12.1. The Changing Proteome
Much work in biochemistry and physiology already has shown us that biochemical pathways are constantly in flux. The use of DNA microarrays has demonstrated that gene-expression patterns in cells are also changing constantly. Indeed, one can use either old or new technologies to observe regular changes in the status of many enzymes during the daily life cycle of an organism or in the cycle of a cell. All this suggests that the proteomes of cells are constantly changing as well. In addition to these changes that are essential to life, other changes are induced by environmental stimuli, chemicals, drugs, and growth and disease processes. Many of these latter changes interest those trying to understand complicated pathologies (e.g., cancer) or trying to identify novel targets for therapeutic drugs. Perhaps the ultimate challenge of proteomics is to measure the status of all cellular proteins as they change with time. Unfortunately, the technology is not yet there. Nevertheless, the problem of comparing proteomes between two states of a cell or organism is relevant and important.
The most fundamental task in protein-expression profiling is to measure the expression of a set of proteins in two samples and then compare them. To do this, we simply need a method that detects and identifies the same proteins in the two samples and provides a basis for comparing their levels. This can be done in two different ways, as is outlined below. However, these methods may indicate
From: Introduction to Proteomics: Tools for the New Biology By: D. C. Liebler © Humana Press, Inc., Totowa, NJ
only the level of the polypeptide gene product per se, but not how proteins are changed by modification. This requires not only use of the techniques outlined in this chapter, but also integration of the approaches outlined in subsequent chapters.
Perhaps the most widely used approach to comparative proteomics is to subject two samples to 2D-SDS-PAGE and compare the spot patterns. Two-dimensional SDS-PAGE is particularly well-suited to comparative proteome analysis because it effectively resolves many proteins. With recent improvements in 2D gel technology (see Chapter 4), the task of running reproducible 2D gels has been made easier. Even before the introduction of MS-based protein identification, this approach provided a useful means of comparing proteomes. However, identification of the proteins was cumbersome and difficult. Application of peptide mass fingerprinting and LC-MS-MS analyses now makes it possible to identify essentially any protein one can detect by staining the gel. Thus, the critical task in comparative proteomics with 2D gels is identifying the features that differ between gels.
A great deal of work has been done to develop software tools to analyze the patterns of protein spots on 2D gels. In addition, extensive databases to archive this information have been developed. Among the most widely used programs for 2D gel-image analysis is Melanie™, which was developed at the Swiss Institute for Bioinformat-ics. Melanie™ works with images of stained 2D gels. These images can be acquired by use of a document scanner (to produce .gif or .tif files) or preferably by the use of a CCD camera. The program does several things. First, the gel is evaluated for "features," which simply refer to any significant deviation from the background. The features correspond to the protein spots on the gel (see Fig. 1). The features can be characterized by optical density (OD), by size, and by volume, which integrates OD over the spot area. These characteristics comprise the basis for comparing features within a gel and between multiple gels.
Of course, for protein-expression profiling, we want to compare the 2D gels from two different samples for differences in the occurrence or intensity of features. The problem with this is that it is very hard to run multiple 2D gels with exact reproducibility. There usually are
slight variations in the location of spots for specific proteins. This makes it hard to compare spots on two gels if their locations are slightly different. To accommodate these differences, the software allows the user to identify "landmarks," which are proteins that occur in both (or all) of the gels to be compared. These features then are "paired" by the software to create a series of pairs by which the gels can be aligned or "matched." The matching process involves alignment of the two gels so that the landmarks have the same relation to each other in 2D space. In other words, the gel images are lined up pixel-wise so that all the landmark features match. This process can entail some transformations or spatial "warping" of the images to compensate for local geometric distortions in the gel.
Once the gels are matched, then comparison of the features may be done. These comparisons examine the OD volume differences between features on the gels and provide a graphical output that assigns numbers to the observed differences (see Fig. 2). The software also enables statistical analyses of these data to facilitate interpretation of significant differences. It is this operation that allows the user to identify those features or spots that differ between two or more samples. Gels may be visually "stacked" to enable comparison of images. Alternatively, virtual gels can be synthesized from the images collected from multiple gels to provide a master archive of composite proteomes in different states of an organism.
Identification of the proteins in these spots of interest then involves excision of the spots, in-gel digestion, and MS analysis. In addition to detecting differences between features on multiple gels, the software also allows the user to annotate the features and to link them to database files containing MS data, gene sequences and functional genomics data.
The ability to compare two gels and then identify differently expressed spots is the essence of protein-expression profiling with 2D gels. However, the development of this approach by a number of groups has led to the development of 2D-SDS-PAGE databases, which archive large numbers of annotated images from 2D gel analyses. These databases are an increasingly powerful resource for the comparison of data generated in different laboratories. A powerful feature of these databases and software programs such as Melanie is the ability to compare large numbers of gels to a single gel or to groups of gels and compile statistical summaries of patterns in protein spot variation.
Another unique software tool to compare 2D gel images over the Internet is the Flicker program, which was developed by Lem-kin and colleagues at the National Cancer Institute (http://www-lecb.ncifcrf.gov/flicker/). Flicker uses many of the same approaches to the evaluation and matching of gels described earlier. An important feature of Flicker is that it permits the user to compare gel images from different sources on a web browser. This makes possible not only the facile comparison of images from different databases, but also the comparison of one's own 2D gel images with images from different databases.
The use of 2D gels is a powerful approach to protein profiling and it is unique in providing a visual-image basis for proteome comparisons. However, there is one major drawback to this approach: staining of 2D gels only detects the more abundant proteins in a sample. There is approximately a million-fold range of protein expression in cells, whereas gel staining is limited by about a hundred-fold dynamic range. It is possible to enhance detection of low-abundance proteins by loading more protein for analysis, but abundant proteins eventually overwhelm many of the features on the gel. A related problem is that many proteins exist in multiply modified forms, which may display different isoelectric points and are thus separated on 2D gels. For less abundant proteins, spreading out into multiple spots can lower detectability by staining. Finally, identification of very weakly stained (and thus low-abundance) proteins by in-gel digestion and MS can be hampered by poor recovery of peptides from the digestion and the gel. As noted in Chapter 5, recovery of peptides from in-gel digestions is usually less than quantitative and is frequently less than 60%. These drawbacks to 2D gel protein profiling all stem from the limited dynamic range of 2D gel staining for protein detection. Although staining and visualization methods are continuing to evolve and improve, this problem may ultimately limit 2D gels to analysis of relatively abundant proteins. This is adequate for many circumstances, however, and 2D gel-based proteome profiling will continue to be a valuable, widely used technique.
12.3. Comparative Proteomics with LC-MS and Isotope Tags
The LC-MS approach to proteome comparisons is conceptually the opposite of the 2D gel approach. Whereas the 2D gel approach separates proteins and begins with an image comparison, the LC-MS approach separates peptides and ends with data mining to assess differences between samples. Here's the LC-MS approach in a nutshell. Two protein samples are treated with reagents to "tag" them. The tags are chemically identical, except that one contains heavy isotopes and the other contains light isotopes. The samples are digested and the peptides are analyzed by LC-MS-MS. Analysis of the MS-MS data (e.g., with Sequest) allows identification of the proteins present. Examination of the full-scan spectra corresponding to each MS-MS scan then allows measurement of the ratio of the light- and heavy-isotope tagged peptides. This ratio corresponds to the ratio of that protein in the two samples. This approach provides not an absolute quantitation of proteins, but rather a relative quantitation of the level of a particular protein in two samples. This approach was first applied to analyzing differences in proteomes by Gygi and Aebersold. We will examine their approach in the following paragraphs, and also consider some possible variants on this approach.
Perhaps the best place to start is with the isotope-labeled tags. The use of isotopic tagging is a variation of the technique known as "stable isotope dilution." To understand why they are used, let's recall that stable isotopes are forms of elements that vary in the number of neutrons in their nuclei, yet are not radioactive. For example, hydrogen has no neutrons, whereas its less abundant isotopomer, deuterium, has one. Thus, hydrogen (1H) has a mass of one and deuterium (2H) has a mass of two. Other frequently used stable isotopes are 13C (one amu heavier than 12C), 15N (one amu heavier than 14N), and 18O (two amu heavier than 16O). Compounds labeled with deuterium atoms in place of some of their hydrogen atoms (referred to as "deuterated") have greater molecular mass because each deuterium confers an extra unit of mass. Thus, a compound with eight deuteriums has a mass 8 amu higher than the same unlabeled compound. However, the two compounds will have essentially identical chemical properties, at least in the context of the analytical techniques we are considering. This means that two otherwise identical peptides tagged with an unlabeled and a deuterated reagent will exhibit identical chromatography and display essentially identical ionization and fragmentation in MS. However, the MS instrument distinguishes them as separate species because of their different m/z values. These features make stable isotope-labeled tags ideal agents for labeling, tracking, and quantifying the same peptide in two different samples.
Let's apply stable isotope labeling to the analysis of a particular peptide that is present in two samples. Our hypothetical sample A has 100 pmol of the peptide, whereas sample B has 50 pmol. We treat sample A with a reagent that adds a chemical tag (to the N-terminus, for example) and sample B with a d10 (deuteriumlabeled) tag (Fig. 3). The samples then are mixed and analyzed by LC-MS. Because the unlabeled tagged peptide (from sample A) and the d10-tagged peptide (from sample B) exhibit virtually identical chemical behavior, they elute together from the HPLC column and enter the ESI source at the same time. A full-scan spectrum (Fig. 3) indicates signals for both tagged versions. The instrument records a
Fig. 3. Relative quantification of a protein in two samples by stable isotope-labeled N-terminal tags and LC-MS-MS analysis.
Fig. 3. Relative quantification of a protein in two samples by stable isotope-labeled N-terminal tags and LC-MS-MS analysis.
full-scan spectrum that records the singly and doubly charged ions for the d0- (unlabeled) and d10-tagged peptides. These are selected and subjected to MS-MS. The MS-MS spectra are essentially identical (except for expected mass differences owing to the difference between the d0 and d10 tag masses), which shows that these two precursor ions in the full-scan spectrum represent the same peptide. Moreover, the MS-MS spectra can be analyzed with Sequest to establish the protein from which they originated. Moreover, the ratio of intensities for the d0- and d10-tagged peptides ions indicates their ratio in the original samples A and B. In other words, the intensity of the doubly charged ion for the d0-tagged peptide is twice that for the d10-tagged peptide. This reflects the presence of twice the amount of the peptide in sample A compared to sample B.
Now that we have established how isotope tagging can help us quantify peptides by LC-MS-MS, lets take a closer look at the application of this approach by Gygi and Aebersold. Remember, our "real world" problem is comparative quantification of many proteins between two samples. We face the problems of analyzing a complex protein mixture that can give rise to an even more complex peptide mixture. Each protein may yield many peptides upon tryptic digestion, but we really only need to generate tagged derivatives of one or two representative peptides from each of the proteins to identify them and to provide a basis for measuring relative amounts. The Gygi and Aebersold approach deals with this problem by employing an innovative, multifunctional tagging reagent (Fig. 4). The reagent is called an "isotope-coded affinity tag" (ICAT). The reagent has three parts. The first is a thiol-reactive iodoacetamide functional group. This allows the tag to covalently label free cysteinyl thiols in proteins. The second feature is a linker, which may contain either hydrogens (unlabeled, d0) or deuteriums (labeled, d8). The third feature is a biotin moiety, which confers high affinity for avidin.
The procedure for the analysis is summarized in Fig. 5. Protein sample A is treated with the light d0-ICAT reagent, whereas sample B is treated with the heavy d8-ICAT reagent. The reagents label one or more cysteinyl thiols on the proteins. The samples then are combined and digested together with trypsin to generate a very complex digest containing a relatively small proportion of peptides with ICAT tags. The entire mixture is then applied to a column of avidin beads and
the ICAT-tagged peptides bind tightly through their biotin moieties. The majority of the peptides in the combined sample are then washed away. What remains on the avidin beads are ICAT-tagged peptides from both samples A and B. Thus, the initially very complex tryptic digest has been simplified considerably to a group of bound ICAT-tagged peptides that serve as "representatives" of the proteins from which they originated. The ICAT-tagged peptides are then eluted from the column and analyzed by LC-MS-MS with data-dependent scanning. The analysis of these data is now essentially identical to that illustrated for the simple sample of two peptides in Fig. 3. A Sequest analysis of the MS-MS data (with correction of the cysteine residue mass for the presence of the ICAT tag) establishes the peptide and protein sequences that correspond the peptides analyzed. Thus, the ICAT-tagged peptides yield sequence information that permits identification of the proteins from which they originated. Examination of the full scan that corresponds to each MS-MS scan reveals the precursor ions bearing the d0- and d8-ICAT tags. The tagged peptides are carried through the entire analysis in proportions that matched those of the proteins from which they originated. These proportions are indicated by the ratio of the d0- to d8-ICAT-tagged peptide in the full-scan spectrum. This ratio thus indicates the ratio of the corresponding protein in sample A to that same protein in sample B.
4 copies of target protein
Was this article helpful?
Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.