Proteomics is the study of the protein complement present in a cell at any given moment in time. In part they reflect the genome, but they also mirror changes due to further cellular processes resulting in co- and post-translational modification (PTM) events, including for example acetylation, phosphorylation and oxidation. Figure 2 is an example of a silver-stained 2-D PAGE gel where the proteins from a non-failing, ‘normal’ samples of human ventricle are separated in the first dimension on a 3–10 pH gradient and in the second dimension on a 10% acrylamide gel.
The excitement surrounding proteomics focuses on the potential to elucidate the molecular mechanisms that control normal cell and organ functions as well as, the potential to identify proteins responsible for initiation and progression of disease. In this respect, the proteins that change as well as those that do not are key to our understanding of the subtle and complex manner in which the cells adapt to pathophysiological stimuli resulting in disease.
This is especially true with heart failure, a complex multi-factorial disease/disorder ultimately diagnosed on the basis of the inability of the heart to maintain sufficient cardiac output to meet the metabolic needs of the body.
Potentially, there are many sub-proteomes within cardiomyocytes capable of causing reduced cardiac output. Proteomic analysis can certainly help in sorting out which families of proteins are affected. The outcome of a proteomic experiment is a long list of proteins can change (or not change) due to the heart failure process. The ultimate goal is to differentiate the protein changes that are causative versus those that occur as a result/effect of the disease. To discern ‘cause vs. effect’ requires foresight in experimental design and quantitative analyses of the proteomic data.
Proteomic experimental design
There are several experimental strategies that can assist in addressing this issue. The first is based on the assumption that protein(s) altered early after the initiation or trigger of heart failure may subsequently cause downstream protein changes. This connectivity is key to understanding the far-reaching cellular dynamics and interconnections.
To obtain this information it is prudent to undertake temporal protein profiling. In other words, to carry out in-depth proteomic analysis on a series of tissue samples, preferably from the same patient or animal model, obtained over the time course of the disease progression.
It would be even more powerful to match clinical and physiological measurements (such as cardiac output or ejection fraction), and pharmacological records obtained over the same time course. For example, using the reversible, fast, pacing-induced canine model of heart failure, proteomic analysis was carried out on dogs in end-stage heart failure. Cardiac output was tracked in these animals by echocardiography giving raise to physiologically relevant time points for the collection of proteomic data. In this study, we (Heinke et al., 1998, 1999) showed that a relatively small number of proteins from a 2-D gel database of around 2000 protein spots changed with end-stage heart failure.
Of the 70 proteins that showed quantitative changes in the failing heart, 42 were present at reduced abundance, while 28 were increased. After identification, these proteins were assigned to three broad functional classes: (1) proteins associated with mitochondria and energy metabolism; (2) cytoskeletal and myofibrillar proteins; and (3) proteins associated with cellular stress responses (Corbett et al., 1998).
An interesting aspect of this model is that after pacing was stopped, the hearts reverted to ‘normal’ function. An investigation into whether there is concomitant ‘normalization’ of the alterations to the cardiac proteome will of great interest.
Temporal profiling
Temporal profiling of the human myocardium proteome during the development of heart failure is not possible for many reasons. Thus, an alternative strategy is required that will provide equal insight into ‘cause vs. effect’ of various changes within the proteome of the cardiac myocyte.
The second strategy is to undertake proteomic analysis of the same disease (or a particular phenotype) produced by different causative factors. For example, heart failure can be caused by ischemia, viral infection, genetic disorders and structural problems with the heart (valve insufficiency), each ultimately producing the same disease phenotype.
The same approach can be used where multiple animal models of the same disease are analysed. Regardless of whether experiments are conducted using human or animals, this second approach is based on the assumption that a disease caused by different triggers will have overlapping (common) protein changes. These represent either convergent points for the various mechanism or end-effectors of the disease phenotype.
It is these commonalities in mechanisms that may reveal key pathways or groups of proteins/targets for therapeutic intervention. However, one must keep in mind that in some cases the same end phenotype or function may arise by different groups of proteins. In the heart, contractile dysfunction can arise either through a change of Ca-handling proteins and/or the contractile/myofilament proteins.
An example is the comparison of two different animal models of myocardial stunning. Myocardial stunning is defined as reversible cardiac dysfunction due to brief ischemic episode. In the rat isolated heart model, the myofilament protein, troponin I, is selectively proteolysed while in the in vivo swine model, the Ca-handling protein, phospholamban, undergoes target phosphorylation (for a review see Murphy and Van Eyk, 2001). Both defects are able to produce the same overt contractile dysfunction, the hallmark of stunning. Therefore, although the phenotype is the same, a different sub-proteome is being altered. The lesson to be learned from this example is that unless one understands the physiological consequences of the protein changes, the link between disease pathways may not become clear.
Proteomic analysis of human end-stage heart failure has been carried out (Corbett et al., 1998), in which samples of failing hearts resulting from both dilated cardiomyopathy and ischemic heart disease were compared with a small number of samples from ‘normal’ unused transplant donor hearts.
In spite of the inherent heterogeneity of the 2-D gel protein profiles obtained from this human population, we were able to show significant changes in some 93 proteins (80 decreased, 13 increased in abundance) from a total of a 2-D gel database of around 1600 protein spots. Interestingly, identification of these altered proteins showed them to be in the same three functional classes as described in our proteomic studies of heart failure in the paced dog model (Heinke et al., 1998, 1999).
A major limitation to these studies is the lack of large numbers of clinically well-defined failing hearts. The greatly enlarged collection of human heart samples now available to us provides the power to weed out the background noise from the common disease-induced changes. What would be truly insightful is to have several hundreds heart samples to analyse using proteomic techniques. We are nearly there.
Proteomic technologies – 2DGE
In order to understand the underlying molecular mechanisms of heart disease, it is critical that the most extensive observation of the proteome is achieved and that proteins are identified and PTMs characterized. Broad-based proteomics involves the systematic characterization of myocardial proteins most often by two-dimensional gel electrophoresis (
2DGE) followed by mass spectrometry, although alternative and complementary technologies exist. We and others have found that this type of broad-based screening represents a major experimental advance and is ideally suited to documenting the subtle, complex and dynamic changes to the proteome.
The cornerstone of proteomic analysis is 2DGE. This technique separates intact proteins in the first dimension (isoelectric focusing, IEF) based on their intrinsic pIs, and in the second dimension (SDS-PAGE), according to their molecular masses. With 2DGE, we are limited by the solubility of the proteome and therefore, we need to optimise the homogenization and IEF steps using various combinations of detergents (i.e. Stanley et al., in press) to maximize the quantity and types of proteins loaded onto the gel.
In some cases, optimization is required for a single key protein of interest (Labugger et al., 2002). Furthermore, protein spot resolution and increased coverage of the proteome can be achieved using a variety of overlapping narrow pH gradient IEF gels for the first dimension (i.e. Westbrook et al., 2001) as well as by varying the extent of crosslinking used for the SDS PAGE dimension. Even so, with myocardial tissue we still have the problem of the highly abundant myofilament proteins dominating the myocyte proteome. Therefore, it can be helpful to use of modified extraction methods to remove the myofilament sub-proteome and enrich for the cytoplasmic and membrane proteins (Arrell et al., 2001).
Alternative approaches involve isolation of a particular organelle (i.e. mitochondria – Taylor et al., 2003) or of a binding partner complex i.e. protein kinase C (Edmonson et al., 2002). In any case, complex 2DGE gel image analysis is carried out to compare multiple gels and indicate protein changes (reviewed in Dowsey et al., in press). Proteins are identified by in-gel digestion and mass spectrometry analysis of the resulting peptide fragments.
With peptide mass fingerprinting, the observed mass spectrum of the peptides are compared in silico to the known protein databases. In the case where the species or isoform of the protein is not present, identification can only be made on the basis of amino acid sequence homology. Amino acid sequencing maybe carried out with MS/MS to validate protein identification.
One important advantage to 2DGE analysis is the ability to detect PTMs that result in a change in pI or Mr. However, it is still extremely challenging to characterize PTMs. Examples are phosphorylation of the elongation factor-2 (Jager et al., 2002) or myosin light chain 1. The latter is a protein previously thought to be unphosphorylatable (Arrell et al., 2001). Another example is the apparently aberrant ubinquitination of several cardiac proteins (Weekes et al., 2003).
A high level of skill and expertise is required to generate identical or nearly identical 2DGE gels. The technique is expensive, time consuming and limited to soluble proteins. These inherent limitations have driven the development of alternative or complementary proteomic methods. Examples of such techniques are liquid chromatography (one-dimensional or two-dimensional (2D/LC)) of proteins (Neverova and Van Eyk, 2002) or peptides obtained from the global enzymatic digestion of the sample of interest (Shot gun and Mudpit (review: McDonald and Yates JR 3rd, 2002)) or based on quantitative MS using methods such as ICAT (isotope coded affinity tag reagents) (see review: Turecek, 2002). Another possibility is protein arrays (described below).
As with 2DGE, all of these methods have technical limitations, but if used collectively they can expand the observable proteome. Thus, in the future it will be come essential to use more than one method to analyse a particular proteome. Then limitations will be on sample size, technical expertise and funding.
Answers to all your Biology Questions
