic plasticity at the molecular level. To isolate and identify dendritically synthesized proteins we have developed a new protein tagging strategy that can be used in combination with mass spectrometry. The protein tagging is based on an azide-alkyne ligation using the azide-group bearing non-canonical amino acid azidohomoalanine (AHA), which serves as a surrogate for methionine. Proteins bearing AHA can subsequently be tagged with an alkyne-bearing affinity tag. After tryptic digestion of affinity-purified proteins, mass spectral analysis is achieved by utilizing MudPIT (Multidimensional Protein Identification Technology) followed by bioinformatical analysis. Initial experiments show, that AHA can be incorporated into newly synthesized proteins of HEK293 cells and cultured hippocampal neurons. In control experiments where AHA was replaced with methionine, no biotinylated proteins were recovered following avidin-chromatography. In tandem mass spectrometry analysis of avidin-purified proteins from AHA (2 hr)-treated whole cell lysates of HEK293 more than 190 proteins, including an overexpressed control protein, were identified. To identify the dendritic proteome, newly synthesized proteins from either rat brain synaptoneurosomes or isolated dendrites of hippocampal cultures are analyzed.

D.C.D. was supported by the German Academy for Natural Scientists LEOPOLDINA (BMBF-LPD9901/8-95).

3.1

Innovative Technology for the Study of Cell Signaling

Donald F. Hunt
Departments of Chemistry and Pathology, University of Virginia, Charlottesville, VA

This lecture will focus on the use of electron transfer dissociation (ETD) to identify proteins and to characterize their post-translational modifications on a chromatographic time scale. In this experiment, proteins are fractionated by nano-flow HPLC, converted to gas-phase, positive ions by electrospray ionization, and allowed to react with fluoranthene radical anions inside a linear trap mass spectrometer. Electron transfer to the multiply charged protein promotes random fragmentation of amide bonds along the protein backbone. Multiply charged fragment ions are then de-protonated in a second ion/ion reaction with the carboxylate anion of benzoic acid. MS and ETD-MS/MS spectra are recorded every 500 msec. The m/z values for the resulting singly, doubly, and triply charged ions are used to read a sequence of 15-60 amino acids at both the N and C termini of the protein. This information, along with the measured mass of the intact protein, is used to identify unknown proteins, to confirm the amino acid sequence of a known protein, to detect post-translational modifications, and to determine the presence of possible splice variants.
Presented first will be results of studies to characterize post-translational modifications that involve acetylation, methylation and O-GlcNAcylation.
Part two of the presentation will describe results of studies to define sites of phosphorylation that regulate the formation of focal adhesions involved in cell migration. For this work we employ immobilized metal affinity chromatography (IMAC) to identify phosphorylaton sites present at levels as low as 0.1% of the parent protein. Stable isotopes are employed to follow changes in site usage as a function of cellular perturbation. Further information on this topic is available at cellmigration.org (site guide, phosphoproteomics).
Part three of the presentation will focus on signaling between cancer cells and cytotoxic killer cells by the class I, antigen-processing pathway. Since signal transduction pathways in cancer cells are highly dysregulated, we hypothesized that this might manifest itself in the presentation of unique phosphopeptides by the cancers to the immune system in association with Class I MHC molecules. By using a combination of IMAC, stable isotope labeling, and nano-flow HPLC-tandem mass spectrometry, we are able to detect cancer-specific, Class I phosphopeptides present at levels as low as 1-5 copies per/cell. Recent studies show that the Class II antigen processing pathway also presents phosphopeptides. Results of studies on melanoma, ovarian, breast, and lymphoma cancers will be described.

1. Peptide and Protein Sequence Analysis by Electron Transfer Dissociation Mass Spectrometry, Syka JEP, Coon JJ, Schroeder MJ, Shabanowitz J, Hunt DF, Proc Natl Acad Sci USA 2004;101:9528-9533
2. Protein Identification Using Sequential Ion/Ion Reactions and Tandem Mass Spectrometry, J.J. Coon, B. Ueberheide, Syka JEP, Dryhurst DD, Ausio J, Shabanowitz J, Hunt DF. Proc Natl Acad Sci USA 2005;102:9463-9468.
3. Novel Linear Quadrupole Ion Trap/FT Mass Spectrometer: Performance Characterization and Use in the Comparative Analysis of Histone H3 Post-Translational Modifications, Syka JEP, Marto JA, Bai DL, Hornung S, Senko MW, Schwartz JC, Ueberheide BM, Garcia BA, Busby SA, Muratore T, Shabanowitz J, and Hunt DF, J Proteome Res 2004;3:621-626.
4. Dynamic Regulation of HP1/Chromatin Interaction by Methylation and Phosphorylation of Histone H3. Fischle W, Tseng BS, Dormann H, Ueberheide BM, Garcia BA, Shabanowitz J, Hunt DF, Funabiki H, Allis CD. Nature 2005;438, 1116-1122.
5. Identification of Class I MHC-Associated Phosphopeptides as Targets for Cancer Immunotherapy, Zarling, AL, Polefrone, JM, Evans, AM, Mikesh, LM, Shabanowitz J, Lewis AT, Engelhard, VH, and Hunt DF, Proc. Natl. Acad. Sci. USA, 2006, 103, 14889-14894.
6. Analysis of Intact Proteins on a Chromatographic Time Scale by Electron Transfer Dissociation Mass Spectrometry, Chi, A, Bai, DL, Geer, LY, Shabanowitz, J, and Hunt, DF, Int. J. Mass Spectrom. 2007, 259, 197-203.

This research was supported by NIH grants, GM-37537, AI-33993 and U54 GM-64346.

3.2

TiMAC - A new phosphoproteomic strategy for the separation of mono- from multiply phosphorylated peptides combined with optimized tandem MS mass spectrometric analysis

T.E. Thingholm and Martin R. Larsen
Department of Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark.

The ability to selectively isolate and analyze the complete complement of phosphorylated peptides from complex peptide mixture is essential for performing comprehensive phosphoproteomics. However, despite the numerous strategies used in modern phosphoproteomics including the use of IMAC, TiO2 chromatography or phosphoramidate chemistry for affinity enrichment of phosphorylated peptides prior to mass spectrometric analysis no single method is able to provide a complete coverage of the phosphoproteome of a given sample. This has recently been shown by (Bodenmiller et al., Nature Methods 2007). One of the reasons for this large discrimination between the strategies is most likely that phosphopeptides are very different both with respect to amino acid contents and number of phosphate groups attached to each peptide, providing a different binding affinity to the chelating material. In addition, the micro-environment in which a phosphate group is located, represented by amino acid functional groups, will eventually have a large influence on the physio-chemical properties of the resulting phosphopeptide and thereby also contribute to the difference in the binding affinity to different chelating material. The comprehensive analysis of phosphopeptides is furthermore compromised by common mass spectrometric instrumentation which favors the analysis of mono-phosphorylated peptides over multiply phosphorylated peptides due to ion suppression effects and inadequate peptide fragmentation.
To circumvent some of the problems in large scale phosphoproteomics of low amount of sample, listed above, we have developed a new method, TiMAC, which combine the advantages of IMAC and TiO2 chromatography into one strategy without the use of additional material. We show that by using the TiMAC strategy we are able to separate mono-phosphorylated peptides from the multiply phosphorylated peptides prior to MS analysis. The separation of the two species allowed us to optimize the subsequent pdMS³ experiment to favor analysis of either mono-phosphorylated or multiply phosphorylated peptides. When applying this new method to a total of 120 μg protein from human mesenchymal stem cells more than twice the number of phosphorylation sites was identified compared to an optimized TiO2 protocol.

3.3

Identification of the missing components of a receptor kinasemediated steroid signal transduction pathway using quantitative proteomic profiling

W. Tang1, Z. Deng1, J.A. OsesPrieto2, X. Zhang2, N. Suzuki2, S. Guan2, R.J. Chalkley2, A.L. Burlingame2, and ZhiYong Wang1
Department of Plant Biology, Carnegie Institution, Stanford, CA1 ; Mass Spectrometry Facility, Department of Pharmaceutical Chemistry, University of California, San Francisco, CA2 ; Institute of Molecular Cell Biology, Hebei Normal University, Shijiazhuang, Hebei, China3

Brassinosteroid (BR) is a steroid hormone that plays essential roles in plant growth and development. BR is perceived by the cell surface receptorlike kinase BRI1, and downstream signal transduction involves another receptorlike kinase (BAK1), a GSK3like kinase (BIN2), and nuclear transcription factors (BZR1 and BZR2/BES1). BR binds to the extracellular domain of BRI1 to induce BRI1BAK1 dimerization and receptor kinase activation, leading to inhibition of BIN2 and dephosphorylation of the BZR1 and BZR2/BES1 transcription factors. When BR levels are low, BIN2 phosphorylates BZR1 and BZR2/BES1 to inhibit their nuclear accumulation and DNA binding. How the receptor kinases regulate BIN2 activity remains the only major gap in our understanding of the pathway. In order to identify additional proteins involved in BR signal transduction and BRregulated physiological response, we used twodimensional difference gel electrophoresis (2D DIGE) to identify BRregulated proteins in total protein, plasma membrane (PM), and phosphoprotein fractions. While proteomic changes were detected in total protein samples only after 3 hr of BR treatment, obvious responses were observed in PM and phosphoprotein fractions after a 0.5 hr BR treatment. Using tandem mass spectrometry (LCMS/MS), we identified over 65 BRregulated proteins. While none of the known BR signaling proteins were identified in total protein samples, BAK1 and BZR1 were identified in the PM and phosphoprotein fractions, respectively, as proteins that show BRinduced phosphorylation changes (spot shift along the IEF dimension). In addition, several novel proteins were identified and shown to function in BR response. Among these are two novel kinases that showed BRinduced phosphorylation changes. These two kinases are closest homologs of each other in the Arabidopsis genome, and we named them BRregulated kinase 1 and 2 (BRK1 and BRK2). To understand the function of BRK1 and BRK2 in BR signaling, we have shown that overexpression of BRK2 and its homologs suppresses the receptor mutant bri1, suggesting that BRK1 represents a family of positive regulators of BR signaling. Preliminary results indicate that BRK1 and BRK2 interact with both BRI1 and BIN2 in vivo, and are phosphorylated by BRI1 in vitro, suggesting that BRK1 and BRK2 mediate signal transduction from the receptor kinase to BIN2. Therefore, it is likely that our proteomic study has closed the last major gap in the BR signaling pathway, establishing the BR pathway as the first complete signal transduction pathway from cell surface receptor kinases to the nucleus in plants as well as for steroid hormones.

3.4

Comprehensive Phosphoproteome Analyses of Leukemic Hematopoietic Stem Cells to Uncover the Molecular Basis of Self-Renewal

M. Trost1 ; O. Hérault1 ; M. Marcantonio1 2 ; A. Faubert1 ; C. Pomies1 ; G. Sauvageau1; Pierre Thibault123
Institute of Research in Immunology and Cancer1, Department of Biochemistry2, Department of Chemistry, Université de Montréal, Montréal, Canada3

Hematopoietic stem cells (HSC) maintain blood formation throughout life time and are typified by their ability to self-renew indefinitely and differentiate into myeloid and lymphoid blood cell lineages. Pathways regulating the maintenance of adult HSC remain poorly understood primarily due to the inability of expanding HSC in vitro and by the difficulty of isolating pure in vivo stem cells in sufficient amount. Significant advances in HSC research have been obtained through the development of different leukemia model systems, and our group previously reported the isolation of HSC with self-renewal capacity by introducing oncogenes into highly purified cell populations (Ska+, C-Kit+, Lyn-) [1-2]. In this study, we compared the phosphoproteome of mice leukemia model systems showing different frequency of self renewal to investigate specific cell signaling events associated to their distinctive phenotypes. Nuclear and cytosol extracts from 108 cells of leukemic HSC and normal granulocytes from mouse models were were lysed and centrifuged to isolate nuclear and cytosol protein extracts. Phosphopeptides were enriched using TiO2 affinity media and analyzed by 1D and 2D LC-MS/MS on an Orbitrap mass spectrometer. In-house software was used to obtain expression profiles and MS/MS spectra were searched against the IPI database using Mascot. On-line 2D-LC-MS/MS provided enhanced sample loading and capacity for phosphopeptide identification with 2915 different phosphopeptides assigned to 1117 unique proteins. Comparison of phosphopeptide abundances in nuclear and cytosol extracts of HSC enabled the identification of phosphoproteins showing cellular translocation including kinases (e.g. Dyrk 1A, STK 10) and low abundance transcription factors (e.g. ETV6, ATRX) that were also correlated with leukemic stem cells of different self-renewal capacity. This study will highlight the analytical advantages of the present approach to identify cellular markers of self renewal in HSC mouse model systems of leukemia.

1. A. Mamo, et al, Blood, 2006, 108, 622-629.
2. J.K. Krosl, A. Faubert, G. Sauvageau, Hematol J., 2004, 5, S118-21

3.5

New Technology to Detect and Monitor the Post-Translational Modification Events that Commit Human Embryonic Stem Cells to Exit the Pluripotent State

D.L. Swaney1, G.C. McAlister1, D. Phanstiel1, J. Brumbaugh2, J. Keith1, S.B. Ficarro4, X. Feng3, V. Ruotti5, R. Stewart5, J.A. Thomson3, T.Berggren5, and Joshua J. Coon1 2
Departments of Chemistry1, Biomolecular Chemistry2, and Anatomy3, University of Wisconsin, Madison, WI; Department of Genomics, Novartis Foundation, San Diego, CA4; WiCell Research Institute, Madison, WI5

We describe the use of new mass spectrometry technology — electron transfer dissociation coupled with ultra high resolution mass analysis (orbitrap) — to discover and quantify the post-translational modification events that signal human ES cells to exit the pluripotent state. Human ES cells are grown in feeder-independent, chemically defined (TeSR1) culture and exposed to doses of various differentiating agents over multiple timepoints. Following treatment, the cells are lysed, the proteins isolated, and enzymatically digested. Using affinity chromatography (IMAC) coupled with ETD-MS/MS on our ETD-enabled orbitrap we have discovered nearly 4000 sites of phosphorylation in human ES cells. Next, we have characterized the modification status of histone H4 in pluripotent and differentiating human ES cells. This work reveals no less than 70 unique histone H4 modification patterns exist in ES cells and that these patterns change upon differentiation. Finally, we describe new technological capabilities afforded by the implementation of ETD on the hybrid orbitrap mass spectrometer. For example, we have designed a decision tree-based data-dependent method that automatically selects whether to perform CAD or ETD, based on the precursor charge, mass and abundance, and whether to perform mass analysis at high or low resolving power. For complex mixture analysis this method nearly doubles the probability that any given MS/MS event will lead to a successful identification.

4.1

Interrogation of cellular signalling pathways using protein microarrays

Jan van Oostrum
Novartis Institutes of BioMedical Research, Novartis AG, Basel, Switzerland

Molecular signalling pathways are frequently triggered by extracellular molecules binding receptors and activating relay systems inside cells, leading to processes that affects cellular behaviour and fate. For many genetic disorders a link between disease and signalling pathways have been established consequently a systematic analysis of dynamic cellular networks provides an opportunity for pharmaceutical discovery, by taking into consideration the complex biological context of drug targets, rather than observing the targets in isolation. Such analyses are, perhaps, ideally suited for a systems biology approach that integrates experimental data with computational modeling with the aims of discovering and validating new drug targets and biomarkers, as well as predicting potential "off target" effects of drug candidates. Informing, calibrating and validating mathematical models with experimental data is a key component of an applied systems biology investigation and a number of genomic and proteomic techniques can be employed to generate these crucial data sets. These techniques range from LC-MS/MS based discovery of phosphopeptides to genome-wide cDNA and RNAi functional screens. Recently, we implemented and optimized a proteomics platform based on "reverse" protein arrays (RPA) that is particularly suitable to monitoring cell signalling events. These arrays are based on the principle that complex protein mixtures or proteomes (such as cell or tissue lysates) are spotted in an array format and probed with selected fluorescent antibodies in a multiplexed manner. To ensure high levels of sensitivity and signal to noise ratio of these RPAs, we are using planar waveguide technology. The advantage of the evanescent field fluorescence detection ensures that only analyte-bound fluorescent antibodies contribute to the signal. This method make it feasible to obtain quantitative and kinetic protein expression profiles and signaling information in a wide variety biospecimen.

4.2

Dynamic Interplay between GlcNAc and Phosphate on Regulatory Proteins

Gerald Hart
Johns Hopkins University School of Medicine, Baltimore, MD

OGlcNAcylation is a dynamic modification of Ser(Thr) residues on signaling, transcription and cytoskeletal proteins in all metazoans. GlcNAcylation is very abundant and widespread, and often competes with phosphorylation, but also it directly plays an important role as a nutrient/stress sensor underlying glucose toxicity and diabetes. Studies during the last two decades have revealed serious limitations in both electrospray MS/MS and MALDITOF MS for the analysis of GlcNAcylation. However, recently, several chemicoenzymatic methods have been developed for the site mapping and quantification of OGlcNAc. In addition, both FTMS with ECD and iontraps with ETD are proving valuable for analyzing labile OGlcNAc residues on proteins. We are currently using these methods to better understand both the global and specific interplay between GlcNAcylation and phosphorylation.

Supported by NIH grants HD13563, CA42486, DK61671, DK71280, and NIH contract N01HV28180. Dr. Hart receives a share of royalty received by the university on sales of the CTD 110.6 antibody. Terms of this arrangement are managed by JHU.

4.3

Mass spectrometry as a detector for protein ubquitination

Steven Gygi
Department of Cell Biology, Harvard Medical School, Boston MA, USA

Of the major cell regulatory pathways, the biochemical complexity and functional diversity of protein ubiquitination make it an excellent candidate for large-scale proteomic studies. From the profiling of ubiquitin-protein conjugates by shotgun sequencing to the quantitative analysis of polyubiquitin chains, mass spectrometry is making critical advances in ubiquitin biology. In this lecture, we will describe a method for the detection and quantification of the levels of substrate protein, total ubiquitin, and each potential chain linkage. We demonstrate these measurements for conjugates formed by APC-catalyzed ubiqutination of cyclin B1.

5.1

Systems-wide analysis of protein complexes in Saccharomyces cerevisiae

Anne-Claude Gavin
European Molecular Biology Laboratory, Heidelberg, Germany

The omics field contributes comprehensive repertoires of the cell building blocks. The next challenge resides in the understanding of how the pieces of this puzzle assemble a coherent entity; a cell. Biology relies on the concerted action of a number interacting proteins and metabolites operationally organized in cellular networks. The current appreciation of the wiring diagram of these networks is scanty. We used tandem-affinity purification and mass spectrometry to achieve a system-wide analysis for protein complexes in a model organism, budding yeast. The study provides one of the largest collections of physically-determined eukaryotic cellular machines; 491 complexes, of which 257 were novel. Beyond the repertoire, the analysis captures the modular nature of proteomes, where protein complexes differentially combine with additional attachment proteins or protein modules to enable a diversification of functions. Support for this organisation comes from integration with available data on expression, localisation, function, evolutionary conservation, protein structure and binary interactions. An innovative scoring system measures the potency of proteins to associate. It represents an attempt to move from static interaction networks to more dynamic maps. This first proteome equation captures some biochemical properties of protein-protein interaction: the likelihood to be in direct physical contact and, weakly, the dissociation constants. In the future, experimental and computational refinements may turn such scoring approaches into suitable parameters for rational modeling of entire systems.

5.2

Degradomics: the Proteolysis of Cell Death

S. Mahrus, J. Trinidad, A. Burlingame and James A. Wells
Department of Pharmaceutical Chemistry, University of California at San Francisco, San Francisco, CA

Apoptosis, or programmed cell death, represents an ultimate fate decision in cell biology. This process is critical for cellular differentiation and remodeling of tissues, and for anti-viral and anti-tumor defense. A distinct molecular feature of apoptosis is the widespread but controlled cellular proteolysis, that is predominantly mediated by the caspase family of cysteine proteases. The study of apoptotic pathways has important ramifications for determining what is critical for cellular homeostasis, and for the development of potential anti-cancer therapeutics. We have developed a robust proteomic method for global profiling of proteolysis ("degradomics") in cells. Key to this is a new method that permits selective labeling and enrichment for the protein N-termini created as a result of proteolysis. Using this approach we have already identified >250 caspase substrates from Jurkat cells that were induced to undergo apoptosis by treatment with the chemotherapeutic agent etoposide. More than 80% of these proteins have not been reported before. These proteins fall into a wide range of functional classes, and reveal much about the molecular components, logic, and timed sequence of events that drive a cell from life to death. We believe this approach will be useful for following the proteolysis of apoptosis induced by various agents or in different cell types, and will be generally useful for dissecting protease signaling pathways.

5.3

Characterization of GSK3 Dependent Circadian Phosphoproteome

Krista Kaasik1, J. Allen2, K. Shokat2, L.J. Ptácek1 and Y-H Fu1
Department of Neurology1, Department of Pharmaceutical Chemistry2, University of California, San Francisco, CA

Circadian rhythms are an adaptation to the daily solar cycle and are produced by transcriptional feedback loops. Defining features like period length and self sustained oscillations are regulated by protein kinases that phosphorylate clock associated transcription factors. Glycogen synthase kinases (GSK3α and GSK3β) are among key regulators of circadian rhythms; however the functional significance of GSK3 catalyzed protein phosphorylation in the circadian proteome is poorly understood. GSK3 protein kinase activity itself has a circadian profile. Phosphorylated N-terminus of GSK3 prevents substrate bindings and blocks access to its catalytic center. Therefore we hypothesized that a subset of GSK3 substrates are regulated in a circadian manner. In order to identify direct GSK3 targets we are using chemical genetics approach and we have designed ATP analog-specific (as) kinases, which utilize bulky ATP analogs that are not utilized by wild-type kinases. We have further refined this method to allow antibody based detection and immunoaffinity isolation of kinase substrates. Immunoprecipitation with a thiophosphate-ester specific antibody allows isolation of direct kinase substrates from complex proteomes in the presence of hundreds of other kinases, which is particularly important for GSK3 which often requires a prior phosphorylation priming event to recognize its substrates.We have determined that a subset of GSK3 substrates in liver and brain tissues have a circadian phosphorylation pattern. Furthermore, we are in the process of identifying potentially novel GSK3 targets. Knowing the specificity of GSK3α and GSK3β substrates and their circadian regulation will assist in design of novel therapies and more efficient use of current GSK3 inhibitors in medical treatment with fewer side effects.

5.4

Interactome Networks

Marc Vidal
Center for Cancer Systems Biology (CCSB) and Department of Cancer Biology, Dana-Farber Cancer Institute, Department of Genetics, Harvard Medical School, Boston, MA, USA

For over half a century it has been conjectured that macromolecules form complex networks of functionally interacting components, and that the molecular mechanisms underlying most biological processes correspond to particular steady states adopted by such cellular networks. However, until recently, systems-level theoretical conjectures remained largely unappreciated, mainly because of lack of supporting experimental data.
To generate the information necessary to eventually address how complex cellular networks relate to biology, we initiated, at the scale of the whole proteome, an integrated approach for modeling protein-protein interaction or "interactome" networks. Our main questions are: How are interactome networks organized at the scale of the whole cell? How can we uncover local and global features underlying this organization, and how are interactome networks modified in human disease, such as cancer?

6.1

The Study of Organelle Dynamics Using Stable Isotope Labeling

Kathryn S. Lilley
Cambridge Centre for Proteomics, Department of Biochemistry, University of Cambridge, UK

Cells are organized spatially and functionally into sub-cellular compartments. Changes in sub-cellular localization are involved in regulation of interactions, stability and activity. Studying global changes in protein localization can provide useful insights into cellular functions.
We have previously developed LOPIT, a high-throughput technique for protein localization to sub-cellular organellesa,b.. Organelles are partially separated by density gradient centrifugation and fractions labeled with stable isotopic iTRAQ tags for quantitation. Proteins from the same organelle co-sediment and exhibit similar distributions in the density gradient. Sub-cellular localization can be assigned by comparing distributions of unknown proteins to those of known organelle markers. Using this approach we have been able to assign hundreds of proteins to different sub-cellular locations within Arabidopsis cultured cells.
We have now extended LOPIT to map protein localization within Drosophila melanogaster embryos and vertebrate cell lines. Data from these studies will be presented along with data which demonstrates that protein super-complexes such as the proteasome can also be mapped using this method.
We are currently using the LOPIT technique to dynamically map protein redistribution upon a given perturbation. The key to this process is to be able to control for technical variability within the LOPIT experimental schema, which otherwise may obscure genuine translocation of protein species between sub-cellular locations.
Methods to account for the technical variability associated with this approach will be introduced.

1. Dunkley TP, Hester S, Shadforth IP, Runions J, Weimar T, Hanton SL, Griffin JL, Bessant C, Brandizzi F, Hawes C, Watson RB, Dupree P, Lilley KS. (2006) Proc Natl Acad Sci 103(17):6518-23.
2. Sadowski PG, Dunkley TP, Shadforth IP, Dupree P, Bessant C, Griffin JL, Lilley KS. (2006) Nat Protoc. 1(4):1778-89.

6.2

Elucidating Interacting Regions Within a Large Protein Complex Through Chemical Crosslinking and Mass Spectrometry

O.W. Nadeau1, A. Artigues1, M.D. Jeyasingham1, J. Sage1, M.T. Villar1, G.J. Wyckoff2, and Gerald M. Carlson1
University of Kansas Medical Center, Kansas City, KS1; University of Missouri–Kansas City, Kansas City, MO2

Recent developments in MS technologies have significantly reduced the amount of material required to fully analyze modified peptides, which has led to a resurgence of protein chemistry techniques, including chemical crosslinking. For proteins that are not readily amenable to X-ray diffraction analysis, such as large oligomeric complexes, the identification of crosslinked amino acids by MS provides an alternative approach for obtaining relatively high resolution structural information. Despite the resolving power of MS, however, considerable computational analysis of digests from crosslinked proteins is required to account for all the possible masses that can arise not just from intermolecular crosslinking, but also from intramolecular crosslinking, incomplete digestion, monoderivatization, and hydrolysis and other side reactions that can occur on the crosslinker itself. Consequently, assignment of masses from crosslinked digests is invariably the slowest step in this structural approach. To facilitate analysis and assignment, we have constructed a search engine that generates best matches for crosslinked peptides, based on the identity of the crosslinker and on theoretical peptide maps of the proteins in question as digested by a given protease. This engine was used to identify crosslinked residues among the various subunits of the large model oligomeric complex phosphorylase kinase (PhK).
PhK from skeletal muscle is a stable complex of sixteen subunits, (α β γ δ)4, and a mass of 1.3 MDa. Its γ subunit is catalytic, while the remaining subunits are regulatory, giving rise to activation through such mechanisms as their phosphorylation or the binding of Ca2+ ions. In fact, 90% of PhK's mass is involved in its regulation. Thus, it has been of considerable interest to determine which regions of PhK's regulatory subunits interact with its catalytic γ subunit to control activity. The γ subunit contains 386 amino acids having a mass of 44.7 kDa and is composed of an N-terminal catalytic domain of ca. 286 residues and a basic C-terminal domain of ca. 100 residues that will be referred to as its regulatory domain, or γCRD. A variety of chemical crosslinkers, zero-length or short when possible, were screened to identify those that could form γ–δ, γ–β, and γ–α dimeric conjugates when reacted with the hexadecameric (α β γ δ)4 PhK complex. These dimers were subsequently fractionated and analyzed by MS, following tryptic digestion. A γ–δ dimer was formed by the zero-length crosslinker N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline and was linked through an isoamide bond between K325 of the γCRD and D93 of the 16.7 kDa δ subunit, an intrinsic molecule of calmodulin. A γ–α dimer was selectively formed in large amounts by the very short crosslinker formaldehyde. Two-hybrid screening had initially indicated that the regions of interaction between these subunits occur somewhere within residues 343-386 of the γCRD and 1060-1237 of the 138.4 kDa α subunit, a region just C-terminal to a hyperphosphorylated sequence and that extends to the C-terminus of α. Our search engine identified peptides from digests of the γ–α complex that corroborate these regions of interaction; however, the data have yet to be fully analyzed to determine the residues crosslinked. A longer and more complex crosslinker, N-[γ-maleimidobutyryloxy]succinimide ester (GMBS), was required to form the γ–β conjugate. GMBS, which appeared to act as an affinity crosslinker, formed an unusual linkage between K303 of the γCRD and R18 of the 125.2 kDa β subunit. This residue near the N-terminus of β lies between its phosphorylatable seryl residues 11 and 26. Inasmuch as these residues are subject to intramolecular autophosphorylation by γ, the crosslinking of βR18 to γK303 also places the γCRD near the catalytic site of γ. The sum of our crosslinking results indicates that regulatory regions of the δ, α and β subunits of PhK directly interact with, or at the very least in the case of β lie near, the C-terminal regulatory domain of the catalytic γ subunit. The juxtaposition of these regions provides a physical rationale for the activation of PhK by Ca2+ (through the δ subunit) and phosphorylation (through the α and β subunits).

Supported by NIH grant DK32953 (G.M.C.).

7.1

Sperm Chromatin Proteomics Identifies Evolutionarily Conserved Fertility Factors.

Diana Chu1, H. Liu2 3, J. Yates2, and B. Meyer4 5
Department of Biology, San Francisco State University, San Francisco, CA, USA1 ; Dept. of Cell Biology, Scripps Research Institute, La Jolla, CA 2 ; Agilent Technologies, Inc., Wilmington, DE, USA 3 ; Howard Hughes Medical Institute, Berkeley, CA4 ; Department of Molecular and Cell Biology, University of California, Berkeley, CA, USA5

Human male infertility and in vitro fertilization failure can arise from defects in sperm chromatin composition and structure. To understand the underlying molecular causes of male infertility, we combined proteomics, cytology, and functional analysis in C. elegans to identify conserved spermatogenesis-enriched chromatin proteins important for fertility. Our strategy involved purification of chromatin from C. elegans spermatogenic or oogenic germ cells, identification of chromatin proteins through Multidimensional Protein Identification Technology, subtraction of common proteins, and prioritization of spermatogenesis-enriched proteins for functional analysis based on abundance. This approach reduced 1099 spermatogenesis-enriched chromatin proteins, the most extensive catalog of such proteins available, to 132 proteins for analysis. Functional analysis in C. elegans revealed conserved spermatogenesis-specific proteins are vital for DNA compaction, chromosome segregation, and fertility. Proteins with diverse functions in other cell types showed enrichment on sperm chromatin, revealing unexpected roles for general factors in spermatogenesis. Of 25 candidates with genetically disrupted mouse homologs, 11 cause infertility in male mice, validating our strategy to identify conserved fertility factors. Our list provides significant opportunity to identify causes of male infertility and targets for male contraceptives.

7.2

Walking in the Proteomic Footprints of Viral Infection

Ileana M. Cristea1, M.P. Rout2, and B.T. Chait1
Laboratory of Mass Spectrometry and Gaseous Ion Chemistry1, Laboratory of Cellular and Structural Biology, Rockefeller University, New York, NY, USA2

Protein interactions, of stable or transient nature, underline the majority of cellular processes. Through such interactions, viruses succeed in sabotaging complex organisms and turning their cellular machineries to their own use. Although many virus-host interactions have been studied, our knowledge of the interactions between viral and host proteins remains quite limited. We have recently developed a genomic-proteomic approach to study the dynamic localizations and interactions of viral proteins during the course of viral infections. This presentation will report on technical aspects of our strategy and results from our studies of Sindbis virus and human cytomegalovirus (HCMV). Our findings led to hypotheses as to how viruses manipulate host cellular processes. These studies demonstrate the importance of proteomic approaches in helping us gain a better understanding of the molecular details of viral infections, and just as importantly, the biology of the cell.
Our approach has the following steps: (1) Creation of replication-competent mutant viruses, genomically tagged with green fluorescent protein, protein A, or FLAG. (2) Natural infection of mammalian cells followed by fluorescence microscopy to visualize tagged viral proteins at various stages of infection. (3) Immediate Freezing of the cells followed by lysis in their cryogenic state to help maintain protein complexes close to their original condition in the cell. (4) Rapid immunosolations (5-60 min) using magnetic beads coated with antibodies against the tag. (5) Identification of the isolated proteins by mass spectrometry. (6) Confirmation of the observed interactions by co-localization and reciprocal immunoisolation of the host proteins. (7) Further studies are conducted for the functional characterization of the interactions.
Using this approach, our studies of the Sindbis virus, an Alphavirus genus member that in humans causes arthritis, provided details of temporally-regulated viral-host interactions during the course of infection. One resulting hypotheses is that Sindbis utilizes G3BP, a putative nuclear transport factor, to capture host RNAs undergoing nuclear export, for subsequent sequestration in structures involved in translational shut off. Additional studies of the Alphavirus Ross River and Flavivirus Yellow Fever viruses indicated that G3BP may be generally important during Alphavirus infection.
We have extended our studies to other viral systems, including HCV, BVDV, and HIV. Most recently, we have initiated a comprehensive study of the HCMV interactome. HCMV establishes a latent infection in 80% of adults by the age of 40. Congenital infection with HCMV is the leading viral cause of birth defects in U.S.A. To study the HCMV interacting network, we have constructed a library of 155 tagged HCMV viruses. Results from our studies of the early infection environment will be presented. The functional significance of the observed interactions will be discussed, including our finding that HCMV ensures cell growth and ribosome biogenesis by blocking normal cellular responses to stress.

7.3

Exploiting viruses to identify critical interactions and therapeutic targets in the cellular networks that regulate growth and survival

C. Soria1, A. Miller2, K. Espantman1, J. Smyth2, R. Chalkey2, A. Burlingame2, Clodagh O'Shea1
Salk Institute for Biological Studies, MCBL, La Jolla, CA, USA1 ; Department of Pharmaceutical Chemistry, University of California, San Francisco, CA, USA2

Protein-protein interactions provide a framework for understanding biology and disease at an integrated and mechanistic level. A major challenge is to define the critical cellular hubs and interactions that are subverted to elicit growth deregulation, a phenotype that is at the heart of many diseases, including cancer.
Unfortunately, human tumors acquire a myriad of molecular lesions, which makes it difficult to define the critical therapeutic targets. In contrast, the small DNA tumor virus, adenovirus, encodes 16 early growth deregulatory proteins that have been selected over millions of years to target the critical cellular hubs that control human cell growth and survival. It is the very same cellular networks that are disrupted by mutations in cancer. As such, there is a striking overlap between the key cellular targets of tumor mutations and DNA viral proteins. Indeed, many of the critical growth regulatory targets, for example p53, E2F and PI-3 kinase, were initially identified in studies with DNA viral proteins. Proteomics has progressed a long way in the interim. Tandem Affinity Purification (TAP) combined with Mass Spectrometry is a powerful approach to defined the cellular targets of early adenoviral growth deregulatory proteins, which have yet to be systematically identified.
We are exploiting adenovirus as a powerful paradigm for an integrative systems understanding of a natural, dynamic and pleiotropic growth deregulation program in normal human cells. In collaboration with Dr. Alma Burlingame°Øs lab at UCSF, we are comprehensively identifying the cellular interactions targeted by early adenoviral proteins both in uninfected and infected cells. This is providing new insights into key growth regulatory pathways and how they could be modulated for cancer therapy, such as, PI-3 Kinase/mTOR and p53.
In adenovirus infection, E1B-55K mimics mdm2 in binding and degrading p53. On this basis an adenovirus mutant lacking E1B-55K (ONYX-015) was proposed and tested as a p53-selective cancer therapy. However, we found that p53 plays a minor role in determining the tumor selectivity of this virus. This is because, although the absence of E1B-55K results in very high levels of p53 in infected primary cells, p53 activity is nonetheless suppressed even upon irradiation. Recently, to determine the mechanism whereby p53 is inactivated, we screened for p53 activation in primary cells infected with adenoviruses that have compound mutations in E1B-55K and other early viral genes. These studies have revealed that there are additional early viral genes which dominantly suppress p53 activity in the absence of p53 degradation. Using TAP tags and mass spectrometry, we have identified novel cellular proteins that interact differentially with active versus inactive p53 in E1B-55K mutant virus infected cells and upon adriamycin treatment. The results of these studies have important implications for understanding p53 regulation and how it could be modulated for tumor therapy.

7.4

High Throughput Characterization of Amplified Nucleic Acids by ESI-TOF Mass Spectrometry: Applications in Pathogen Detection and Characterization

Steven A. Hofstadler
Ibis Biosciences, Carlsbad, CA, USA

High throughput ESI-TOF analysis of amplicons represents a novel and universal strategy for the detection and characterization of microorganisms associated with emerging infectious diseases. The process uses mass spectrometry, signal processing, and base composition analysis of PCR amplification products from biologically conserved regions of microbial genomes to simultaneously identify the organisms present in a sample without the need for culture. The sample can be derived from environmental samples, clinical specimens, or other sources. Core to this approach are broad range PCR primers that target broadly conserved regions of microbial genomes that flank variable regions. Using high-performance electrospray ionization mass spectrometry (ESI-MS), the base composition (i.e., the number of As, Gs, Cs, and Ts) of each amplicon is unambiguously determined.
Bacterial Example: We have examined cultures and direct throat swabs obtained from individuals suspected to be suffering from Group A Streptococcus (GAS) infections. Samples were first analyzed using a panel of survey primers that readily identified the infectious agent as Streptococcus pyogenes, clearly distinguishable from all other organisms, including other streptococci and staphylococci. Subsequent analysis with Streptococcus-specific primers rapidly yielded emm-types for each sample which were later corroborated with conventional MLST analyses. This study demonstrated that this approach can be used to detect and identify infectious agents directly from a throat swabs. In the present configuration, hundreds of samples can be analyzed within 12 hours allowing near real-time evaluation of patient samples and will make possible more rapid and appropriate treatment of patients in an ongoing epidemic. The use of "drill down" primers allows closely related strain variants to be distinguished and accurately identified. This is of particular importance when tracking the spread of particularly virulent strains of disease-causing organisms.
Viral Example: The base compositions of amplicons from six influenza genes were used to provide sub-species identification and infer H and N subtypes. The method detected and correctly identified 92 mammalian and avian influenza isolates, representing 30 different H and N types, avian H5N1 isolates. Further, PCR/ESI-MS enabled correct identification and mapping of 656 human clinical respiratory specimens collected over a seven-year period (1999-2006) to previously established clades. Thus, PCR followed by rapid ESI-MS analysis can be used to simultaneously identify all species of influenza viruses with clade-level resolution, identify mixed viral populations, and monitor viral evolution. This method promises to become an integral component of high-throughput influenza surveillance.
This strategy distinguishes this approach from other detection/identification strategies in that it requires little or no prior knowledge about an organism in order to identify it in a sample. The approach requires that high-performance mass measurements be made on PCR products in the 80 – 140 bp size range in a high-throughput, robust modality. The base compositions from multiple primer pairs are used to "triangulate" the identity of the organisms present in the sample. Use of species-specific primers allows rapid strain-typing of the organism. The concept is equally applicable to bacteria and viruses and has recently been applied to detection of variations in human mtDNA associated with mitochondrial diseases.

8.1

Exploration of the Human Muscle Proteome in Diabetes

Susan Weintraub1, C.A. Carroll1 2, S. Kamath², A. Monroy2, A.O. Chavez-Velazquez2, R.A. DeFronzo2 and F. Folli2
Department of Biochemistry1 and Diabetes Division2, Department of Medicine, University of Texas Health Science Center, San Antonio, TX, USA

Type 2 diabetes has increased dramatically world-wide over the last few decades, potentially becoming a global pandemic with more than 300 million people projected to be afflicted by 2025. The primary cellular cause of this disease is still unclear. Insulin resistance plays an early role in the pathogenesis of the disease along with a decrease in insulin production by pancreatic beta cells; the result, ultimately, is hyperglycemia associated with diabetes. Skeletal muscle and liver are the primary tissues in the body that respond to insulin and, as such, play a major role in maintaining normal homeostatic glucose levels. Our research efforts have been aimed at characterizing the defects in enzymes and associated metabolic pathways that lead to insulin resistance in these tissues. Our studies have encompassed several different approaches in order to gain insight into this highly prevalent and insidious disease.
Phosphorylation in response to insulin. Insulin signaling in muscle is mediated by a complex, highly integrated network of serine, threonine and tyrosine phosphorylation/dephosphorylation events. With this in mind, we used a 2-dimensional gel electrophoresis (2-DE) approach to assess the phosphorylation status of human cytosolic skeletal muscle proteins in the basal state and after insulin stimulation. Global phosphorylation was evaluated using a phosphoprotein-specific gel stain, Pro-Q Diamond (Invitrogen); this stain is able to detect serine, threonine and tyrosine phosphorylation at the femtomole level in proteins. Percutaneous muscle biopsy samples were obtained from the vastus lateralis of normal, healthy volunteers before and after 4 hours of insulin infusion; 20% glucose was simultaneously administered to maintain normoglycemia. Cytosolic protein fractions were subjected to 2-DE. After image acquisition of phosphoproteins stained with Pro-Q Diamond, gel images of total protein were obtained after staining with SYPRO Ruby (Invitrogen). From image analysis by PDQuest (Bio-Rad), we found that of the 75 spots stained by Pro-Q Diamond, phosphorylation was significantly increased in 22 spots and decreased in 43 after insulin infusion. Among the proteins exhibiting changes in phosphorylation were aldose reductase, creatine kinase, GAPDH and triosephosphate isomerase. There were approximately 300 spots visualized by SYPRO Ruby; of these, the intensity increased in 37 spots after insulin treatment.
Type 2 diabetes and the muscle mitochondrial proteome. Two-dimensional gel electrophoresis studies were conducted on mitochondrial protein extracts obtained from normal (control) and type-2 diabetes subjects. For some experiments, samples from pairs of normal and type-2 diabetes subjects were directly compared, while for others, extracts from several individuals were combined before 2-DE. Of the 388 protein spots detected, approximately 120 have been identified so far by mass spectrometry. From evaluation of the results from all sample sets, seven proteins exhibited consistent, statistically significant changes in quantity for all groups that were compared. Of particular interest is mitofilin (mitochondrial inner membrane protein); this protein was detected as a distinct "charge train" on the 2-D gels and was found to decrease substantially in all diabetic subjects compared to control. Since mitofilin has been shown to be involved in mitochondrial morphology, and the shape of mitochondria has been shown to change in diabetes, observation of decreases in the level of mitofilin isoforms could have important physiological implications in the disease process.
Islet cells in culture. Islet cells have frequently been cultured for use in a wide range of metabolic studies. However, there have been no studies reported on changes that take place in islet cells at the protein level during culture. For these experiments, islets were isolated from human pancreas obtained at autopsy. Proteins were isolated before culture and at 2 and 4 days in culture. From 2-DE and staining with SYPRO Ruby, approximately 320 spots were detected; proteins in 125 of these spots have been identified to date by mass spectrometry. From PDQuest analysis we found that 27 spots increased significantly in intensity while 18 were decreased. Included in the proteins that exhibited changes in quantity were a number of participants in the apoptotic pathway. The most striking change was seen in cofilin-1, a protein that binds cooperatively to two strands of F-actin. In our study, cofilin was found to increase approximately 50-fold at 2 and 4 days of islet cell culture. Our observation of significant changes in islet cell proteins during culture indicates the need for prevention of apoptosis in islet cells being cultured for transplantation.

8.2

Protein Biomarker Identification in the Cerebrospinal Fluid in Patients with Central Nervous System Lymphoma

S. Roy1, S.A. Josephsons2, J. Fridlyand3, J. Karch4, C. Kadoch4, J. Karrim4, L. Damon4, P. Treseler5, S. Kunwar6, M.A. Shuman4, T. Jones¹, C.H. Becker¹, H. Schulman¹, and James L. Rubenstein4
PPD Biomarkers, Menlo Park, CA, USA1 ; Departments of Neurology 2 , Epidemiology & Biostatistics3 , Medicine4, Pathology5, and Neurological Surgery6, University of California, San Francisco, CA, USA

Purpose: Elucidation of the cerebrospinal fluid (CSF) proteome may yield insights into the pathogenesis of central nervous system (CNS) disease. We tested the hypothesis that individual CSF proteins distinguish CNS lymphoma from benign focal brain lesions.
Methods: We used a liquid chromatography-mass spectrometry-based method to differentially quantify and identify several hundred CSF proteins in CNS lymphoma and control patients. We used ELISA to confirm results for one of these markers in an additional validation set of 101 cases.
Results: Approximately 80 CSF proteins were identified and found to be present at significantly different concentrations, both higher and lower, in training and test studies which were highly concordant. To further validate these observations, we defined in detail the expression of one of these candidate biomarkers, a serine protease inhibitor, AT. AT RNA transcripts were identified within CNS lymphomas and AT protein was localized selectively to tumor neovasculature. Determination of AT concentration by ELISA was significantly more accurate (>75% sensitivity; >98% specificity) than cytology in the identification of cancer. Measurement of CSF AT levels was found to potentially enhance the ability to diagnose and predict outcome.
Conclusion: Our findings demonstrate for the first time that proteomic analysis of the CSF yields individual biomarkers with greater sensitivity than CSF cytology in the identification of cancer. We propose that the discovery of CSF protein biomarkers will facilitate early and noninvasive diagnosis in patients with lesions not amenable to brain biopsy as well as provide improved surrogates of prognosis and treatment response.

8.3

Progress Toward a Biomarker Discovery-to-Verification Pipeline in Clinical Proteomics

Steven A. Carr
Proteomics Platform, Broad Institute of MIT and Harvard, Boston, MA, USA

Better biomarkers are urgently needed to improve diagnosis, guide molecularly targeted therapy, and monitor activity and therapeutic response across a wide spectrum of disease. Proteomics methods based on mass spectrometry hold special promise for the discovery of novel biomarkers that might form the foundation for new clinical blood tests, but to date their contribution to the diagnostic armamentarium has been disappointing. This is due in part to the lack of a coherent pipeline connecting marker discovery with well established methods for validation. Advances in methods and technology now enable construction of a comprehensive biomarker pipeline from six essential process components: candidate discovery, qualification, verification, research assay optimization, biomarker validation, and commercialization. Better understanding of the overall process of biomarker discovery and validation and of the challenges and strategies inherent in each phase should improve experimental study design, in turn increasing the efficiency of biomarker development and facilitating the delivery and deployment of novel clinical tests. In this talk I will illustrate discovery-through-quantitative verification steps with examples from our ongoing studies in breast cancer and cardiovascular disease.

8.4

The Protein Microscope: An Application of Cell Map Proteomics

John J.M. Bergeron
Department of Anatomy and Cell Biology, McGill University, Montreal, Canada

The application of hierarchical clustering to a quantitative representation of comprehensive protein abundance in isolated organelles provides a ready method to sort contaminants from resident proteins and gain insight into organelle identity and function. Organelle based proteomics provides hundreds of proteins of unknown function. An application of the protein microscope is to assign locations and mechanistic functions to these proteins.


National Institute of General Medical Sciences Adelson Medical Research Foundation