1 The application of proteomic tools for a systems approach to elucidating signaling pathways Jessica Rouleau Abstract— The post genome era is seeing a shift towards the study of cellular proteomes. It is becoming increasingly clear that proteins are the dynamic systems of the cell which control the net responses. Therefore, a wide range of novel techniques are being used for the large scale study of proteins in a physiological context. These techniques include mass spectrometry, two hybrid arrays, tandem affinity purification and RNA interference. Application of these techniques to study cellular processes will provide a non-intuitive understanding of complex signaling pathways. The union of systems biology and proteomics holds vast potential for understanding multiple pathways of a cell in a manner that is spatially and temporally regulated. Index Terms—Mass spectrometry, Proteomics RNA interference, Signal transduction pathway, Systems biology, Tandem affinity purification, Two hybrid arrays. I. INTRODUCTION R ESEARCH in the post genome era is shifting towards protein based studies. It is now understood that the study of gene expression is not sufficient to describe dynamic processes occurring in cells [1]. The human genome contains approximately 30000 to 40000 genes but the resulting protein products are estimated to be composed of over a million distinct molecular species [1]. Therefore, it has been recognized that studying gene expression and regulation will not provide the complete picture of cellular processes. The large scale analysis of proteins has formed a discipline of research referred to as proteomics. Proteomics includes studying the abundance, structure, formation of multi-unit complexes, modifications and isoforms of the proteome [2]. The proteome is a dynamic entity that corresponds to all the proteins in a given cell [3]. To elucidate the many aspects of proteomics research, a number of different high-throughput proteomic techniques have been developed. These include mass spectrometry (MS), two hybrid arrays, tandem affinity purification techniques (TAP) and RNA interference (RNAi) approaches. Each of these techniques has great potential to characterize aspects of the proteome but as these techniques are still in their infancy, there are many problems that need to be overcome. Many systems biology approaches to model signaling pathways and intracellular mechanisms have used large sets of gene expression data to identify gene interaction networks [4]. The shortcomings of genomic approaches is they do not give a clear indication of the abundance of proteins translated, the lifetime of these proteins, the state of these proteins and their interactions in the cell [1]. Therefore, proteomic data will provide a systems approach, whereby large data sets will explain the molecular interactions occurring in the system. Systems biology based modeling will use this data and put it into a context that allows non-intuitive predictions of cellular behaviour [5]. These models will explain the states of cellular systems and not just the connectivity of components. To achieve a systematic understanding of signaling pathways, an integration of extracellular stimuli, protein dynamics and gene regulation need to be incorporated into models. The improvement and increased use of proteomic techniques will greatly aid in mapping out and integrating many signal transduction pathways to develop an overall systems understanding of cellular processes. This will lead to a greater understanding of many disease states, the normal physiology of the cell, targets for drug discovery and more specific requirements for tissue engineered scaffolds. II. PROTEOMIC TECHNIQUES A. Mass Spectrometry With the completion of genome sequencing, mass spectrometry is a technique that has become central in proteomic research. Traditionally, mass spectrometry based proteomics has been used for peptide sequencing, the study of protein-protein interactions and post-translational modifications [6]. Recently, novel ways to study proteins have been developed using this technology [7, 8]. Mass spectrometry requires protein samples to be digested into peptide fragments where they can then be separated using high performance liquid chromatography (HPLC) or ion exchange columns [9]. These peptides are then ionized using one of two techniques, electrospray ionization (ESI) or matrixassisted laser desorption/ionization (MALDI) [9]. ESI ionizes peptides out of solution and MALDI ionizes the samples from a dry crystalline matrix. Generally ESI techniques are used to analyze complex mixtures of proteins while MALDI is better for analyzing simple mixtures [6]. Once inside the mass spectrometer, the ionic fragments are in a vacuum where they are manipulated by electric fields. Mass analyzers determine the mass to charge ratio of the peptides. Three main types of mass analyzers are used to identify peptide fragments: Fourier transform mass spectrometers, time of flight mass spectrometers and ion traps [6]. Ion traps tend to be the most widely used to date but have relatively low mass accuracy while Fourier transform mass analyzers have high sensitivity and resolution but are expensive and complex to run [6]. After obtaining the mass to charge ratio the primary structure of the peptide can be determined. Once all the signatures of the 2 peptides are collected, databases are then used to match these sequences to known sequences. Since the samples generally do not contain one type of protein, these searches have to be coupled to statistical techniques to determine the probability of each protein in the sample being present. Researchers may use filtering criteria to make a more accurate analysis of the sample composition [6]. Mass spectrometry can provide other types of data besides peptide identification. Post-translational modifications can be identified through analysis of the spectra. Proteins that are modified have different spectral features. For example, phosphorylated proteins have a prominent peak added due to the loss of a phosphoryl group [9]. Phosphorylated proteins can also be selected by affinity separation through the use of anti-phosphotyrosine, anti-phosphoserine and antiphosphothreonine antibodies. The purified extract can then be analyzed in the mass spectrometer [7]. There have been several methods created to use isotope labels to identify two protein populations in different states or at different time points [7, 9]. These methods allow for a more dynamic analysis of protein populations. One other technique that may gain popularity is a technique termed AQUA and it involves producing known concentrations of specific peptides to be used as internal standards for absolute quantification of proteins [8]. Large scale proteomic studies using mass spectrometry show great potential for characterizing different aspects of the proteome but there are still many obstacles to overcome. The databases used to match peptide fragment signatures need to become more extensive, containing all potential protein sequences, the variable splicing domains, and the post translational modifications. Statistical analysis and the use of filters to identify proteins from complex protein mixtures needs to be improved and standardized. New mass spectrometry techniques must continue to be developed so this instrument can be used to its greatest potential. B. Yeast two hybrid arrays The two hybrid method is an experimental method that is used to study protein-protein interactions. It can be used to generate large interaction maps and identify novel proteins that are associated by the interaction [10]. The function of these new proteins can then be hypothesized because of the known function of the interacting protein. Yeast two hybrid studies have been used to functionally map out signal transduction pathways [10]. This method is also valuable to studying specific protein interactions between two proteins of a larger protein complex. This method was devised by exploiting the knowledge about transcription factors, which usually have separate domains for DNA-binding (DBD) and transcriptional activation (AD) [11]. To study the interaction between two proteins using the two hybrid method, two yeast strains need to be created. One strain will express the protein of interest as a fusion to a DBD and the other strain will have the other protein of interest fused to an AD domain. Yeast strains are created by cloning PCR products for the specific protein of interest into a plasmid vector [10]. When these two strains are mated the fusion proteins are present together in the progeny. If there is an interaction between the two proteins of interest, the DBD and AD will come together to activate transcription of a reporter gene and this process is shown in figure 1 [11]. There are many reporter genes that can be used such as luciferase, betagalactosidase or green fluorescent protein. Different assays can then be used to detect the presence of these products. If no interaction occurs then the transcription of the reporter genes will not occur. This technique can be used in a high throughput fashion through the creation of many yeast strains from cDNA libraries of the organism of interest [10]. Therefore arrays of bait and prey proteins can be studied in a systematic nature. Data from these experiments can produce large interaction maps of proteins [11]. Strain A Strain B A B DBD Reporter X AD Reporter Progeny AB Reporter Figure 1 [11]: Schematic of yeast two hybrid method. The advantages of this experimental approach are that it has good resolution, it is independent of endogenous protein expression, and transient and unstable interactions can be detected [3]. Disadvantages to using this method are that interactions take place in the nucleus so that many of the proteins tested are not in the same compartment as in their native environment. Therefore, these interactions are not related to their physiological setting. There is relatively high rate of false positives using the yeast two hybrid method [3]. To validate the results from two hybrid studies, repetition of the experiments and analysis using other techniques can be performed [11]. C. Tandem affinity purification (TAP) There have been many different affinity purification techniques that have been developed to study protein complexes. These include immunoprecipitation, epitope tagging, and tandem affinity purification. Tandem affinity purification is a technique that can be used in a high throughput fashion and can identify a large number of protein 3 complexes within a cellular environment [1]. The other techniques have limitations which make them less likely to be used in a large scale proteomic study. As more proteomic data is generated, it is becoming clear that cellular proteins organize themselves through a dynamic arrangement of protein complexes [1]. Protein complexes can vary from a few proteins in size to large complexes with over 80 components [12]. Analysis of complexes can facilitate a functional understanding of many unknown proteins. TAP isolates protein complexes through the use of a tag which is attached to a protein of interest in the cell [12]. The components of this tag are shown in figure 2. Proteins that complex with this bait protein will then be purified using the tag. Tandem affinity purification begins with the insertion of the tag into the host DNA [12]. The tag can be inserted at the end of the gene representing the carboxy terminus end or the amino terminus end of the protein [1]. The TAP tag contains specific elements, each corresponding to a specific function. The first component of the tag contains the sequence for a protein that will be used to bind to specific antibodies in a separation column. This protein will selectively bind to the column while all other proteins will pass through. The tag contains a TEV (tobacco etch virus) site [12]. This site is highly susceptible to attack by the TEV protease. Cleavage by this site specific protease releases the rest of the tag with the protein complex. The remainder of the tag contains a calmodulin binding peptide (CBP) and the complex of interest [12]. The CBP site is used for a second purification step, whereby the eluant from the first purification step will be passed through a second affinity column of calmodulin coated beads [1]. Elution of the protein complex is achieved using chelating agents. This purified protein complex can then be analyzed by mass spectrometry to identify the proteins in this complex [12]. This purification technique has several advantages over some of the other techniques [1]. The two step purification tends to greatly reduce background noise caused by unspecific binding. This reduction also reduces the complexity of the mass spectrometry analysis. The purification steps in this technique do not require strong detergents or high salt concentrations, which can result in the loss of proteins within the complex [1]. The separation steps used in TAP are kept near physiological conditions and this gives the technique greater physiological relevance. TAP has been shown to be highly reproducible, using different bait proteins. Caveats of this technique could be tag interference in protein function or complex formation [1]. Repetition of experiments and the use of different entry points to verify complexes will provide stronger experimental evidence. D. RNA interference In the recent past, it was thought that transcription of DNA to RNA was the main regulatory mechanism in the cell and that RNA played a relatively simple role in the cell. This role is translation, where genetic information is used to produce proteins. The identification of double stranded RNA (dsRNA) PCR of TAP tag and bait protein Spacer CBP TEV site Protein Homologous Recombination Gene Bait Protein Spacer [Gene Targeting] Gene CBP [ TAP cassette] TEV site [Chromosome] Protein Collection of Lysates and Affinity Purification Figure 2 [2]: Process of creating a TAP tag. molecules in plants and Caenorhabditis elegans has altered these views [13]. RNA interference has now been identified as a mechanism in many cell types that cause inhibition of protein production at a post transcriptional level [13]. Large dsRNA molecules are processed by an enzymatic protein called Dicer to yield 19-23 base dsRNA oligonucleotides [14]. These oligonucleotides then associate with specific mRNA molecules that match their sequence. This association promotes the formation of the RNA induced silencing complex (RISC) which has RNase activity. This complex degrades the RNA and prevents transcription [14]. The understanding of these mechanisms has lead researchers to use the cell’s machinery as a research tool [13]. Using the cDNA sequence of the gene to be silenced, short interfering RNA (siRNA) strands can be synthesized and then inserted into the cell in a number of ways which include soaking the cells in siRNA and injection [13]. Plasmids and viral vectors have also been constructed to produce specific siRNA within a cell population through transfection. There are many methods that are currently being optimized for RNAi based studies [13]. These studies let researchers block the action of a specific protein within the cell and look at the resulting phenotypes. Large scale arrays of RNAi can now be performed and this will help to identify functions of different proteins, their activity in signaling pathways and targets for drug therapies [13]. The other method of gene silencing is to perform gene knockouts. Compared with RNAi these techniques are more complex and time consuming. Gene knockouts often produce embryonic lethalities, so observation of the effects of the knockout is prevented. RNAi technology can produce knockouts during different stages of development and target specific tissues. III. SYSTEMS BIOLOGY APPROACH TO PROTEOMIC MODELING OF SIGNAL TRANSDUCTION PATHWAYS As large scale proteomic techniques evolve and are 4 optimized, the large protein networks that make up signal transduction pathways will be elucidated. Different techniques will need to be combined to determine the connectivity between different proteins, their dynamic interactions, the states of these proteins and lifetime in the cell. Proteomics and genomics will need to work in synchrony to determine how signaling pathways affect transcription of signaling pathway components and other processes going on in the cell. For example, it is thought that the transforming growth factor pathway regulates cell proliferation and tissue remodeling. Therefore, proteomic techniques will be used to map out the components of the pathway. These components have a differential effect on gene expression of not only proteins in the TGF- signaling pathway but proteins likely involved in cell cycle regulation and extracellular matrix synthesis [10, 15]. A combination of proteomic and genomic techniques will elucidate the dynamic activity of proteins in a temporal and spatial fashion and describe the resulting changes in gene expression that these protein pathways influence. Current proteomic techniques are able to get a basic understanding of the large compendium of proteins within a cell. Mass spectrometry approaches allow for identification of proteins in complex samples. This technique can also identify phosphorylation states and abundance of these proteins in a snapshot of time. Interactions of these proteins can be identified using the yeast two hybrid techniques. These techniques are able to build upon mass spectrometry data to begin construction of a protein interaction network. Another layer of complexity is the determination of larger complexes within the cell through affinity purification. This will refine the protein interaction network of signaling pathways. Throughout the process RNAi techniques can be used to look at the importance of specific proteins in a pathway and how each protein affects downstream signaling and interaction of protein complexes. These techniques will be able to identify redundant pathways or processes as well as points of cross-talk between different signaling pathways. V. CONCLUSIONS Overall, these techniques will aid in understanding the cell as a system. Large scale studies can be fit into models that systems biology attempts to create. Models will be developed in an iterative fashion to identify areas that need to be altered and expanded. It is the goal of systems biology to combine signaling pathways, cellular metabolism and nuclear processes of a system to develop a model that can be used for patient based treatments and therapies. Proteomic techniques will aid in elucidating signal transduction networks. In the future, models of these networks will be integrated with other models to develop a mechanism whereby cellular physiology can be understood and predicted at the systems level. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] IV. FUTURE CHALLENGES [12] Proteomic techniques still face many challenges. These techniques need to be used in way that enables a more dynamic look at signaling pathways instead of just snapshots. The reduction in the number of false positives and false negatives needs to occur. More stringent procedures need to be implemented in these studies. Proteomics needs to address the low abundance of many signaling molecules and the high sensitivity needed to be able to measure and monitor these proteins. The large range of protein levels is also a major barrier to analysis, there are proteins that only have a couple of copies per cell but others have millions of the same protein within a cell. A technique needs to be sensitive enough to detect the low abundance of proteins but have large enough range to be able to monitor higher levels of other proteins. The similar challenge is seen when looking at time scales within a cell. Some processes occur in seconds while others take hours and these large ranges need to be addressed. [13] [14] [15] A. Bauer and B. Kuster, “Affinity purification-mass spectrometry,” Eur. J. Biochem, 270: 570-578, 2003. M. Tyers and M. Mann, “From genomics to proteomics” Nature, 422: 193-197, 13 March 2003. C. von Mering, et al. “Comparative assessment of large-scale data sets of protein-protein interactions”. Nature, 417: 399-403, 23 May 2002. P. Brazhnik, A. de la Fuente, P. Mendes, “Gene networks: how to put the function in genomics”. Trends in Biotech. 20: 467-472, November 2002. H. Kitano, “Looking beyond the details: a rise in system-oriented approaches in genetics and molecular biology”. Curr. Genetics. 41: 110, 2002. R. Abersold and M. Mann. “Mass spectrometry-based proteomics” Nature, 422: 198-207, 13 March 2003. B. Blagoev, S.E. Ong, I. Kratchmarova, M. Mann. “Temporal analysis of phosphotyrosine-dependent signaling networks by quantitative proteomics”. Nature Biotech. 22: 1139-1145, September 2004. S.A. Gerber et al. “Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS”. Proc. Nat. Acad. Sci. 100: 6040-45, 2003. H. Steen and M. Mann, “The ABC’s (and XYZ’s) of peptide sequencing”. Nature Reviews: Molecular Cell Biology. 5: 699-710, Septermber 2004. F. Colland, X. Jacq, V. Trouplin, C. Mougin, C. Groizeleau, A. Hamburger, A. Meil, J. Wojcik, P. Legrain, J.M. Gauthier. “Functional Proteomics Mapping of a Human Signaling Pathway”. Genome Research. 14: 1324-1332, 2004. P. Utez. “Two-hybrid arrays” Current Opinions Chem. Bio. 6:57-62, 2001. A-C. Gavin et al. “Functional organization of the yeast proteome by systematic analysis of protein complexes”. Nature. 415: 141-147, 2002. O. Milhavet, D.S. Gary, M.P. Mattson, “RNA interference in biology and medicine”. Pharma. Reviews. 55: 629-648, 2003. C.C. Mello and D. Conte, “Revealing the world of RNA interference”. Nature. 431: 338- 342, 16 September 2004. M. Schiller, D. Javelaud, A. Mauviel, “TGF--induced SMAD signaling and gene regulation: consequences for extracellular matrix remodeling and wound healing”. J. of Derm. Sci. 35: 83-92, 2004.